Method, device and system for detecting breeding cattle status

Abstract

The present invention provides a method, device and system for detecting the status of a breeding cattle, the method comprising: obtaining individual status information of the breeding cattle based on multi-source data; obtaining individual behavior information based on the individual status information; inputting a sample set into a status detection model to obtain status detection results of all the breeding cattle. The method, device and system for detecting the status of the breeding cattle provided by the present invention are based on the collection and quantification of multi-source data to generate individual behavior information, extract core features from the individual behavior information, and use the extracted individual behavior information of various cattle as a sample set, and the status detection model performs status detection on the individual activities of the breeding cattle. It realizes the comprehensive analysis of the integration of multiple habitual data of the breeding cattle, which can improve the accuracy of abnormal detection of the breeding cattle and avoid misjudgment caused by monitoring by a single habitual data.

Description

Technical Field

[0001] The present invention relates to the technical field of breeding management, and in particular to a method, device and system for detecting the status of breeding cattle. Background Art

[0002] With the rapid development of large-scale, intensive and information-based beef cattle breeding technology, a huge amount of beef cattle breeding data has also been accumulated. During the growth process, individual breeding cattle are easily disturbed by various external factors, which affects the development of individual breeding cattle and the comprehensive benefits of breeding farms. The development of information technology and intelligent equipment technology in recent years has provided a new entry point for judging abnormal breeding cattle. Abnormal breeding cattle refer to cattle that show abnormalities during normal production activities. Such individuals may have abnormal behaviors such as illness and estrus.

[0003] In actual production, the prevention and control measures for breeding cattle in farms are divided into two categories. One is to combine experience in the manual inspection process to make supervisory judgments, predictions, and measurement and perception of individual abnormal situations. This method may cause redundant equipment and human resources investment. The other is for animals to wear standard wearable smart equipment, collect core production data in the daily production of individual breeding cattle, and use detection algorithms corresponding to different data to monitor abnormal breeding cattle in real time.

[0004] However, due to the perception type of smart equipment and the built-in algorithms of each manufacturer, certain specific behaviors of abnormal individual breeding cattle are often judged through single indicators. This may lead to misjudgment of abnormal individual breeding cattle due to the lack of combination of multi-source inertial data of individual breeding cattle, that is, the accuracy of detecting abnormal individual breeding cattle is poor. Summary of the invention

[0005] The present invention provides a method, device and system for detecting the status of a breeding cattle, so as to solve the defect of poor accuracy in detecting abnormal breeding cattle individuals in the prior art.

[0006] The present invention provides a method for detecting the status of a breeding cattle, comprising:

[0007] Based on multi-source data, obtain individual status information of breeding cattle;

[0008] Based on the individual status information, obtaining individual behavior information;

[0009] Input the sample set into the state detection model to obtain the state detection results of all breeding cattle;

[0010] Among them, the sample set includes individual behavior information of multiple breeding cattle, the state detection model is a collection of attribute trees established according to each target indicator contained in the individual state information, and the multi-source data includes sensor data collected by various monitoring devices on the breeding cattle.

[0011] According to a method for detecting the status of a breeding cattle provided by the present invention, the target indicator includes at least one of the amount of exercise, exercise duration, rest duration, eating duration, and rumination duration.

[0012] According to a method for detecting the status of a breeding cattle provided by the present invention, the method of obtaining individual status information of the breeding cattle based on multi-source data includes:

[0013] When it is determined that the multi-source data has missing values ​​in the time dimension, the missing values ​​are generated based on sensor data of neighboring individual cattle;

[0014] Based on the missing values ​​and the multi-source data, integration in the time dimension is performed to generate the individual status information.

[0015] Wherein, the distance between the neighboring breeding bull individual and the breeding bull individual is less than or equal to a preset threshold.

[0016] According to a method for detecting the status of a breeding cattle provided by the present invention, the step of obtaining current behavior information based on the individual status information includes:

[0017] Based on the individual state information, obtaining a current activity sequence;

[0018] Based on the current activity sequence and the historical activity sequence, obtaining the current behavior information;

[0019] The current behavior information includes the change entropy value of the target indicator and the discrete representation of the activity law.

[0020] According to a method for detecting the status of a breeding cattle provided by the present invention, the current behavior information is input into a status detection model to obtain the status detection result of the breeding cattle, including:

[0021] For any of the attribute trees, using the weighted value of each attribute node, obtaining the attribute cut value of the individual cattle;

[0022] Based on each of the attribute cut values, a state detection result is obtained.

[0023] According to a method for detecting the state of a breeding cattle provided by the present invention, before inputting the current behavior information into a state detection model to obtain the state detection result of the breeding cattle, the method further comprises:

[0024] Based on the number of the sample sets and the number of target indicators, the number of attribute nodes included in the attribute tree in the state detection model and the target value of the number of cuts are determined.

[0025] The present invention also provides a device for detecting the status of a breeding cattle, comprising:

[0026] The data fusion module is used to obtain the individual status information of the breeding cattle based on multi-source data;

[0027] A feature extraction module, used for acquiring individual behavior information based on the individual state information;

[0028] A state detection module, used for inputting the individual behavior information into a state detection model to obtain a state detection result of the breeding cattle;

[0029] The state detection model is a set of attribute trees established according to each target indicator included in the individual state information, and the multi-source data includes sensor data collected by various monitoring devices on the breeding cattle.

