Field of the invention

The present invention relates to a device and method for determining an animal’ activity of and, in particular, to determine whether the animal is in a state of rumination and in a state of food consumption.

Background art

Current dairy farming requires knowledge of each animal’s activity features, in particular, food consumption and rumination. Any changes or deviations in the behavior of the animal require immediate identification, as they may indicate an animal's health problems that affect its productivity.

In this regard, devices attached to animals have been developed that allow the tracking of animal activity in real time.

US 2016/0165851 A1 discloses a device that is attached to the animal’s neck and contains an accelerometer that generates the first and second signals indicating the animal’s movement and the raised and lowered position of the animal's head. Based on the counting of the number of positive and negative peaks of these signals, it is determined whether the animal is in a state of rumination, rest, or feeding.

The disadvantage of this device is the lack of accuracy in determining the state of the animal’s activity.

WO 2015/036303 A1 discloses a device for determining the animal’s rumination that is attached to an animal. The device contains an accelerometer and a microprocessor. The microprocessor processes the signal coming from the accelerometer by removing the constant component from it, periodically determines the value of the received signal in the selected frequency range and compares this value with the threshold value. When the signal is exceeded in the selected frequency range of the threshold value, it is concluded that the animal is in a state of rumination.

The said device does not allow determining other states of the animal’s activity, except for rumination. In addition, the said device does not have a sufficiently high accuracy of determining the state of the animal’s activity. Thus, it is necessary to create a method and device that allows increasing the accuracy of determining the state of the animal’s activity.

Substance of the invention

In one aspect, to solve the problem of creating a device and a method for improving the accuracy of determining the state of the animal's activity, a device is proposed for determining the animal’s activity, comprising: a three-axis accelerometer attached to the animal's neck and configured to obtain data containing acceleration values; a data processing unit connected to the accelerometer and configured to process data from the accelerometer containing acceleration values; characterized in that the data processing unit is additionally configured to: filter incoming data using a third-order Chebyshev filter with a cutoff frequency of 0.5 -2.5 Hz; averaging the data in a given interval; comparing the averaged data with an adaptively calculated threshold value and determining, based on a comparison, the state of the animal’s activity.

In another embodiment, it is possible to mount sensors on other parts of the animal’s body, for example, on the ear or tail. In this case, data from such devices are transmitted to the main device, which is mounted on the neck of the animal and is the central computing unit of the entire system.

In one embodiment of the device, the predetermined interval is approximately 30 seconds, and the adaptively calculated threshold value is the minimum value of the averaging calculated over a period of at least 30 seconds preceding the specified predetermined interval; however, the data processing unit is configured to determine whether the animal is in a state of food consumption, if the average value exceeds the threshold value.

In another embodiment of the device, the specified interval is 20-50 seconds, and the adaptively calculated threshold value is the average value of the obtained filtered data in the interval of 0.5-2 seconds, the origin of which is ahead of the origin of the specified interval by 8-16 seconds; wherein the data processing unit is adapted to determine whether the animal is in the rumination state if the average value exceeds the threshold value.

In yet another embodiment of the device, specific values for the durations of the intervals and the advance values for each animal are selected based on the training by comparing the results obtained by the processing unit with the actual state of the animal’s activity.

In another aspect of the invention, a method for determining the activity of an animal is proposed, comprising: obtaining data containing acceleration values using a three-axis accelerometer attached to the neck of the animal; process the data from the accelerometer containing the acceleration values using a processing unit connected to the accelerometer; characterized in that at the data processing stage: they filter incoming data using a third-order Chebyshev filter with a cutoff frequency of 0.5-2.5 Hz; average the data obtained at a given interval; compare the averaged data with an adaptively calculated threshold value and determine the state of the animal’s activity based on a comparison.

In one embodiment, the method specified interval is approximately 30 seconds, and the adaptively calculated threshold value represents the minimum value of the averaging calculated over a period of at least 30 seconds preceding the specified predetermined interval; however, the method further comprises determining, by using a data processing unit, the animal being in a state of food consumption, if the average value exceeds the threshold value.

In another embodiment, the method specified interval is 20-50 seconds, and the adaptively calculated threshold value represents the average value of the obtained filtered data in the interval of 0.5-2 seconds, the origin of which is ahead of the origin of the specified interval by 8-16 seconds; however, the method further comprises a stage at which it is determined, using a data processing unit, that the animal is in a state of rumination, if the average value exceeds the threshold value.

In yet another embodiment, the method specific values for the durations of the intervals and the advance values for each animal is selected based on the training by comparing the results obtained by the processing unit with the actual state of the animal’s activity.

Brief description of the drawings

Fig. 1 represents an example of the attachment of a three-axis accelerometer to the animal’s neck.

Fig. 2 represents graphs showing the source data of the animal, coming from the accelerometer, and the data obtained by the conversion to determine the activity of the animal

Fig. 3 represents a graph showing the data obtained by the conversion to determine rumination.

Fig. 4 represents a graph illustrating the extraction of peaks from the converted data to determine the rumination.

Fig. 5 represents a graph showing the comparison of peaks with a threshold value to determine the state of rumination.

