Animal survival status early warning method, system and device
By combining environmental sensors and biological vital signs parameters into a predictive model, the problems of high cost and low accuracy of traditional equipment have been solved, enabling low-cost and high-precision early warning of animal survival status, thus improving animal welfare and economic benefits in the breeding process.
Patent Information
- Application Number
- CN202210099542.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-27
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2042-01-27
AI Technical Summary
Existing technologies struggle to provide low-cost and high-precision early warnings of animal survival status, especially in livestock farming, where traditional wearable devices are expensive and have low monitoring accuracy.
By collecting environmental parameters and human factors in the animal's habitat using environmental sensors, combining them with a biological trait parameter prediction model, setting a monitoring frequency, and using a survival status classification model for early warning, energy consumption can be reduced and prediction accuracy improved.
It achieves low-cost and high-precision prediction of animal survival status, improves prediction accuracy and reduces equipment energy consumption, and ensures animal welfare and economic benefits.
Smart Images

Figure CN114428810B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of animal breeding, and in particular to an animal survival status early warning method, system and equipment. Background Art
[0002] With the development and progress of the times, higher requirements have been put forward for animal welfare and quality, and modern animal farming has gradually shifted to intensive farming. Among them, the survival status of livestock is an important indicator in the breeding process, which requires breeders to pay great attention to. Generally, breeders can use existing experience to observe the survival status of livestock, but it is difficult to achieve early warning mechanisms and precision animal farming. At present, there is technical support for wearing wearable devices on livestock to monitor livestock status indicators and achieve early warning of abnormal animal survival indicators, but due to the high cost, it cannot be invested and the monitoring accuracy is low. Summary of the invention
[0003] In view of this, the present invention provides an animal survival status early warning method, system and equipment, which predict the changes in the physiological and behavioral characteristics of animals by combining environmental factors with human operation parameters, and further predict the animal's survival status, thereby achieving low-cost and high-precision survival status prediction.
[0004] To achieve the above object, the present invention provides the following solutions:
[0005] An animal survival status early warning method, comprising:
[0006] Based on the environmental sensors, various environmental parameters in the living area where the animals are located are collected to obtain a real-time environmental parameter data set, and various human factors in the living area where the animals are located are obtained to obtain a real-time initial human factor data set;
[0007] Based on the changes of the real-time environmental parameter data set and the real-time initial human factor data set, a monitoring frequency is set, and based on the monitoring frequency, an environmental parameter prediction data set and a human factor prediction data set are obtained;
[0008] Based on the environmental parameter prediction data set and the human factor prediction data set, combined with the biological sign parameter prediction model, a biological sign parameter prediction set is obtained;
[0009] Based on the biological sign parameter prediction set and in combination with the survival status classification model, the survival status of the animal is obtained;
[0010] Based on the survival status, combined with the early warning mechanism, the survival status of the animals is monitored and warned.
[0011] Preferably, the environmental parameters include ambient temperature, relative humidity, mixed gas concentration, light intensity and noise;
[0012] The human factors include feeding, passing of people, construction, chasing, catching, depilation, blood drawing and gathering of feed during animal breeding;
[0013] The living areas of the animals include: the activity areas of animals that are in the same stage of their growth cycles and whose average living distance does not exceed 5 meters, and the activity areas of animals of the same sex whose average living distance does not exceed 5 meters.
[0014] Preferably, the step of setting a monitoring frequency based on changes in the real-time environmental parameter data set and the real-time initial human factor data set, and obtaining an environmental parameter prediction data set and a human factor prediction data set based on the monitoring frequency includes:
[0015] The real-time initial human factor data set is converted from text form to vector form to obtain a real-time human factor data set; specifically, a vocabulary set V is defined according to the interference type, and each real-time initial human factor data X in the real-time initial human factor data set is represented as a vector X∈R of |V| dimension. |V| , R is a positive real number, R |V| represents |V|-dimensional Euclidean space; the value of the i-th dimension in each interference vector x in the real-time initial human factor data represents whether the i-th word in the vocabulary set appears in x, if it appears, the value is 1, otherwise it is 0, and the real-time human factor data set is obtained;
[0016] For any environmental parameter in the real-time environmental parameter data set, a set environmental change threshold is corresponding. The parameter value of the current environmental parameter at the start time t0 of the current collection cycle and the parameter value at t0+1 are obtained. If the parameter value change rate at t0 and t0+1 is greater than the corresponding set environmental change threshold, the current environmental parameter is continuously collected in the current collection cycle. If the parameter value change rate at t0 and t0+1 is less than or equal to the corresponding set environmental change threshold, the current environmental parameter is intermittently collected in the current collection cycle, and each environmental parameter is traversed to obtain the environmental parameter prediction data set; the monitoring frequency includes the continuous collection and the intermittent collection; the parameter value change rate calculation formula is as follows:
[0017]
[0018] Where: α is the rate of change of parameter value, is the parameter value at time t0+1, t θ0 is the parameter value at time t0;
[0019] For any human factor in the real-time human factor data set, a set human change threshold is corresponding, and the factor value of the current human factor at the start time t0 of the current collection cycle and the factor value at time t0+1 are obtained. If the factor value change rate at time t0 and time t0+1 is greater than the corresponding set human change threshold, the current human factor is continuously collected in the current collection cycle. If the factor value change rate at time t0 and time t0+1 is less than or equal to the corresponding set human change threshold, the current human factor is intermittently collected in the current collection cycle, and each human factor is traversed to obtain the human factor prediction data set;
[0020] For any environmental parameter in the real-time environmental parameter data set, there is a corresponding set environmental threshold range. When the parameter value of any environmental parameter exceeds the corresponding set environmental threshold range, an environmental warning signal is issued.
