METHOD FOR OBTAINING INFORMATION ABOUT A FARM ANIMAL

Abstract

The invention relates to a method for obtaining information about a farm animal, wherein a device (a) is attached to the head region of an animal, said device containing at least one acceleration sensor by means of which recurring acceleration data is measured, wherein the acceleration data is evaluated using automatic data processing means, and, as an outcome of the evaluation, is indicative of the activities and/or conditions of the animal. The acquired acceleration data is evaluated in order to detect the swallowing processes carried out by the animal.

Claims

1. A method for obtaining information about a farm animal, wherein a device is attached to the head area of an animal, which device contains at least one acceleration sensor by means of which acceleration data is measured repeatedly, wherein the acceleration data is evaluated with automated data processing means and activities and/or states of the animal are inferred as a result of the evaluation, wherein the recorded acceleration data is evaluated to the effect that swallowing processes which the animal carries out are detected.

2. The method as claimed in claim 1, wherein acceleration data is evaluated to the effect that time periods in which the animal swallows are differentiated from time periods in which the animal chews, and in that for this purpose the variants between the measured acceleration data within adjoining time windows is calculated and compared.

3. The method as claimed in claim 1, wherein the swallowing processes are counted.

4. The method as claimed in claim 1, wherein the recorded acceleration data is additionally evaluated to the effect that chewing actions which the animal carries out are detected and counted.

5. The method as claimed in claim 2, wherein the counting values which each relate to a previously determined time period, in particular to time periods which each have a duration of 24 hours, are generated.

6. The method as claimed in claim 4, wherein the number of chewing actions which take place in a time period are calculated by carrying out a Fourier transformation for this time period over the chronological profile of the measured accelerations, the fundamental frequency determined here is interpreted as being a chewing action frequency and is multiplied by the length of the time period.

7. The method as claimed in claim 6, wherein acceleration is measured in three respective coordinate directions, and in that for a multiplicity of different directions which are assumed in the space and for the acceleration vectors which are measured at the individual measuring times the absolute value of each directional component of the respective acceleration vector is calculated, which directional component is parallel to the respective direction assumed in the space, with the result that a data series is formed for each of the directions assumed in the space, over the chronological sequence of the acceleration components oriented in the respective direction, and in that the Fourier transformation is carried out at one or more of these data series.

8. The method as claimed in claim 5, wherein the time period over which chewing actions are counted extend from the end of one swallowing phase to the start of the next swallowing phase.

9. The method as claimed in claim 1, wherein swallowing processes which are part of a drinking process are counted separately.

10. The method as claimed in claim 9, wherein the process of drinking is identified by variance analysis of the measured acceleration data, wherein fluctuations in the variance profile within a predetermined frequency range are evaluated as an indication of a drinking process.

11. The method as claimed in claim 1, wherein said method is applied to a ruminant, and in that the acceleration data is also evaluated to the effect of detecting whether the swallowing processes and, if appropriate, chewing actions take place during the consumption of feed or during the rumination, and in that swallowing processes within the scope of the consumption of feed are counted separately from swallowing processes within the scope of ruminating.

12. The method as claimed 3 wherein an automated data processing operation one or more of the following information relating to the animal are generated on the basis of the measured counting values and further stored data items: number of swallowing processes per day, number of swallowing processes during the consumption of feed per day, ratio of numbers of chewing actions per swallowing process during the rumination, ratio of numbers of chewing actions per swallowing process during the consumption of feed, number of swallowing processes while drinking, deviations from the average value of the individual previously mentioned numbers and ratios of numbers with respect to a standardized value and/or an average value for a herd, deviations from the average value of the previously mentioned numbers and ratios of numbers with respect to the average value for the same animal in preceding time periods, information about a state of the animal for which the previously mentioned numbers and/or ratios of numbers and/or deviations are an indication, in particular information about the time of mating, information about the predicted optimum insemination time period, information about the ratio of the long fiber content with respect to the protein content in the feed, information about the supposed milk yield, information about the state of health.

13. The method as claimed in claim 1, wherein the farm animal is a ruminant, and in that the recorded acceleration data is evaluated to the effect that the measured value of a quantifiable variable which describes an instantaneous state of a ruminating process is determined.