[0030] The present invention also provides a system for detecting the status of a breeding cattle, comprising a first monitoring device and a second monitoring device, and also comprising the device for detecting the status of a breeding cattle as described above;

[0031] Among them, the first monitoring device is fixedly installed in the breeding farm, and the second monitoring device is wearable on individual breeding cattle. The second monitoring device includes at least one of an exercise monitoring sensor, a positioning monitoring device, a eating monitoring device and a rumination monitoring device.

[0032] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for detecting the status of a breeding cattle as described in any one of the above is implemented.

[0033] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for detecting the status of a breeding cow as described in any one of the above is implemented.

[0034] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for detecting the status of a breeding cattle.

[0035] The method, device and system for detecting the status of breeding cattle provided by the present invention are based on the collection and quantification of multi-source data to generate individual behavior information, extract core features from the individual behavior information, use the extracted individual behavior information of various cattle as a sample set, and use the status detection model to perform status detection on the individual activities of the breeding cattle. It realizes the comprehensive analysis of the integration of multiple habitual data of breeding cattle, can improve the accuracy of abnormal detection of breeding cattle, and avoid misjudgment caused by monitoring by single habitual data. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are 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.

[0037] Figure 1 It is a schematic diagram of the flow of the method for detecting the status of breeding cattle provided by the present invention;

[0038] Figure 2 It is a structural schematic diagram of a detection device for breeding cattle status provided by the present invention;

[0039] Figure 3 It is a structural schematic diagram of a detection system for breeding cattle status provided by the present invention;

[0040] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0043] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present invention, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0044] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0045] Figure 1 FIG. 1 is a flow chart of the method for detecting the status of a breeding cattle provided by the present invention. Figure 1 As shown, the method for detecting the status of a breeding cow provided by an embodiment of the present invention includes: step 101, obtaining individual status information of the breeding cow based on multi-source data.

[0046] Among them, multi-source data includes sensor data collected by various monitoring devices on breeding cattle.

[0047] It should be noted that the executor of the method for detecting the status of a breeding cow provided in the embodiment of the present invention is a device for detecting the status of a breeding cow.

[0048] The application scenario of the method for detecting the status of breeding cattle provided by the embodiment of the present invention is to collect core production data in the daily production of individual breeding cattle, so that the electronic device on which the detection device for the status of breeding cattle relies uses data mining technology to perform convergence analysis of multi-source sensor data on individual breeding cattle monitoring data, and combines the occurrence and development laws of individual abnormal behaviors of breeding cattle, using corresponding detection algorithms to monitor abnormal individual breeding cattle in real time.

[0049] The above-mentioned electronic device can be implemented in various forms. For example, the electronic device described in the embodiment of the present application may include mobile terminals such as mobile phones, smart phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), smart bracelets, smart watches, etc., and fixed terminals such as digital TVs, desktop computers, farm servers, etc. Below, it is assumed that the electronic device is a mobile terminal. However, it will be understood by those skilled in the art that, in addition to components specifically used for mobile purposes, the construction according to the embodiment of the present application can also be applied to fixed type terminals.

[0050] Among them, commonly used monitoring equipment may include collection nodes set at fixed locations in breeding farms, wearable devices set on individual breeding cattle, collection equipment held by information officers and other intelligent monitoring equipment, which is not specifically limited in the embodiments of the present invention.

[0051] Preferably, multi-source data refers to the sensor data of individual breeding cattle collected by at least one monitoring device in the corresponding working cycle. Multi-source data includes but is not limited to data collected by basic individual information collection nodes of breeding farms, data collected by wearable devices, and electronic records collected manually by information personnel.

[0052] Among them, the wearable devices of individual breeding cattle may include pedometers for obtaining exercise data, collar devices for sensing rumination movements, and other intelligent monitoring devices.

[0053] Accordingly, the multi-source data collected by various wearable devices may include:

[0054] (1) The sensor data collected by the exercise monitoring sensor may include individual information of the breeding cattle, individual exercise information and collection time.

[0055] (2) The sensor data collected by the locator may include individual information of the breeding cattle, the individual real-time location, and the data collection time corresponding to the real-time location.

[0056] (3) The sensor data collected by the feeding monitoring collar may include individual information of the breeding cattle, individual feeding start time and individual feeding end time;

[0057] (4) The sensor data collected by the rumination monitoring collar may include individual information of the breeding cattle and the accumulated rumination time.

[0058] Specifically, in step 101, the detection device for the status of the breeding cattle can receive multi-source data from various monitoring devices, integrate the multi-source data for the purpose of unifying the data subject, and obtain the individual status information of the breeding cattle.

[0059] Preferably, the ear tag number or pen binding ear tag data of the breeding cattle can be pre-bound as identification information, and after binding with the individual information of the breeding cattle collected by each monitoring device, the remaining sensor data corresponding to the individual information of the breeding cattle are gathered, and the sensor data are integrated in chronological order to obtain the individual status information of the breeding cattle within a monitoring period.

[0060] Step 102: Obtain individual behavior information based on the individual status information.

[0061] Specifically, in step 102, the detection device for the status of the breeding cattle can extract core features from the individual status information obtained in step 101, and compose the corresponding individual behavior information with the extracted features that are highly correlated with the individual behavior.

[0062] Among them, the feature extraction method includes but is not limited to principal component analysis (PCA), linear discriminant analysis (LDA), multidimensional scaling (MDS), independent component analysis (ICA), kernel principal component analysis and other algorithms, and the embodiments of the present invention do not make specific limitations on this.

[0063] Step 103: input the individual feature vector into the state detection model to obtain the state detection result of the breeding cattle.

[0064] The state detection model is a collection of attribute trees established based on each target indicator contained in the individual state information.