[0021] Preferably, the biological sign parameter prediction set is obtained based on the environmental parameter prediction data set and the human factor prediction data set in combination with a biological sign parameter prediction model, including:
[0022] Obtaining environmental parameter data training sets, human factor data training sets and biological sign parameter training sets;
[0023] Constructing an initial biological sign parameter prediction model based on the environmental parameter data training set, the human factor data training set and the biological sign parameter training set, and training the initial biological sign parameter prediction model to obtain the biological sign parameter prediction model;
[0024] Preprocessing the environmental parameter prediction data set and the human factor prediction data set to obtain a prediction input matrix;
[0025] Inputting the prediction input matrix into the biological sign parameter prediction model to obtain the biological sign parameter prediction set;
[0026] The biological sign parameter prediction set is judged, and when any biological sign parameter prediction value in the biological sign parameter prediction set exceeds its corresponding biological sign threshold range, a biological sign warning signal is issued; when any biological sign parameter prediction value is within its corresponding biological sign threshold range, the survival status is obtained based on the biological sign parameter prediction set.
[0027] Preferably, the environmental parameter data training set includes an environmental temperature data training set, a relative humidity data training set, a mixed gas concentration data training set, a light intensity data training set and a noise data training set;
[0028] The human factor data training set includes a feeding data training set, a personnel passing data training set, a construction data training set, a chasing data training set, a capturing data training set, a hair removal data training set, a blood drawing data training set and a material gathering data training set;
[0029] The biological sign parameter training set is obtained by collecting data from animals based on a flexible wearable biosensor; the biological sign parameter training set includes a physiological parameter data training set and a behavioral characteristic parameter data training set; the physiological parameter data training set includes a heart rate data training set, a body temperature data training set, a blood pressure data training set and a respiratory data training set; the behavioral characteristic parameter data training set includes a step data training set, a standing and lying data training set, an acceleration data training set, a call data training set and an angular velocity data training set;
[0030] Preprocessing the environmental parameter data training set, the human factor data training set and the biological sign parameter training set to obtain a preprocessed training set;
[0031] Extracting the preprocessed training set based on a random extraction method to obtain training samples, and training the initial biological sign parameter prediction model based on the training samples to obtain the biological sign parameter prediction model;
[0032] The environmental parameter prediction data set and the human factor prediction data set are preprocessed to obtain a prediction input matrix, and the specific calculation formula is as follows:
[0033]
[0034] Where: a` is the environmental parameter prediction data or human factor prediction data after preprocessing; a is the environmental parameter prediction data or human factor prediction data before preprocessing, a i is the value of the ith data in a certain environmental parameter prediction data or human factor prediction data before preprocessing, and m is the total number of data in a certain environmental parameter prediction data or human factor prediction data.
[0035] Preferably, the survival state includes a normal state, a panic state, a depressed state and a death state;
[0036] The method of obtaining the animal's survival status based on the biological sign parameter prediction set and combining the survival status classification model includes:
[0037] The biological sign parameter prediction set is preprocessed to obtain a preprocessed biological sign parameter set. The specific formula is as follows:
[0038]
[0039] Where: y miis the value of the i-th biological sign parameter in the m-th group of predicted data in the biological sign parameter prediction set, Y mi is the preprocessed value of the ith biological sign parameter in the mth group of prediction data in the biological sign parameter prediction set, and M is the number of groups of data in the biological sign parameter prediction set;
[0040] Classifying the preprocessed biological sign parameter set to obtain a qualified data set and a non-qualified data set; the qualified data set corresponds to the normal state and the panic state; the non-qualified data set corresponds to the depressed state and the death state;
[0041] Based on the pre-processed biological sign parameter set, the animal survival status index set is obtained, and the calculation formula is as follows:
[0042]
[0043] Where: is the survival status index obtained by the k-th prediction, k∈1,2,3, when k=1, the pre-processing biological sign parameter set is taken The pre-processed data of each group is 1, and the qualified data set is obtained. For each group of pre-processed data with k=-1, a substandard data set is obtained. For the standard data set, k=2 is set and the calculation is continued to obtain a second survival state index set. For the substandard data set, k=3 is set and the calculation is continued to obtain a third survival state index set. The survival state index set includes the second survival state index set and the third survival state index set.
[0044] When k = 1: M1 is the number of groups of preprocessed data in the preprocessed biological sign parameter set, α m is the vector factor of the mth group of preprocessed data in the preprocessed biological sign parameter set, z m is the survival status index of the animal corresponding to the mth group of preprocessed data in the preprocessed biological sign parameter set, z b Y is the survival status index of the animal corresponding to the pre-processed data of group b in the pre-processed biological sign parameter set, mi is the i-th biological sign parameter value in the m-th group of preprocessed data in the preprocessed biological sign parameter set, Y mj is the jth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, Y mn is the nth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, y mn is the nth biological sign parameter value in the mth group of predicted data in the biological sign parameter prediction set, Y is the sum of the standard deviations of the biological sign parameters in the mth group of preprocessed data in the preprocessed biological sign parameter set, bis the preprocessed data of group b; similarly, when k=2, M2 is the number of groups of qualified data in the qualified data set, and when k=3, M3 is the number of groups of non-qualified data in the non-qualified data set;
[0045] The survival state is obtained based on the survival state index set and the survival state classification model.
[0046] Preferably, the warning mechanism includes yellow warning, orange warning and red warning;
[0047] When the survival state is the normal state, no warning is issued; when the survival state is the panic state, the yellow warning is issued; when the survival state is the depressed state, the orange warning is issued; when the survival state is the death state, the red warning is issued;
[0048] When any one of the environmental warning signal, the biological sign warning signal, the orange warning signal or the red warning signal is issued, a reminder signal is issued, and the breeder checks the living area where the animal is located based on the reminder signal.
[0049] Preferably, the method further comprises:
[0050] The missing data in the environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set are supplemented to obtain the supplemented environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set; the supplementation formula is as follows:
[0051]
[0052] Where: x it is the missing i-th environmental parameter data at time t, x it-j is the jth forward data of the i-th environmental parameter data at time t, x it+j is the jth backward data of the i-th environmental parameter data at time t, λ j For x j The weight value of
[0053] Correcting the supplemented environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set to obtain corrected environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set;
[0054] when When x kt Exception, for x kt Make corrections; where xkt represents the value of the kth parameter at time t, x kq represents the value of the kth parameter at time q, The kth parameter The mean of the group;
[0055] The correction formula is as follows:
[0056]
[0057] Where: x′ kt For x kt Correction value, x at represents the value of the ath parameter at time t, x aq represents the value of the ath parameter at time q, x qk is the value of the kth parameter at time q, is the total number of parameters.