14. The method as claimed in claim 13, wherein the quantified data which is obtained by evaluation relates to the frequency of the chewing actions during the rumination.

15. The method as claimed in claim 13, wherein evaluation of the acceleration data it is inferred whether the individual acceleration data items are to be assigned to a chewing phase or a swallowing/regurgitating phase of a ruminating process, and in that further evaluations are carried out only with that measurement data which is assigned to a chewing phase.

16. The method as claimed in claim 15, wherein the variance of the measurement data or a variable which correlates to the variance of the measurement data is calculated over two time windows which adjoin one another and which each relate to a partial duration of the measuring duration of acceleration data, and in that a pronounced capability of differentiating the two variable values calculated therewith is interpreted as a change between a chewing phase and a swallowing/regurgitating phase during the rumination.

17. The method as claimed in claim 15, wherein the evaluation is used for the purpose of differentiating the assignment of the measurement data to a chewing phase or a swallowing/regurgitating phase on the basis of the series of those chronologically successive acceleration values which each signify the absolute value of the total acceleration at a measuring time and can be calculated for the respective measuring time by geometric addition of the individual acceleration values measured in the individual coordination directions at the respective measuring time.

18. The method as claimed in claim 7, wherein each case a Fourier transformation is carried out for a multiplicity of data series which are each assigned to another direction in the space, and in that from the obtained multiplicity of results that result is filtered out in which in each frequency range in which the frequency of the chewing actions can in principle occur, a single oscillation dominates to the greatest extent in terms of amplitude over all the other oscillations in this frequency range, and in that the frequency of this dominant oscillation is interpreted as the frequency of the chewing actions.

19. The method as claimed in claim 7, wherein each case a Fourier transformation is carried out for a multiplicity of data series which are each assigned to another direction in the space, and in that from the obtained number of results that result is filtered out in which, in each frequency range in which the frequency of the chewing actions can in principle occur, three oscillations which occur one next to the other with respect to frequency dominate to the greatest extent in terms of amplitude over all the other oscillations in this frequency range, and in that that frequency at which, in the Fourier representation, the apex point of that parabola occurs with an access of symmetry which is parallel to the ordinate, and which runs through those points which characterize said three oscillations with respect to the frequency and amplitude, is interpreted as the frequency of the chewing actions.

20. The method as claimed in claim 15, wherein the number of chewing actions per portion of feed, which is regurgitated from the rumen into the mouth, chewed and then swallowed again is calculated by multiplying the frequency of the chewing actions by the duration of the chewing phase.

Description

[0037] The method according to the invention is explained in more detail below with reference to drawings and the detailed description of the “rumination” example:

[0038] FIG. 1: shows a flowchart of the central, cyclically repeating process sequence of an advantageous exemplary method according to the invention, and

[0039] FIG. 2: shows a profile of the absolute value of the acceleration plotted over time, measured by means of acceleration measurement on an animal which is currently ruminating.

[0040] The phase in which a ruminant ruminates is divided into a multiplicity of individual cycles occurring chronologically one after the other, wherein an individual cycle comprises a regurgitating phase, a chewing phase and a swallowing phase. During the regurgitating phase, a quantity of feed which is referred to as a “bolus” is moved from the rumen of the animal into its mouth. During the chewing phase, the bolus is comminuted more finely in the mouth by means of chewing movements. The individual bites which are carried out during the chewing movements are referred to as “chewing actions”. During the swallowing phase, the bolus is swallowed again. The chewing phase can be differentiated from the outside compared to the other two phases through, for example, relatively large head movements of the animal which are periodic with the frequency of the chewing actions, and also through characteristic noises.

[0041] For the method according to the invention, the head movements are measured with the aid of acceleration sensors. The sensors which are necessary for measuring acceleration can be fabricated without difficulty in such a small size that they can readily be placed, for example, in an ear tag (also including a necessary energy accumulator such as typically a battery).

[0042] The necessary acceleration sensors are, if required, of course also available for other measurements than for the purpose according to the invention here. Examples of this are gait analyses and the determination of miscellaneous activities or positions of the animal such as e.g. eating, walking or lying.

[0043] The position numbers 1 to 11 according to FIG. 1 signify processes which are additionally symbolized in FIG. 1 by labeled boxes or other symbols and occur essentially one after the other in the chronological sequence characterized by arrows.