[0065] It should be noted that the state detection model can be an artificial intelligence model, and the embodiment of the present invention does not specifically limit the model type.

[0066] Preferably, since the individual status information of the breeding cattle contains sensor data collected by multiple monitoring devices, it has multiple characteristic attributes (i.e., target indicators) inside, so that the individual behavior information of multiple breeding cattle can be aggregated into a hyperplane data set (i.e., sample set). Abnormal individual breeding cattle can be considered as data anomalies caused by their abnormal performance feedback in daily production.

[0067] Therefore, based on the spatial segmentation technology, an attribute tree can be established for each special type corresponding to the target indicator. By using the idea of ​​binary division, the hyperplane data set can be cut layer by layer in space to achieve the classification of various target indicators.

[0068] For any attribute tree, if the value of the individual attribute is not less than the cutting attribute, it is stored in the right child node, and it is considered to meet the attribute and can be split next time. If the value of the individual attribute is less than the cutting attribute, it is stored in the left child node, and it is considered to not meet the attribute and needs to be cut out.

[0069] If the individual cut out is a single breeding cattle individual, the breeding cattle individual is removed from the original hyperplane data set, and the cutting is continued as a new data set until the cutting is completed or the number of cutting times is reached.

[0070] If the individual cut out is not a single individual, the child node of the root node is used as the sub-root node for cutting, that is, the original hyperplane data set is divided into two or more data sets for cutting until the cutting is completed or the number of cutting times is reached.

[0071] Specifically, in step 103, the detection device for the status of breeding cattle forms a sample set with the individual behavior information of all breeding cattle obtained in step 102, and uses each attribute tree in the status detection model to spatially cut the sample set. The cutting results are respectively stored in the left and right child nodes, so as to distinguish abnormal individuals from normal individuals with respect to the target indicators among multiple breeding cattle individuals, and realize abnormality detection.

[0072] It is understandable that during the detection process, the detected abnormal individuals will be identified as a minority. If there are multiple abnormal individuals in the sample set, it means that the data is highly discrete and the structural division of the attribute tree needs to be adjusted to be more complex to improve the accuracy of the division results.

[0073] The embodiment of the present invention is based on the collection and quantification of multi-source data to generate individual behavior information, extract core features from individual behavior information, use the extracted individual behavior information of various cattle as a sample set, and use the state detection model to perform state detection on individual activities of breeding cattle. It realizes the integration of multiple habitual data of breeding cattle for comprehensive analysis, which can improve the accuracy of abnormal detection of breeding cattle and avoid misjudgment caused by monitoring by single habitual data.

[0074] Based on any of the above embodiments, the target indicator includes at least one of the amount of exercise, exercise duration, rest duration, eating duration, and rumination duration.

[0075] Specifically, in step 101, the detection device for the status of the breeding cattle can quantify the sensor data involved in the multi-source data according to the pre-set target index, and then summarize them into the individual status information. The embodiment of the present invention does not specifically limit the quantification process of different target indicators.

[0076] Optionally, the target indicator may be the amount of exercise, and the quantification method thereof may be to normalize the amount of exercise to obtain the amount of exercise scalar S i,n,scale , which is calculated as follows:

[0077]

[0078] Among them, S i,n is the individual exercise information of the nth individual cattle at time i collected by the exercise monitoring sensor, S min,n and S max,n They are respectively the minimum and maximum exercise amounts of the nth individual breeding cattle calculated based on the data collected by the exercise amount monitoring sensor up to the i-th time.

[0079] Optionally, the target indicator may be exercise duration, and its quantification method may be to normalize the acquisition time of the individual exercise information of the nth individual cattle at the i-th time by the exercise monitoring sensor to obtain the exercise duration scalar T i,n,scale The normalization method is similar to the above process, and will not be described in detail in the present embodiment.

[0080] Optionally, the target indicator can be the rest time, and its quantification method can be to filter out the position information of the nth individual cattle in the resting area at the i-th time collected by the locator, and then normalize the data collection time corresponding to the position information to obtain the rest time scalar R1 i,n,scale .

[0081] Optionally, the target indicator may be the eating time, and its quantification method may be to calculate the cumulative eating time of the nth individual cattle collected by the eating monitoring collar within the i-th time, and then normalize the cumulative eating time to obtain the eating time scalar E i,n,scale .

[0082] Optionally, the target indicator can be the rumination duration, and its quantification method can be to normalize the cumulative rumination time of the nth individual cattle collected by the rumination monitoring collar in the i-th time, and then obtain the rumination duration scalar R2 i,n,scale .

[0083] Preferably, the amount of exercise, exercise duration, rest duration, eating duration, rumination duration and other information of the individual breeding cattle are counted and quantified from the multi-source data, and the information is integrated into the individual state information X corresponding to the identification information of the individual breeding cattle. i,n , this information can be represented as a data tuple as follows:

[0084] X i,n =[S i,n,scale ,T i,n,scale ,R1 i,n,scale ,E i,n,scale ,R2 i,n,scale ]

[0085] Among them, X i,n is the status information of the nth individual breeder at time i, S i,n,scale , T i,n,scale 、R1 i,n,scale 、E i,n,scale and R2 i,n,scale They are respectively the exercise amount scalar, exercise duration scalar, rest duration scalar, eating duration scalar and rumination duration scalar of the nth breeding cow at the i-th time, so that the value range of each scalar in the tuple is [0, 1].