[0058] An animal survival status early warning system, comprising:
[0059] A data acquisition module collects various environmental parameters in the living area where the animal is located based on environmental sensors to obtain a real-time environmental parameter data set, and obtains various human factors in the living area where the animal is located to obtain a real-time initial human factor data set; the real-time environmental parameter data set and the real-time initial human factor data set both include real-time, dynamic and continuous data;
[0060] A data processing module is used to complete missing data in the real-time environmental parameter data set and the real-time initial human factor data set, and to correct abnormal data;
[0061] A collection frequency module, which sets a monitoring frequency based on changes in the real-time environmental parameter data set and the real-time initial human factor data set, and obtains an environmental parameter prediction data set and a human factor prediction data set based on the monitoring frequency;
[0062] A biological sign prediction module, based on the environmental parameter prediction data set and the human factor prediction data set, combined with a biological sign parameter prediction model, obtains a biological sign parameter prediction set; and stores the biological sign parameter prediction set;
[0063] A survival status prediction module, based on the biological sign parameter prediction set and combined with the survival status classification model, obtains the survival status of the animal;
[0064] The early warning module monitors and warns the survival status of animals based on the survival status and in combination with the early warning mechanism.
[0065] An animal survival status early warning device, the early warning device is applicable to the above-mentioned early warning method, the early warning device comprises: a receiving device, a sending device, a processor and a communication bus;
[0066] The communication bus enables communication between the receiving device, the sending device and the processor.
[0067] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0068] The present invention relates to an animal survival status early warning method, system and equipment. First, various environmental parameters in the survival area where the animal is located are collected based on environmental sensors to obtain a real-time environmental parameter data set, and various human factors in the survival area where the animal is located are obtained to obtain a real-time initial human factor data set; then, based on the changes in the real-time environmental parameter data set and the real-time initial human factor data set, a monitoring frequency is set, and an environmental parameter prediction data set and a human factor prediction data set are obtained based on the monitoring frequency; further based on the environmental parameter prediction data set and the human factor prediction data set, combined with a biological sign parameter prediction model, a biological sign parameter prediction set is obtained; based on the biological sign parameter prediction set, combined with a survival status classification model, the survival status of the animal is obtained; finally, based on the survival status, combined with an early warning mechanism, the survival status of the animal is monitored and warned. The present invention combines environmental parameters and human factors to predict the survival status, which reduces the cost and improves the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0069] 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 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 labor.
[0070] Figure 1 This is a schematic diagram of the process of the animal survival status early warning method of the present invention;
[0071] Figure 2 This is a structural diagram of the animal survival status monitoring system of the present invention;
[0072] Figure 3 This is a schematic diagram of the animal survival status monitoring system of the present invention;
[0073] Figure 4 This is a specific flow chart of the animal survival status early warning method of the present invention;
[0074] Figure 5 This is a structural diagram of the animal survival status early warning device of the present invention.
[0075] Explanation of symbols: 201 - data acquisition module, 202 - data processing module, 203 - acquisition frequency module, 204 - biological sign prediction module, 205 - survival status prediction module, 206 - early warning module, 301 - receiving device, 302 - sending device, 303 - processor, 304 - memory, 305 - communication bus. DETAILED DESCRIPTION
[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.
[0077] The purpose of the present invention is to provide an animal survival status early warning method, system and equipment, which predicts changes in animal physiological and behavioral characteristics through environmental factors combined with human operation parameters, and further predicts the animal survival status, thereby achieving low-cost and high-precision survival status prediction.
[0078] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0079] Figure 1 This is a flow chart of the animal survival status early warning method of the present invention; Figure 4 The following is a specific flow chart of the animal survival status early warning method of the present invention. Figure 1 and Figure 4 The present invention provides an animal survival status early warning method, comprising:
[0080] Step S1, collecting various environmental parameters in the living area where the animal is located based on environmental sensors to obtain a real-time environmental parameter data set, obtaining various human factors in the living area where the animal is located to obtain a real-time initial human factor data set.
[0081] The environmental parameters include ambient temperature, relative humidity, mixed gas concentration, light intensity and noise. The mixed gas concentration includes the concentration of carbon dioxide, nitrogen and ammonia in the ambient air of the animal. The real-time environmental parameter data set is collected by multiple environmental sensors and a noise meter.
[0082] The human factors include feeding, passing of people, construction, chasing, capturing, depilating, blood drawing and feed gathering during the breeding process of target animals; the feeding includes feeding concentrated feed, feeding grass, artificial feed gathering and feeding other livestock and poultry animals; the artificial feed gathering can be divided into human activities in three different time periods before feeding, during feeding and after feeding; the construction can be divided into construction inside the house and construction outside the house; the construction also includes transporting livestock and poultry, transporting manure, cleaning the house and transporting feed, etc.
[0083] Since the environmental parameter data obtained by the sensor is within a large area, and under the same environmental conditions, the corresponding survival status of mutton sheep in different growth cycles and different genders may not be the same, it is necessary to divide the areas where mutton sheep with the same or similar survival status live, so as to facilitate the subsequent establishment of a biological sign status assessment model. That is, the living area of the animals includes the activity area of animals in the same growth cycle and with an average living distance of no more than 5 meters and the activity area of animals of the same gender with an average living distance of no more than 5 meters.
[0084] Step S2, based on the changes of the real-time environmental parameter data set and the real-time initial human factor data set, a monitoring frequency is set, and based on the monitoring frequency, an environmental parameter prediction data set and a human factor prediction data set are obtained.