Implementation of the invention

To determine the state of activity of the animal, a three-axis accelerometer is mounted on the neck of the animal using a collar mount, as shown in Fig. 1. Data from the accelerometer in the form of acceleration values are received in the information processing unit. In the future, the processed data is transmitted to the center of accumulation and complex analysis of data over the air. This data is used to determine the state of the animal’s activity, the state of food consumption and the state of rumination.

In one embodiment, the initial data of the acceleration values along the three axes enter the information processing unit at a frequency of 28 and 55 measurements per second.

To determine the activity of the animal, these data are processed by a Chebyshev filter of the third frequency of the third order with a cutoff frequency of 0.4 Hz. Then the processed data is integrated over a period of 1 minute by summing the increments of adjacent measurements of the values obtained at the output of the specified filter. The device calculates the activity of the animal, as an average for the interval of 10, 15 and 30 minutes. The calculated value is transmitted to the server, where it is analyzed together with other parameters obtained from the device.

In one embodiment, the data conversion is performed according to the following formula:

where

y is the output; x - input data, n - number of measurement;

a[l] = -2,8116; a[2] = 2,6405; a[3] = -0,8281;

b[0] = 0,0000954; b[l] = 0,0002863; b[2] = 0,0002863; b[4] = 0,0954.

Fig. 2 shows the data before and after the conversion.

Then the data obtained are summed at a given interval in accordance with the following formula:

where N is the number of samples for a given interval.

The above operation is performed for each of the axes of the accelerometer. The results are summarized and taken as the animal’s activity level.

In one embodiment, the source data is processed to determine if the animal is in a state of food consumption. With this treatment, uniform periodic movements of the jaws of the animal are detected in the frequency range 0.5 -2.5 Hz, which lasts at least 30 seconds. The source data obtained by a three-axis accelerometer mounted on the neck of an animal, with a sampling frequency of 55 or 28 measurements per second, is processed with a third-order Chebyshev bandpass filter with cutoff frequencies of 0.5-2.5 Hz. The processed data is subjected to straightening and averaging according to the rules of the moving average for a period of 30 seconds, after which the obtained value is compared with the adaptively calculated threshold value for at least 30 seconds. In this case, the threshold value is the minimum value of the averaging in the previous period.

The state of rumination is different from the state of chewing food in that in this state the animal makes gaps in chewing in order to burp the gum from the rumen. According to the duration of these intervals and the period between them, we can conclude that the animal is in a state of rumination.

In one embodiment, determining whether an animal is in a state of rumination consists in identifying uniform interruptions in the movement of the jaws of an animal in the frequency range 0.5-2.5 Hz with a duration of 0.5-8 seconds, occurring at intervals of 20-180 seconds. It uses the original data obtained by a three-axis accelerometer mounted on the animal’s neck, with a sampling rate of 55 or 28 measurements per second. The initial data is subjected to subsequent digital processing with a third-order Chebyshev bandpass filter with cutoff frequencies of 0.5- 2.5 Hz.

The results are averaged by the simple moving average rules at two intervals. The first, long interval is 20-50 seconds. The second, short interval is 0.5-2 seconds. The beginning of the countdown of the long interval is 8-16 seconds behind the countdown of the short interval. If the value of a simple moving average of a long interval exceeds the value of a simple moving average of a short interval 1.5-2.5 times within 0.5-8 seconds, then the decision is made whether the animal is in a state of rumination.

The values of intervals, delays and exceeding the values of a simple moving average of a long interval of a simple moving average of a short interval can be adjusted during the training period, which can last 1-3 days. During this period, parallel calculations are performed to find the animal in a state of rumination using several values of the above parameters. During this period, the animals are monitored and periods of the animal being in a rumination state are noted. The parameters at which use the estimated finding of the animal in the state of rumination as closely as possible with the observed states are used to further calculate the location of the animal in the state of rumination.

In one embodiment, the determination of the location of the animal in a state of rumination is as follows.

The data received from the accelerometer is converted in accordance with the following formula:

where y is the output value; x is the input value; n is the measurement number; a[l]= -4,8480; a[2]= 10,2498; a[3]= -12,0485; a[4]= 8,3007; a[5]= -3,1805; a[6]= 0,5321; b[0] = 0,0029; b[l] = 0; b[2] = -0,0087; b[3]= 0; b[4]= -0,0087; b[5]= 0; b[6]= -0,0029.

The data obtained after processing is presented in Fig. 3

Then, the peak values of the positive pulses are calculated, the peaks at a distance of less than 10 to 17 samples are ignored, and as a result we obtain an array of data, an example of which is shown in Fig. 4.

The obtained data is compared with a threshold value of 50-30% of the average peak value in 40-60 seconds, and an array of data is obtained in which units correspond to exceeding the threshold, zeros correspond to values below the threshold level (Fig. 5). The zero sequences with a duration of less than 1-3 counts are ignored (replaced by units).

The dip in the graph in Fig. 5 corresponds to the absence of chewing, this can also occur due to regurgitation of animals by animals, which is a sign of rumination. To make a decision on whether the animal is in a rumination state, the duration of the failures and the distance between them is calculated. The duration of the dip should be in the range of 0.4-3 s, the distance between the dips should be in the range of 35-80 seconds. If these conditions are met, then the rumination counter is increased by the corresponding number of time intervals.