[0086] The embodiment of the present invention quantifies the amount of exercise, exercise duration, rest duration, eating duration, and rumination duration of multi-source data, and integrates them into individual behavior information. Then, by extracting core features from the individual behavior information, the individual behavior information of various cattle extracted is used as a sample set, and each attribute tree in the state detection model cuts the activities of the breeding cattle contained in the sample set. This realizes the comprehensive analysis of the fusion of multiple habitual data of breeding cattle, which can improve the accuracy of abnormal detection of breeding cattle and avoid misjudgment caused by monitoring by a single habitual data.

[0087] On the basis of any of the above embodiments, individual status information of breeding cattle is obtained based on multi-source data, including: when it is determined that there are missing values ​​in the time dimension of the multi-source data, the missing values ​​are generated based on the sensor data of neighboring breeding cattle individuals.

[0088] Among them, the distance between the adjacent breeding cattle individuals and the breeding cattle individuals is less than or equal to a preset threshold.

[0089] It should be noted that, since the monitoring devices are manufactured by different manufacturers and have different built-in collection modes, it is necessary to unify the sensor data based on the time zone of the region or country where the detection device for the status of the breeding cattle is located as the time dimension.

[0090] For example, when the longitude of the location of the detection device for the status of the breeding cattle is between 112.5 degrees and 127.5 degrees east longitude, the collection time of the multi-source data is converted to the time dimension of UTC / GMT+8 East Eight Zone and stored in a remote server or cloud platform. However, the working cycle of the monitoring equipment and the breeding environment are complex, and the multi-source data obtained by aggregation may have a small number of missing values.

[0091] Specifically, when the detection device for the status of a breeding cattle finds that data is missing at a certain moment when reading multi-source data, it can find neighboring breeding cattle individuals similar to the breeding cattle individual with missing values, and extract corresponding values ​​from the neighboring breeding cattle individuals to fill in the missing values. The embodiment of the present invention does not specifically limit the method of filling in missing values.

[0092] Preferably, the nearest neighbor principle is combined with the individual similarity theory, that is, when a certain target indicator of a certain cattle individual is missing (such as exercise data), the sensor data of the remaining target indicators are used to find the remaining individuals adjacent to the cattle individual on the hyperplane formed by the sample set.

[0093] Each individual breeder is regarded as a sample node in a hyperplane. In a multidimensional vector space, the distance measurement method used includes but is not limited to Euclidean distance, Minkowski distance, Euclidean distance, cosine value, correlation, Manhattan distance or other methods, which are not specifically limited in the present invention.

[0094] Taking the k closest sample nodes according to the Euclidean distance as an example, two sample nodes x 1 and x 2 The distance calculation formula is as follows:

[0095]

[0096] Among them, x 1 is a sample node with missing values, x 2 is a sample node without missing values. When the original data dimension of the sample node is N+1, z is the other data dimension in the sample node except for the missing values, and its value range is [1, N]. 1 With any x 2The distances between them are arranged in descending order to extract k x whose distance is less than or equal to the preset threshold. 2 Node. Then, the k x 2 The missing fields corresponding to the nodes are supplemented by taking their weighted average.

[0097] It can be understood that the value of k can be determined according to actual conditions. In the embodiment of the present invention, k=3. If the neighbor data is empty in actual conditions, the value of k can be appropriately expanded.

[0098] Based on missing values ​​and multi-source data, integration in the time dimension is performed to generate individual status information.

[0099] Specifically, the detection device for the status of breeding cattle supplements the missing values ​​into the multi-source data, and integrates and quantifies the multi-source data with complete time series under the current time dimension, and stores the scalar of each target indicator of the individual breeding cattle at each moment during the monitoring time into the individual status information.

[0100] The embodiment of the present invention supplements missing values ​​based on the sensor data of neighboring individual cattle, and integrates individual status information in a time series through missing values ​​and multi-source data. Then, by extracting core features from individual behavior information, the extracted individual behavior information of various cattle is used as a sample set, and each attribute tree in the state detection model cuts the cattle activities contained in the sample set. This achieves the purpose of supplementing the missing information of sparse multi-source data, ensuring the integrity of data analysis and mining, improving the accuracy of characterizing the behavior status of cattle, and thus improving the accuracy of abnormal detection of cattle.

[0101] On the basis of any of the above embodiments, obtaining current behavior information based on individual state information includes: obtaining a current activity sequence based on the individual state information.

[0102] Specifically, in step 102, the detection device for the status of the breeding cattle, during the current monitoring period, generates the individual status information X of the nth breeding cattle at the time i obtained in step 101. i,n , stored in the current activity sequence p in chronological order n The specific expression formula is as follows:

[0103] p n ={X i,n}={X 1,n ,X 2,n ,…,X end,n}

[0104] ={[S 1,n,scale ,T 1,n,scale ,R1 1,n,scale ,E 1,n,scale ,R2 1,n,scale],[S 2,n,scale ,T 2,n,scale ,R1 2,n,scale ,E 2,n,scale ,R2 2,n,scale ],…,[S end,n,scale ,T end,n,scale ,R1 end,n,scale ,E end,n,scale ,R2 end,n,scale ]}

[0105] The subscript end is the end time of the equal division of this monitoring period.

[0106] The embodiment of the present invention does not specifically limit the division of the monitoring cycle. For example, it can be divided into user-defined time intervals, such as 12 hours or 24 hours. It can also be set according to the breeding rules of the farm and the collection frequency of the monitoring equipment.