[0085] The real-time initial human factor data set is converted from text form to vector form to obtain a real-time human factor data set; specifically, a vocabulary set V is defined according to the interference type, and each real-time initial human factor data X in the real-time initial human factor data set is represented as a vector X∈R of |V| dimension. |V| , R is a positive real number, R |V| express |V| dimensional Euclidean space; the value of the i-th dimension in each interference vector x in the real-time initial human factor data indicates whether the i-th word in the vocabulary set appears in x. If it appears, the value is 1, otherwise it is 0. For example, start catching sheep and stop catching sheep have three words: start, stop and catch sheep, then start catching sheep is represented by [1 1 1] T , stop catching sheep is represented by [1 0 1] T , T is transposed, and the real-time human factor data set is obtained.
[0086] In this embodiment, different human factor buttons such as start catching sheep, start feeding, start construction, start depilation, stop catching sheep, stop feeding, stop construction and stop depilation are designed to obtain human factor data. When the staff performs the corresponding human activities, they press the corresponding buttons to obtain the human factor data. For example, when the staff is catching sheep, they press the start catching sheep button to obtain the start catching sheep data.
[0087] Furthermore, when there is missing data in the real-time environmental parameter data set, the missing data in the real-time environmental parameter data set is supplemented; the supplement formula is as follows:
[0088]
[0089] Where: x it is the missing i-th environmental parameter data at time t, x it-j is the jth forward data of the i-th environmental parameter data at time t, x it+j is the jth backward data of the i-th environmental parameter data at time t, λ j For x j The weight value of .
[0090] Correcting the completed real-time environmental parameter data set to obtain a corrected real-time environmental parameter data set;
[0091] when When x kt Exception, for x kt Make corrections; where x kt represents the value of the kth parameter at time t, x kq represents the value of the kth parameter at time q, The kth parameter The mean of the group;
[0092] The correction formula is as follows:
[0093]
[0094] Where: x′ kt For x kt Correction value, x at represents the value of the ath parameter at time t, x aq represents the value of the ath parameter at time q, x qk is the value of the kth parameter at time q, is the total number of parameters.
[0095] For any environmental parameter in the real-time environmental parameter data set, a set environmental change threshold is corresponding. The parameter value of the current environmental parameter at the start time t0 of the current collection cycle and the parameter value at t0+1 are obtained. If the parameter value change rate at t0 and t0+1 is greater than the corresponding set environmental change threshold, the current environmental parameter is continuously collected in the current collection cycle. If the parameter value change rate at t0 and t0+1 is less than or equal to the corresponding set environmental change threshold, the current environmental parameter is intermittently collected in the current collection cycle, and each environmental parameter is traversed to obtain the environmental parameter prediction data set; the monitoring frequency includes the continuous collection and the intermittent collection, the continuous collection is collected once every 5 seconds, and the intermittent collection is collected once every 1 minute. In this embodiment, each set environmental change threshold is 5%.
[0096] The parameter value change rate calculation formula is as follows:
[0097]
[0098] Where: α is the rate of change of parameter value, is the parameter value at time t0+1, t θ0 is the parameter value at time t0;
[0099] For any artificial factor in the real-time artificial factor data set, a set artificial change threshold is corresponding, and the factor value of the current artificial factor at the start time t0 of the current collection cycle and the factor value at t0+1 are obtained. If the change rate of the factor value at t0 and t0+1 is greater than the corresponding set artificial change threshold, the current artificial factor is continuously collected in the current collection cycle. If the change rate of the factor value at t0 and t0+1 is less than or equal to the corresponding set artificial change threshold, the current artificial factor is intermittently collected in the current collection cycle, and each artificial factor is traversed to obtain the artificial factor prediction data set. Each set artificial change threshold is 5%.
[0100] By setting the monitoring frequency, energy consumption can be reduced and the endurance of each sensor can be improved. For example, the flexible wearable biosensor is powered by a 3.7V / 2000mA lithium battery. After being fully charged, it can work for 6 hours under normal experimental conditions. By setting the monitoring frequency, the working time of the flexible wearable biosensor is 8 to 9 hours, which improves the endurance. The environmental parameter monitoring equipment is continuously powered by the socket. By setting the monitoring frequency, energy consumption is reduced by 25%-35%.
[0101] For any environmental parameter in the real-time environmental parameter data set, there is a corresponding set environmental threshold range. When the parameter value of any environmental parameter exceeds the corresponding set environmental threshold range, an environmental warning signal is issued.
[0102] Specifically, taking the ambient temperature during mutton breeding as an example, the minimum threshold of the ambient temperature is set to 8°C and the maximum threshold is set to 33°C. When the ambient temperature is less than 8°C or greater than 33°C, it can be judged that the living environment is abnormal, and an environmental warning is directly issued, issuing the environmental warning signal.
[0103] Step S3, based on the environmental parameter prediction data set and the human factor prediction data set, combined with the biological sign parameter prediction model, obtain a biological sign parameter prediction set.
[0104] Specifically, step S3 includes:
[0105] Step S31, obtaining an environmental parameter data training set, a human factor data training set and a biological sign parameter training set.
[0106] The environmental parameter data training set includes an environmental temperature data training set, a relative humidity data training set, a mixed gas concentration data training set, a light intensity data training set and a noise data training set.
[0107] The human factor data training set includes a feeding data training set, a personnel passing data training set, a construction data training set, a chasing data training set, a capturing data training set, a hair removal data training set, a blood drawing data training set and a material gathering data training set.
[0108] The biological sign parameter training set is obtained by collecting data from animals based on flexible wearable biosensors; the biological sign parameter training set includes a physiological parameter data training set and a behavioral characteristic parameter data training set; the physiological parameter data training set includes a heart rate data training set, a body temperature data training set, a blood pressure data training set and a respiratory data training set; the behavioral characteristic parameter data training set includes a step data training set, a standing and lying data training set, an acceleration data training set, a call data training set and an angular velocity data training set.
[0109] When there are missing and abnormal data in the environmental parameter data training set and the biological sign parameter training set, the method in "step S2" is used to complete and correct them.
[0110] The environmental parameter data training set, the human factor data training set and the biological sign parameter training set are preprocessed to obtain a preprocessed training set.
[0111] The preprocessing training set is extracted based on a random extraction method to obtain training samples, and the initial biological sign parameter prediction model is trained based on the training samples to obtain the biological sign parameter prediction model.
[0112] Specifically, the training samples include an environmental training set, an artificial training set and a biological sign training set.
[0113] Step S32, training the initial biological sign parameter prediction model based on the environmental training set, the artificial training set and the biological sign training set to obtain the biological sign parameter prediction model.
[0114] The environmental training set and the artificial training set are input into the initial biological sign parameter prediction model to obtain a training biological sign parameter prediction set.
[0115] The calculation formula of the training biological sign parameter prediction set is as follows:
[0116]
[0117] Among them, z m (x) is the training prediction value of the mth biological sign parameter; is the kth basis function, b is the number of basis functions, w m [k] is the connection weight of the kth basis function of the mth biological sign parameter; β is a constant; x is the set of environmental training set and artificial training set.
[0118] Based on the training biological sign parameter prediction set and the biological sign training set, an error value is obtained. The calculation formula of the error value is as follows:
[0119]
[0120] Where: ε is the error between the predicted value and the true value; z mj is the true value of the jth test sample of the mth biological sign parameter; z m (x j ) is the predicted value of the jth test sample of the mth biological sign parameter; x j is the input of the jth test sample; n is the number of input test samples.
[0121] If the error value is greater than or equal to the error setting value, the connection weight of the initial biological sign parameter prediction model is corrected, and iterative optimization is continued until the error value is less than the error setting value to obtain the biological sign parameter prediction model. The correction formula of the connection weight is as follows:
[0122]
[0123] is the initial connection weight of the kth basis function of the mth biological sign parameter, is the connection weight of the kth basis function of the mth biological sign parameter in the next iteration, w m[k] is the connection weight of the kth basis function of the mth biological sign parameter in the current iteration, α is a constant; z mj is the true value of the jth training sample of the mth biological sign parameter.
[0124] In this embodiment, temperature, humidity, light, carbon dioxide content, noise, sheep catching, construction, feeding and depilation are used as inputs, and heart rate, body temperature and acceleration are used as outputs. The prediction results of the biological sign parameter prediction model are tested based on 500 samples as test samples. The prediction test results are shown in Table 1.
[0125] Table 1. Test table of biological sign parameter prediction model
[0126]
[0127]
[0128] Step S33, preprocessing the environmental parameter prediction data set and the human factor prediction data set to obtain a prediction input matrix.
[0129] The specific calculation formula is as follows:
[0130]
[0131] Where: a` is the environmental parameter prediction data or human factor prediction data after preprocessing; a is the environmental parameter prediction data or human factor prediction data before preprocessing, a i is the value of the ith data in a certain environmental parameter prediction data or human factor prediction data before preprocessing, and m is the total number of data in a certain environmental parameter prediction data or human factor prediction data.
[0132] Step S34, inputting the prediction input matrix into the biological sign parameter prediction model to obtain the biological sign parameter prediction set.
[0133] Step S35, judging the biological sign parameter prediction set, when any biological sign parameter prediction value in the biological sign parameter prediction set exceeds its corresponding biological sign threshold range, issuing a biological sign warning signal; when any biological sign parameter prediction value is within its corresponding biological sign threshold range, obtaining the survival status based on the biological sign parameter prediction set.
[0134] Step S4, based on the biological sign parameter prediction set and combined with the survival state classification model, the survival state of the animal is obtained. The survival state includes normal state, panic state, listless state and death state.
[0135] The biological sign parameter prediction set is preprocessed to obtain a preprocessed biological sign parameter set. The specific formula is as follows:
[0136]
[0137] Where: y mi is the value of the i-th biological sign parameter in the m-th group of predicted data in the biological sign parameter prediction set, Y mi is the preprocessed value of the ith biological sign parameter in the mth group of prediction data in the biological sign parameter prediction set, and M is the number of groups of data in the biological sign parameter prediction set.
[0138] The pre-processed biological sign parameter set is classified to obtain a standard data set and a non-standard data set; the standard data set corresponds to the normal state and the panic state; the non-standard data set corresponds to the listless state and the death state.
[0139] The calculation formula for classifying the pre-processed biological sign parameter set is as follows:
[0140]
[0141] Where: is the survival status index obtained in the first prediction. When k = 1, the pre-treatment biological sign parameter set is taken The pre-processed data of each group is 1, and the qualified data set is obtained. The preprocessed data of each group is -1, and the substandard data set is obtained.
[0142] M1 is the number of groups of preprocessed data in the preprocessed biological sign parameter set, α m is the vector factor of the mth group of preprocessed data in the preprocessed biological sign parameter set, z m is the survival status index of the animal corresponding to the mth group of preprocessed data in the preprocessed biological sign parameter set, z b Y is the survival status index of the animal corresponding to the pre-processed data of group b in the pre-processed biological sign parameter set, mi is the i-th biological sign parameter value in the m-th group of preprocessed data in the preprocessed biological sign parameter set, Y mj is the jth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, Y mn is the nth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, y mn is the nth biological sign parameter value in the mth group of predicted data in the biological sign parameter prediction set, Y is the sum of the standard deviations of the biological sign parameters in the mth group of preprocessed data in the preprocessed biological sign parameter set, bPreprocess data for group b.
[0143] For the qualified data set, let k = 2 and use the following formula to calculate and obtain the second survival status index set:
[0144]
[0145] Where: is the normal or panic index, M2 is the number of groups of qualified data in the qualified data set, z m is the animal survival status index of the mth group of qualified data in M2, z b Y is the animal survival status index of the b-group of standard data in M2, mi is the i-th biological sign parameter value in the m-th group of standard-reaching data in M2, Y mj is the jth biological sign parameter value in the mth group of standard data in M2, Y mn is the nth biological sign parameter value in the mth group of standard data in M2, y mn is the nth biological sign parameter value in the mth group of predicted data in the biological sign parameter prediction set, Y is the sum of the standard deviations of the biological sign parameters in the mth group of standard data in M2, b This is the standard data for group b.