[0107] It is understandable that the individual daily data of multiple breeding cattle are aggregated into a hyperplane vector data set. When new data appears, the new production data will be added to the historical activity sequence for repeated calculation. Due to the cyclical nature of breeding, the data can be truncated by setting the vector length. For example, the data length is fixed to a threshold. When the length exceeds this threshold, the original data will be eliminated in the form of a queue to keep the latest data.

[0108] Based on the current activity sequence and historical activity sequence, obtain current behavior information.

[0109] Among them, the current behavior information includes the changing entropy value of the target indicator and the discrete representation of the activity rules.

[0110] Specifically, the detection device for the status of the breeding cattle extracts a feature sequence corresponding to the target indicator from the current activity sequence, combines it with the historical activity sequence, and uses the influence of the newly added behavioral activities on its regular activity sequence as the basis for feature extraction to obtain the current behavior information.

[0111] The embodiment of the present invention does not specifically limit the method for acquiring the current behavior information.

[0112] Optionally, the current behavior information may include the entropy value of the amount of exercise To describe the fluctuation of the amount of movement of individual cattle in the same time dimension, the calculation method is:

[0113]

[0114] Among them, M is the set of all historical activity sequences, p S is the current activity sequence about the motion quantity, and m is a single set of sequences of historical activity sequences in M, that is, m∈M. SThe activity sequence of the nth individual cattle is expressed as:

[0115]

[0116] That is, from p n Each X in i,n In the above equation, we extract the motion scalar S corresponding to each time i. i,n,scale In addition, the way m records the activity sequence is similar to p S similar.

[0117] It is understandable that the entropy value of the change in the amount of motion Distributed between [0,1], the activity volatility is greater when it is close to 1, and the volatility is smaller when it is close to 0.

[0118] Optionally, the current behavior information may include a change entropy value irrT of the exercise duration to describe the fluctuation of the exercise duration of the individual cattle in a unit time, which is calculated as follows:

[0119]

[0120] Among them, from p n Each X in i,n In the above example, we extract the motion duration scalar T corresponding to each time i. i,n,scale As the current exercise duration in this monitoring cycle, is the mean of the preceding motion duration up to the i-th time. It can be seen that when irrT approaches 1, the volatility of the motion duration is greater, and when it approaches 0, the volatility is smaller.

[0121] Optionally, the current behavior information may include a change entropy value irrS of the rest duration to describe the fluctuation of the rest duration of individual breeding cattle in a unit time, which is calculated as follows:

[0122]

[0123] Among them, from p n Each X in i,n In the above example, we extract the rest time scalar R1 corresponding to each time i. i,n,scale As the current rest time in this monitoring period, is the mean of the preceding rest durations up to the i-th time. It can be seen that when irrR1 approaches 1, the volatility of rest duration is greater, and when it approaches 0, the volatility is smaller.

[0124] Optionally, the current behavior information may include the entropy value of the change in the eating time of the nth individual cattle, irrE(n), to describe the regularity of eating during the monitoring period. Combined with the feeding rules and feeding time of the farm, the time is discretized into multiple time stages i, and the eating time and duration regularity in the specified time period are recorded. The calculation method is:

[0125]

[0126] Among them, T b is the standard feeding time period, T n,d is the cumulative time of standard feeding for the nth individual breeder on that day. This value can be obtained from E i,n,scale In the middle fitting, D represents the culture age, that is, the standard feeding time data from the start of production to the day of production.

[0127] Optionally, the current behavior information may include the change entropy value irrR2 of the rumination duration to describe the proportion of the individual rumination duration during the irregular eating time period, that is, the proportion of the total duration of the individual rumination during the specified non-eating time period to the irregular eating time of the whole day. The specific representation is as follows:

[0128]

[0129] Among them, eating during the non-routine eating phase is considered rumination. feed Indicates the cumulative rumination time monitored by the rumination monitoring collar. This value can be obtained from R2 i,n,scale Medium fit, T r The feeding schedule is set for the farm.

[0130] Optionally, the current behavior information may include a discrete representation of the individual activity pattern of the nth breeder cow, namely, the activity location entropy value irrLoc(n), to measure the disorder of the individual's activities at different core locations in the shed. The cowshed is divided into multiple core areas such as rest area, feed area, activity area, and drinking water area. The calculation formula is as follows:

[0131]

[0132] Among them, L represents the entire core area of ​​the cowshed, L n It is the exchange record of all regions when the nth breeder is active, l is L n A regional exchange record of the species, X l For all partitioned areas, X l,n is the number of times the nth breeding bull enters the area within l days.

[0133] It is understandable that irrLoc(n) describes the regularity of the individual activities of the breeding cattle in the cowshed. When the value reaches an extreme value, it often indicates the abnormality of the individual breeding cattle. When it is used as the main feature, it needs to be combined with its logarithm for normalization.

[0134] The embodiment of the present invention extracts key behavior features based on individual state information, and encapsulates them in independent current behavior information in a time-series form, and then aggregates them into a sample set. Each attribute tree in the state detection model cuts the activities of the breeding cattle contained in the sample set. It can ensure that the breeding farm can perform real-time monitoring after setting sequence parameters according to the breeding rules in daily production, improve the health management level of individual breeding cattle, and improve the breeding efficiency of cattle farms.

[0135] Based on any of the above embodiments, the current behavior information is input into the state detection model to obtain the state detection result of the breeding cattle, including: for any attribute tree, using the weighted value of each attribute node to obtain the attribute cut value of the individual breeding cattle.

[0136] Specifically, in step 103, each attribute tree structure has an independent cutting attribute, and a weighted calculation can be performed on each attribute node in each attribute tree structure using a distributed or edge computing method to obtain the discrete situation of each sample node in the entire sample set. The calculation method is as follows:

[0137]

[0138]

[0139] Where x is the sample node in the sample set, n is the sample size, i.e., the total number of individual cattle in the sample set, m is the number of the attribute tree, h(x) is the number of edges that the sample node x goes through to reach the root directory of the attribute tree, E(h(x)) is the expected value of h(x), and H(·) is a fixed harmonic number, which can be a constant or can be calculated approximately according to the sample size n. The calculation method is:

[0140] H(n-1)=ln(n-1)+0.5772156649

[0141] Based on the cut values ​​of each attribute, the status detection results are obtained.

[0142] Specifically, the detection device for the state of the breeding cattle calculates the average value of the attribute cut values ​​corresponding to any sample node and multiple attribute trees to serve as the state detection result corresponding to the sample node in the sample set. The calculation method is as follows:

[0143]

[0144] Among them, Score(x) is the state detection result of sample node x, M is the total number of attribute trees, and its value is equal to the total number of target indicators.

[0145] When Score(x) is closer to 1, it means that the abnormal probability of sample node x is higher, and it can be determined that the individual status of the breeding cattle corresponding to this node is abnormal.

[0146] When Score(x) is closer to 0, it means that the abnormal probability of sample node x is lower, and it can be determined that the individual status of the breeding cattle corresponding to this node is normal.

[0147] Based on the constructed attribute tree, the embodiment of the present invention calculates the attribute cut value of each attribute tree about the individual cattle separately, and makes an overall judgment after summarizing to obtain the status detection result. It can improve the accuracy and efficiency of abnormal detection of cattle, and effectively avoid the high amount of calculation based on feature dimensions.

[0148] Based on any of the above embodiments, before inputting the current behavior information into the state detection model to obtain the state detection result of the breeding cattle, it also includes: determining the number of attribute nodes contained in the attribute tree in the state detection model and the target value of the number of cutting times based on the number of sample sets and the number of target indicators.

[0149] Specifically, before step 103, the initial parameters of the state detection model may be adaptively set according to the existing sample set, so as to improve the processing efficiency of the state detection model to a certain extent.

[0150] The embodiment of the present invention does not specifically limit the setting of the initial parameters.

[0151] Optionally, the initial parameter can be an initial cutting range. Since the attribute cutting value is randomly selected from all points, that is, the dimension cutting is performed between the maximum values ​​of all data, it is necessary to set the dimension measurement. The state detection model uses the Euclidean distance as the measurement method by default, that is, the origin is used as the initial point. There are clear regulations for various production indicators in regular production, and the origin design can also be non-zero point, or it can be customized according to usage habits.

[0152] Optionally, the initial parameter can be the number of attribute nodes. Since the cutting is random, too few constructions may result in insufficient accuracy, and too many attribute nodes in the attribute tree may cause computational redundancy. Therefore, the number of attribute nodes Num can be adaptively designed according to the number of samples and the number of anomalies in the sample set. The calculation method is:

[0153]

[0154] Among them, n is the total number of samples in the sample set, and A is the total number of target indicators.

[0155] Optionally, the initial parameter may be the number of cuts. Since abnormal points in the hyperplane are often prominent points of the data, when the number set is huge, too many cuts are also redundant, since the total number of cuts is lnn.

[0156] It is understandable that if n is too large, it can be set to Any value in between.

[0157] Optionally, the initial parameter can be a cutting attribute. Since the cutting method is random cutting, the cutting dimension can be cut according to the original attributes, or a weighted method can be set for combined cutting, that is, two groups of attributes or multiple groups of attributes are combined for cutting. The combination of attributes can be used as a new feature, which will affect the number of attribute nodes in the attribute tree.

[0158] The embodiment of the present invention optimizes the initial parameters of the constructed attribute tree based on the number of sample sets and the number of target indicators, and then quickly reaches the cutting condition to calculate the attribute cutting value to obtain the state detection result. The efficiency of abnormal detection of breeding cattle can be improved and overfitting can be effectively avoided.

[0159] Figure 2 is a schematic diagram of the structure of the detection device for the status of a breeding cattle provided by the present invention. Figure 2 As shown, the detection device for the state of a breeding cattle provided in an embodiment of the present invention comprises a data fusion module 210, a feature extraction module 220 and a state detection module 230, wherein:

[0160] The data fusion module 210 is used to obtain individual status information of the breeding cattle based on multi-source data.

[0161] The feature extraction module 220 is used to obtain individual behavior information based on individual status information.

[0162] The state detection module 230 is used to input the individual behavior information into the state detection model to obtain the state detection result of the breeding cattle.

[0163] Among them, the state detection model is a collection of attribute trees established according to each target indicator contained in the individual state information, and the multi-source data includes the sensor data collected by various monitoring devices on the breeding cattle.

[0164] Specifically, the data fusion module 210, the feature extraction module 220 and the state detection module 230 are electrically connected in sequence.

[0165] The data fusion module 210 can receive multi-source data from various monitoring devices, integrate the multi-source data for the purpose of unifying the data subject, and obtain individual status information of the breeding cattle.

[0166] The feature extraction module 220 may perform core feature extraction on the individual state information acquired by the data fusion module 210 , and compose the corresponding individual behavior information with the extracted features that are highly correlated with the individual behavior.

[0167] The state detection module 230 forms a sample set with the individual behavior information of all the breeding cattle obtained in the feature extraction module 220, and uses each attribute tree in the state detection model to spatially cut the sample set. The cutting results are respectively stored in the left and right child nodes to distinguish the abnormal individuals from the normal individuals with respect to the target indicators among multiple breeding cattle individuals, thereby realizing abnormality detection.