[0146] when When is 1, the survival state corresponding to the mth group of standard data is normal. When it is -1, the survival state corresponding to the mth group of standard data is a panic state.
[0147] For the substandard data set, let k = 3 and use the following formula to calculate and obtain the third survival status index set:
[0148]
[0149] Where: is the index of malaise or death, M3 is the number of groups of substandard data in the substandard data set, z m is the survival status index of the animals in the mth group of substandard data in M3, z b Y is the survival status index of the animals in group b that do not meet the standard in M3, mi is the i-th biological sign parameter value in the m-th group of substandard data in M3, Y mj is the jth biological sign parameter value in the mth group of substandard data in M3, Y mn is the nth biological sign parameter value in the mth group of substandard data in M3, y mn is the nth biological sign parameter value in the mth group of predicted data in the biological sign parameter prediction set, Y is the sum of the standard deviations of the biological sign parameters in the mth group of substandard data in M3, b These are the substandard data for group b.
[0150] when When is 1, the survival state corresponding to the mth group of substandard data is a depressed state. When it is -1, the survival state corresponding to the mth group of substandard data is the death state.
[0151] The survival status index set includes the second survival status index set and the third survival status index set.
[0152] Based on the survival state index set and the survival state classification model, the survival state is obtained. ;
[0153] The survival status is obtained based on the survival status achievement index and the survival status classification model.
[0154] In this embodiment, the biological sign parameters of the mutton sheep are body temperature 37.8°C, heart rate 72 times / minute and acceleration 0.1g, and the calculated survival state index is not up to standard, and the further calculated survival state is a depressed state. Based on 500 sets of test data, the method of the present invention and the existing traditional method are compared. The results are shown in Table 2. It can be seen that the method of the present invention has higher accuracy and faster running time.
[0155] Table 2 Comparison of prediction between traditional method and the present invention
[0156]
[0157] Step S5, based on the survival status and in combination with the early warning mechanism, monitoring and early warning are performed on the survival status of the animal.
[0158] The warning mechanism includes yellow warning, orange warning and red warning;
[0159] When the survival status is the normal state, no warning is issued; when the survival status is the panic state, the yellow warning is issued; when the survival status is the listless state, the orange warning is issued; and when the survival status is the death state, the red warning is issued.
[0160] When any one of the environmental warning signal, the biological sign warning signal, the orange warning signal or the red warning signal is issued, a reminder signal is issued, and the breeder checks the living area where the animal is located based on the reminder signal.
[0161] Figure 2 This is a structural diagram of the animal survival status monitoring system of the present invention. Figure 3The schematic diagram of the animal survival status monitoring system of the present invention is shown in FIG. Figure 2 and Figure 3 As shown, the present invention provides an animal survival status early warning system, comprising:
[0162] The data acquisition module 201 collects various environmental parameters in the living area of the animal based on environmental sensors to obtain a real-time environmental parameter data set, and obtains various human factors in the living area of the animal to obtain a real-time initial human factor data set; the real-time environmental parameter data set and the real-time initial human factor data set both include real-time, dynamic and continuous data; the data acquisition module 201 stores the real-time environmental parameter data set and the real-time initial human factor data set.
[0163] The data processing module 202 completes the missing data in the real-time environmental parameter data set and the real-time initial human factor data set, and corrects the abnormal data.
[0164] The acquisition frequency module 203 sets a monitoring frequency based on the changes in the real-time environmental parameter data set and the real-time initial human factor data set, and obtains the environmental parameter prediction data set and the human factor prediction data set based on the monitoring frequency. The acquisition frequency module 203 can reduce the power consumption of the data acquisition module 201.
[0165] The biological sign prediction module 204 obtains a biological sign parameter prediction set based on the environmental parameter prediction data set and the human factor prediction data set in combination with a biological sign parameter prediction model; and stores and transmits the biological sign parameter prediction set.
[0166] The survival status prediction module 205 obtains the survival status of the animal based on the biological sign parameter prediction set and the survival status classification model.
[0167] The early warning module 206 monitors and warns the survival status of the animal based on the survival status and in combination with the early warning mechanism.
[0168] The visual output module displays the survival status. When the survival status is abnormal, the alarm signal and alarm time are displayed, and corresponding abnormal data are provided to promptly remind the breeder to take countermeasures.
[0169] Specifically, the data acquisition module 201 includes the functions of data information acquisition, data information recording and data information transmission; the acquisition frequency module 203 has the functions of information processing, information storage, information transmission and information download. The biological sign prediction module 204 has the functions of data reception, data analysis, data storage and data download. The early warning module 206 has a decision-making alarm function, which can issue an alarm when the survival state is abnormal, and can directly issue an early warning when the data obtained by the acquisition frequency module 203 and the biological sign prediction module 204 exceeds the set threshold.
[0170] The number of the data acquisition module 201, the data processing module 202, the acquisition frequency module 203, the biological sign prediction module 204, the survival status prediction module 205, the early warning module 206 and the visual output interface is selected according to actual needs.
[0171] Figure 5 As shown in the figure, the present invention provides an animal survival status warning device, which is applicable to the above-mentioned warning method, and includes: a receiving device 301, a sending device 302, a processor 303, a memory 304 and a communication bus 305.
[0172] The communication bus 305 implements communication between the receiving device 301, the sending device 302, the processor 303 and the memory 304. The number of the receiving device 301, the sending device 302, the processor 303 and the memory 304 is selected according to actual needs.
[0173] An early warning device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. The early warning device is any one of a microprocessor, a dedicated integrated circuit, a programmable gate array, a digital processor and an embedded device; the early warning device may also include a network device and / or a user device; wherein the network device includes but is not limited to a single network server, a server group consisting of multiple network servers, or a cloud platform based on cloud computing consisting of a large number of hosts or network servers, wherein cloud computing is a type of distributed computing, a super virtual computer consisting of a group of loosely coupled computers.
[0174] The receiving device 301 and the sending device 302 can be wired sending ports or wireless devices, for example, including an antenna device for communicating data with other devices; the memory 304 is used to store program codes; the memory 304 can be a circuit with storage function in an integrated circuit without a physical form, such as a random access memory and a FIFO, etc.; or, the memory 304 can also be a memory with a physical form, such as a memory stick, a smart media card and a secure digital card, etc.; the processor 303 can include one or more microprocessors and digital processors; the processor 303 can call the program code stored in the memory to execute related functions.
[0175] The present invention can realize accurate grasp of the survival status of target animals in the breeding process.
[0176] The present invention can improve the animal welfare and quality traceability level in the breeding stage.
[0177] The invention can ensure the growth and reproduction performance of animals and the economic benefits of breeding farms.
[0178] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.
[0179] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of the present invention. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.
Claims
1. A method for early warning of animal survival status, characterized in that: include: Based on the environmental sensors, various environmental parameters in the living area where the animals are located are collected to obtain a real-time environmental parameter data set, and various human factors in the living area where the animals are located are obtained to obtain a real-time initial human factor data set; Based on the changes of the real-time environmental parameter data set and the real-time initial human factor data set, a monitoring frequency is set, and based on the monitoring frequency, an environmental parameter prediction data set and a human factor prediction data set are obtained, including: The real-time initial human factor data set is converted from text form to vector form to obtain a real-time human factor data set; specifically, a vocabulary set V is defined according to the interference type, and each real-time initial human factor data X in the real-time initial human factor data set is represented as a vector X∈R of |V| dimension. V , R is a positive real number, R |V| represents |V|-dimensional Euclidean space; the value of the i-th dimension in each interference vector x in the real-time initial human factor data represents whether the i-th word in the vocabulary set appears in x, if it appears, the value is 1, otherwise it is 0, and the real-time human factor data set is obtained; For any environmental parameter in the real-time environmental parameter data set, a set environmental change threshold is corresponding. The parameter value of the current environmental parameter at the start time t0 of the current collection cycle and the parameter value at t0+1 are obtained. If the parameter value change rate at t0 and t0+1 is greater than the corresponding set environmental change threshold, the current environmental parameter is continuously collected in the current collection cycle. If the parameter value change rate at t0 and t0+1 is less than or equal to the corresponding set environmental change threshold, the current environmental parameter is intermittently collected in the current collection cycle, and each environmental parameter is traversed to obtain the environmental parameter prediction data set; the monitoring frequency includes the continuous collection and the intermittent collection; the parameter value change rate calculation formula is as follows: Where: α is the rate of change of parameter value, is the parameter value at time t0+1, t θ0 is the parameter value at time t0; For any human factor in the real-time human factor data set, a set human change threshold is corresponding, and the factor value of the current human factor at the start time t0 of the current collection cycle and the factor value at time t0+1 are obtained. If the factor value change rate at time t0 and time t0+1 is greater than the corresponding set human change threshold, the current human factor is continuously collected in the current collection cycle. If the factor value change rate at time t0 and time t0+1 is less than or equal to the corresponding set human change threshold, the current human factor is intermittently collected in the current collection cycle, and each human factor is traversed to obtain the human factor prediction data set; For any environmental parameter in the real-time environmental parameter data set, a set environmental threshold range corresponds to each of them. When the parameter value of any environmental parameter exceeds the corresponding set environmental threshold range, an environmental warning signal is issued; Based on the environmental parameter prediction data set and the human factor prediction data set, combined with the biological sign parameter prediction model, a biological sign parameter prediction set is obtained, including: Obtaining environmental parameter data training sets, human factor data training sets and biological sign parameter training sets; Constructing an initial biological sign parameter prediction model based on the environmental parameter data training set, the human factor data training set and the biological sign parameter training set, and training the initial biological sign parameter prediction model to obtain the biological sign parameter prediction model; Preprocessing the environmental parameter prediction data set and the human factor prediction data set to obtain a prediction input matrix; Inputting the prediction input matrix into the biological sign parameter prediction model to obtain the biological sign parameter prediction set; The biological sign parameter prediction set is judged, and when any biological sign parameter prediction value in the biological sign parameter prediction set exceeds its corresponding biological sign threshold range, a biological sign warning signal is issued; when any biological sign parameter prediction value is within its corresponding biological sign threshold range, the survival status is obtained based on the biological sign parameter prediction set; The survival state includes a normal state, a panic state, a depressed state and a death state; The method of obtaining the animal's survival status based on the biological sign parameter prediction set and combining the survival status classification model includes: The biological sign parameter prediction set is preprocessed to obtain a preprocessed biological sign parameter set. The specific formula is as follows: Where: y mi is the value of the i-th biological sign parameter in the m-th group of predicted data in the biological sign parameter prediction set, Y mi is the preprocessed value of the ith biological sign parameter in the mth group of prediction data in the biological sign parameter prediction set, and M is the number of groups of data in the biological sign parameter prediction set; Classifying the preprocessed biological sign parameter set to obtain a qualified data set and a non-qualified data set; the qualified data set corresponds to the normal state and the panic state; the non-qualified data set corresponds to the depressed state and the death state; Based on the pre-processed biological sign parameter set, the animal survival status index set is obtained, and the calculation formula is as follows: Where: is the survival status index obtained by the k-th prediction, k∈1,2,3, when k=1, the pre-processing biological sign parameter set is taken The pre-processed data of each group is 1, and the qualified data set is obtained. For each group of pre-processed data with k=-1, a substandard data set is obtained. For the standard data set, k=2 is set and the calculation is continued to obtain a second survival state index set. For the substandard data set, k=3 is set and the calculation is continued to obtain a third survival state index set. The survival state index set includes the second survival state index set and the third survival state index set. When k = 1: M1 is the number of groups of preprocessed data in the preprocessed biological sign parameter set, α m is the vector factor of the mth group of preprocessed data in the preprocessed biological sign parameter set, Z m is the survival status index of the animal corresponding to the mth group of preprocessed data in the preprocessed biological sign parameter set, Z b Y is the survival status index of the animal corresponding to the pre-processed data of group b in the pre-processed biological sign parameter set, mi is the i-th biological sign parameter value in the m-th group of preprocessed data in the preprocessed biological sign parameter set, Y mj is the jth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, Y mn is the nth biological sign parameter value in the mth group of preprocessed data in the preprocessed biological sign parameter set, y mn is the nth biological sign parameter value in the mth group of predicted data in the biological sign parameter prediction set, Y is the sum of the standard deviations of the biological sign parameters in the mth group of preprocessed data in the preprocessed biological sign parameter set, b is the preprocessed data of group b; similarly, when k=2, M2 is the number of groups of qualified data in the qualified data set, and when k=3, M3 is the number of groups of non-qualified data in the non-qualified data set; Based on the survival state index set and the survival state classification model, the survival state is obtained. Based on the biological sign parameter prediction set and in combination with the survival status classification model, the survival status of the animal is obtained; Based on the survival status, combined with the early warning mechanism, the survival status of the animals is monitored and warned.