[0168] Optionally, the target indicator includes at least one of exercise volume, exercise duration, rest duration, eating duration, and rumination duration.

[0169] Optionally, the data fusion module 210 includes a missing value supplement unit and an integration unit, wherein:

[0170] The missing value supplement unit is used to generate missing values ​​based on the sensor data of neighboring breeding cattle individuals when it is determined that the multi-source data has missing values ​​in the time dimension.

[0171] The integration unit is used to integrate the missing values ​​and multi-source data in the time dimension to generate individual status information.

[0172] Among them, the distance between the adjacent breeding cattle individuals and the breeding cattle individuals is less than or equal to a preset threshold.

[0173] Optionally, the feature extraction module 220 includes a sequence representation unit and a rule representation unit, wherein:

[0174] The sequence representation unit is used to obtain the current activity sequence based on the individual state information.

[0175] The rule representation unit is used to obtain current behavior information based on the current activity sequence and the historical activity sequence.

[0176] Among them, the current behavior information includes the changing entropy value of the target indicator and the discrete representation of the activity rules.

[0177] Optionally, the state detection module 230 includes a cutting unit and a summarizing unit, wherein:

[0178] The cutting unit is used to obtain the attribute cutting value of the individual breeding cattle by using the weighted value of each attribute node for any attribute tree.

[0179] The summary unit is used to obtain the status detection results based on the cut values ​​of each attribute.

[0180] Optionally, the device further comprises a preset module, wherein:

[0181] The preset module is used to determine the number of attribute nodes contained in the attribute tree in the state detection model and the target value of the number of cuts based on the number of sample sets and the number of target indicators.

[0182] The device for detecting the status of a breeding cow provided in an embodiment of the present invention is used to execute the above-mentioned method for detecting the status of a breeding cow of the present invention. Its implementation method is consistent with the implementation method of the method for detecting the status of a breeding cow provided by the present invention, and can achieve the same beneficial effects, which will not be repeated here.

[0183] The embodiment of the present invention is based on the collection and quantification of multi-source data to generate individual behavior information, extract core features from individual behavior information, use the extracted individual behavior information of various cattle as a sample set, and use the state detection model to perform state detection on individual activities of breeding cattle. It realizes the integration of multiple habitual data of breeding cattle for comprehensive analysis, which can improve the accuracy of abnormal detection of breeding cattle and avoid misjudgment caused by monitoring by single habitual data.

[0184] Figure 3 FIG. 1 is a schematic diagram of the structure of the detection system for the status of a breeding cattle provided by the present invention. Figure 3 As shown, the system for detecting the status of a breeding cattle provided by the embodiment of the present invention includes a first monitoring device 310 and a second monitoring device 320 , and also includes a device 330 for detecting the status of a breeding cattle.

[0185] Among them, the first monitoring device 310 is fixedly installed in the breeding farm, and the second monitoring device 320 is worn on the individual breeding cattle. The second monitoring device 320 includes at least one of an exercise monitoring sensor, a positioning monitoring device, a eating monitoring device and a rumination monitoring device.

[0186] Specifically, the detection system of the state of the breeding cattle is composed of a first monitoring device 310, a second monitoring device 320 and a detection device 330 of the state of the breeding cattle. The detection device 330 of the state of the breeding cattle can be integrated into a remote control terminal with a chip or a microprocessor, and the detection device 330 of the state of the breeding cattle communicates with the first monitoring device 310 and the second monitoring device 320 respectively through the detection device 330 of the state of the breeding cattle, and multiple types of sensor data are integrated, combined with the calculation method of biasing non-root nodes after sparse segmentation, to perform abnormal detection of individual breeding cattle.

[0187] The embodiment of the present invention does not specifically limit the types and configurations of the first monitoring device 310 and the second monitoring device 320 .

[0188] The first monitoring device 310 can be used as a collection node fixedly set in each area of ​​the cowshed to monitor the number of times and time when the breeding cattle enter and exit a certain area to characterize the activity pattern of the breeding cattle.

[0189] The second monitoring device 320 can be used as an intelligent monitoring device worn on individual cattle to monitor various activity states of the cattle (such as exercise, eating, rumination, rest, etc.) to characterize the behavioral characteristics of the cattle. The second monitoring device 320 includes but is not limited to an exercise quantity monitoring sensor (such as a pedometer), a positioning monitoring device, an eating monitoring device (such as an eating monitoring collar), and a rumination monitoring device (such as a rumination monitoring collar), etc., and the embodiment of the present invention does not specifically limit this.

[0190] Preferably, the detection device 330 of the breeding cattle status respectively transmits signals with the first monitoring device 310 and the second monitoring device 320 using wireless communication technology.

[0191] Among them, wireless communication technology includes but is not limited to WIFI wireless cellular signals (2G, 3G, 4G, 5G), Bluetooth, Zigbee and other methods, and the embodiments of the present invention do not specifically limit this.

[0192] The embodiment of the present invention is based on the collection and quantification of multi-source data to generate individual behavior information, extract core features from individual behavior information, use the extracted individual behavior information of various cattle as a sample set, and use the state detection model to perform state detection on individual activities of breeding cattle. It realizes the integration of multiple habitual data of breeding cattle for comprehensive analysis, which can improve the accuracy of abnormal detection of breeding cattle and avoid misjudgment caused by monitoring by single habitual data.