2. The animal survival status early warning method according to claim 1, characterized in that: The environmental parameters include ambient temperature, relative humidity, mixed gas concentration, light intensity and noise; The human factors include feeding, passing of people, construction, chasing, catching, depilation, blood drawing and gathering of feed during animal breeding; The living areas of the animals include: the activity areas of animals that are in the same stage of their growth cycles and whose average living distance does not exceed 5 meters, and the activity areas of animals of the same sex whose average living distance does not exceed 5 meters.
3. The animal survival status early warning method according to claim 1, characterized in that: The environmental parameter data training set includes an environmental temperature data training set, a relative humidity data training set, a mixed gas concentration data training set, a light intensity data training set and a noise data training set; The human factor data training set includes a feeding data training set, a personnel passing data training set, a construction data training set, a chasing data training set, a capturing data training set, a hair removal data training set, a blood drawing data training set and a material gathering data training set; The biological sign parameter training set is obtained by collecting data from animals based on a flexible wearable biosensor; the biological sign parameter training set includes a physiological parameter data training set and a behavioral characteristic parameter data training set; The physiological parameter data training set includes a heart rate data training set, a body temperature data training set, a blood pressure data training set and a breathing data training set; the behavioral characteristic parameter data training set includes a step data training set, a standing and lying data training set, an acceleration data training set, a call data training set and an angular velocity data training set; Preprocessing the environmental parameter data training set, the human factor data training set and the biological sign parameter training set to obtain a preprocessed training set; Extracting the preprocessed training set based on a random extraction method to obtain training samples, and training the initial biological sign parameter prediction model based on the training samples to obtain the biological sign parameter prediction model; The environmental parameter prediction data set and the human factor prediction data set are preprocessed to obtain a prediction input matrix, and the specific calculation formula is as follows: Where: a` A certain environmental parameter prediction data or human factor prediction data after preprocessing; a is a certain environmental parameter prediction data or human factor prediction data before preprocessing, a i is the value of the ith data in a certain environmental parameter prediction data or human factor prediction data before preprocessing, and m is the total number of data in a certain environmental parameter prediction data or human factor prediction data.
4. The animal survival status early warning method according to claim 1, characterized in that: The warning mechanism includes yellow warning, orange warning and red warning; When the survival state is the normal state, no warning is issued; when the survival state is the panic state, the yellow warning is issued; when the survival state is the depressed state, the orange warning is issued; when the survival state is the death state, the red warning is issued; When any one of the environmental warning signal, the biological sign warning signal, the orange warning signal or the red warning signal is issued, a reminder signal is issued, and the breeder checks the living area where the animal is located based on the reminder signal.
5. The animal survival status early warning method according to claim 1, characterized in that: The method further comprises: The missing data in the environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set are supplemented to obtain the supplemented environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set; the supplementation formula is as follows: Where: x it is the missing i-th environmental parameter data at time t, x it-j is the jth forward data of the i-th environmental parameter data at time t, x it+j is the jth backward data of the i-th environmental parameter data at time t, λ j For x j The weight value of Correcting the supplemented environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set to obtain corrected environmental parameter data training set, the biological sign parameter training set and the environmental parameter prediction data set; when When x kt Exception, for x kt Make corrections; where x kt represents the value of the kth parameter at time t, x kq represents the value of the kth parameter at time q, Represents the kth parameter The mean of the group; The correction formula is as follows: Where: For x kt Correction value, x at represents the value of the ath parameter at time t, x aq represents the value of the ath parameter at time q, x qk is the value of the kth parameter at time q, is the total number of parameters.
6. An animal survival status early warning system, the early warning system is applicable to the early warning method according to any one of claims 1 to 5, characterized in that: include: A data acquisition module collects various environmental parameters in the living area where the animal is located based on environmental sensors to obtain a real-time environmental parameter data set, and obtains various human factors in the living area where the animal is located to obtain a real-time initial human factor data set; the real-time environmental parameter data set and the real-time initial human factor data set both include real-time, dynamic and continuous data; A data processing module is used to complete missing data in the real-time environmental parameter data set and the real-time initial human factor data set, and to correct abnormal data; A collection frequency module, which sets a monitoring frequency based on changes in the real-time environmental parameter data set and the real-time initial human factor data set, and obtains an environmental parameter prediction data set and a human factor prediction data set based on the monitoring frequency; A biological sign prediction module, based on the environmental parameter prediction data set and the human factor prediction data set, combined with a biological sign parameter prediction model, obtains a biological sign parameter prediction set; and stores the biological sign parameter prediction set; A survival status prediction module, based on the biological sign parameter prediction set and combined with the survival status classification model, obtains the survival status of the animal; The early warning module monitors and warns the survival status of animals based on the survival status and in combination with the early warning mechanism.
7. An animal survival status early warning device, the early warning device is applicable to the early warning method according to any one of claims 1 to 5, characterized in that: The early warning device includes: a receiving device, a sending device, a processor and a communication bus; The communication bus enables communication between the receiving device, the sending device and the processor.