[0193] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the method for detecting the state of the breeding cattle, the method comprising: obtaining individual state information of the breeding cattle based on multi-source data; obtaining individual behavior information based on the individual state information; inputting the sample set into the state detection model to obtain the state detection results of all the breeding cattle; wherein the sample set includes individual behavior information of multiple breeding cattle, the state detection model is a set of attribute trees established according to each target indicator contained in the individual state information, and the multi-source data includes the sensor data collected by each monitoring device on the breeding cattle.

[0194] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0195] On the other hand, the present invention also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the detection method of the breeding cattle status provided by the above-mentioned methods, and the method includes: based on multi-source data, obtaining individual status information of the breeding cattle; based on the individual status information, obtaining individual behavior information; inputting the sample set into the status detection model to obtain the status detection results of all breeding cattle; wherein the sample set includes individual behavior information of multiple breeding cattle, the status detection model is a collection of attribute trees established according to each target indicator contained in the individual status information, and the multi-source data includes sensor data collected by each monitoring device on the breeding cattle.

[0196] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the method for detecting the status of breeding cattle provided by the above-mentioned methods, the method comprising: obtaining individual status information of breeding cattle based on multi-source data; obtaining individual behavior information based on the individual status information; inputting a sample set into a status detection model to obtain status detection results of all breeding cattle; wherein the sample set includes individual behavior information of multiple breeding cattle, the status detection model is a collection of attribute trees established according to each target indicator contained in the individual status information, and the multi-source data includes sensor data collected by each monitoring device on the breeding cattle.

[0197] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0198] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0199] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the status of a breeding cattle, characterized in that: include: Based on multi-source data, obtain individual status information of breeding cattle; Based on the individual status information, obtaining individual behavior information; Input the sample set into the state detection model to obtain the state detection results of all breeding cattle; The sample set includes individual behavior information of multiple breeding cattle, the state detection model is a set of attribute trees established according to each target indicator contained in the individual state information, the multi-source data includes sensor data collected by each monitoring device on the breeding cattle; the target indicators include exercise volume, exercise duration, rest duration, eating duration and rumination duration; The acquiring of individual behavior information based on the individual status information includes: Based on the individual state information, obtaining a current activity sequence; Based on the current activity sequence and the historical activity sequence, obtaining the individual behavior information; Among them, the individual behavior information includes the change entropy value of the target indicator and the discrete representation of the activity pattern; the change entropy value of the target indicator is used to describe the fluctuation of the target indicator per unit time; the discrete representation of the activity pattern includes the activity location entropy value, and the activity location entropy value is used to measure the degree of disorder of the individual's activities in different core positions in the house.

2. The method for detecting the status of a breeding cattle according to claim 1, characterized in that: The method of obtaining individual status information of the breeding cattle based on multi-source data includes: When it is determined that the multi-source data has missing values ​​in the time dimension, the missing values ​​are generated based on sensor data of neighboring individual cattle; Based on the missing values ​​and the multi-source data, integration is performed in a time dimension to generate the individual status information; Wherein, the distance between the neighboring breeding bull individual and the breeding bull individual is less than or equal to a preset threshold.

3. The method for detecting the status of a breeding cattle according to claim 1, characterized in that: The sample set is input into the state detection model to obtain the state detection results of all the breeding cattle, including: For any of the attribute trees, using the weighted value of each attribute node, obtaining the attribute cut value of the individual cattle; Based on each of the attribute cut values, a state detection result is obtained.

4. The method for detecting the status of a breeding cattle according to claim 1, characterized in that: Before inputting the sample set into the state detection model to obtain the state detection results of all the breeding cattle, the method further includes: Based on the number of the sample sets and the number of target indicators, the number of attribute nodes included in the attribute tree in the state detection model and the target value of the number of cuts are determined.

5. A device for detecting the status of a breeding cattle, characterized in that: include: The data fusion module is used to obtain the individual status information of the breeding cattle based on multi-source data; A feature extraction module, used for acquiring individual behavior information based on the individual state information; A state detection module, used for inputting the individual behavior information into a state detection model to obtain a state detection result of the breeding cattle; The state detection model is a set of attribute trees established according to each target indicator included in the individual state information, the multi-source data includes sensor data collected by each monitoring device on the breeding cattle; the target indicators include exercise volume, exercise duration, rest duration, eating duration and rumination duration; The acquiring of individual behavior information based on the individual status information includes: Based on the individual state information, obtaining a current activity sequence; Based on the current activity sequence and the historical activity sequence, obtaining the individual behavior information; Among them, the individual behavior information includes the change entropy value of the target indicator and the discrete representation of the activity pattern; the change entropy value of the target indicator is used to describe the fluctuation of the target indicator per unit time; the discrete representation of the activity pattern includes the activity location entropy value, and the activity location entropy value is used to measure the degree of disorder of the individual's activities in different core positions in the house.

6. A system for detecting the status of a breeding cattle, characterized in that: It comprises a first monitoring device and a second monitoring device, and also comprises a detection device for the state of a breeding cattle as claimed in claim 5; Among them, the first monitoring device is fixedly installed in the breeding farm, and the second monitoring device is worn on individual breeding cattle. The second monitoring device at least includes an exercise monitoring sensor, a positioning monitoring device, a eating monitoring device and a rumination monitoring device.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for detecting the status of a breeding cattle as described in any one of claims 1 to 4 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting the status of a breeding cattle as claimed in any one of claims 1 to 4 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for detecting the status of a breeding cattle as claimed in any one of claims 1 to 4 is implemented.

Images