[0044] The upper rectangle a illustrated using dotted lines according to FIG. 1 surround symbols 1, 2 for processes which also take place in the device which is carried along on the animal. The rectangle a therefore symbolizes a device which is carried along on the ruminating animal, typically an ear tag. This device contains at any rate either a plurality of unidimensionally measuring acceleration sensors or at least one multidimensionally measuring acceleration sensor.

[0045] The lower rectangle b illustrated using dotted lines according to FIG. 1 surrounds symbols for processes 4 to 11 which ideally occur in an evaluation section which is separate from the animal, typically locationally fixed, and which has a connection 3, 10 to the device on the animal, via which connection bidirectional transmission of data is possible. This connection is typically a wireless radio connection. The rectangle b therefore symbolizes an evaluation station.

[0046] In step 1 “3D-ACC” acceleration data is recorded in three coordinate directions which are perpendicular to one another. Good results can be obtained if the measuring frequency is 10 Hz, that is to say if the respective acceleration is measured ten times per second respectively in each of the three coordinate directions.

[0047] Step 2 “STO” means the buffering of the data measured in step 1 in a data memory which is located in the device on the animal. For example, acceleration data is measured over a time period of two minutes and is all written into the data memory.

[0048] In step 3, a radio link is set up between the device a on the animal and the typically fixed evaluation station b, and the acceleration data which is stored in the device a is transmitted via said link to the evaluation station b.

[0049] In step 4 “SW/K” the measured acceleration data, which, of course, represents a chronological profile of accelerations of the device a, are evaluated in order to detect which data can be assigned to chewing phases (“K” in FIG. 2) and which data can be assigned to swallowing/regurgitation phases (“SW” in FIG. 2). During the chewing phases, a nutrition bolus which is located in the mouth of the ruminating animal is finely comminuted by chewing movements. During a swallowing-regurgitation phase, the finely comminuted feed is firstly swallowed and then a new bolus, which is initially composed of feed which has not yet been finely comminuted is regurgitated into the mouth from the rumen. Generally, the swallowing-regurgitation phase is associated with significantly smaller accelerations and changes in acceleration than the chewing phase.

[0050] The differentiation between the chewing phases and swallowing-regurgitation phases is very successful by virtue of an adapted form of the variance analysis of the measurement results. For this purpose, in each case a total amount of an overall acceleration is preferably calculated first from the respective three acceleration values which were recorded per measuring time by geometric addition. (A resulting profile of the absolute acceleration over time is shown by way of example in FIG. 2. Two swallowing-regurgitation phases (“SW”) and a chewing phase (“K”) which lies between these are also marked in FIG. 2.) In each case two partial quantities are used from the total quantity of the data series obtained, at which partial quantities the associated acceleration measurements occur in two adjoining time windows of the measuring time period. In each case the variance is calculated from the two partial quantities. (The variance is the sum of the squares of the distances of the individual values from the mean value over all the individual values divided by the number of individual values). The two obtained variance values which each apply for one of the two time windows are compared with one another. A significant difference in the variance values is an indication that in the region of the boundary between the two time windows there is a transition between a chewing phase and a swallowing-regurgitation phase. These variance calculations are repeated for chronologically shifted pairs of time windows, but the two windows of one pair adjoin one another in each case. In a first stage of the calculation the time window pairs can be shifted from one variance calculation to another by rather large steps so that the entire time range is passed through quickly. For those time ranges for which indications of phase transitions between the chewing phase and swallowing-regurgitation phase are found here, the variance calculations are repeated, wherein from one calculation to the next the time window pairs are shifted by a relatively small time increment, so that the time of the transition between two phases can be set more precisely.

[0051] Since the accelerations are, of course, measured independently of what triggered them, even events which are nothing to do with rumination, for example cattle being frightened by a violent event or an ear being shaken in order to scare away a fly, act on the result of the acceleration measurements. So that such events have as little possible effect on the result of the evaluation of the measurements, only those phases between two phases interpreted as swallowing-regurgitation phases are accepted as chewing phases of the rumination for the further evaluation of the rumination, which chewing phases are not shorter than a certain minimum duration (for example 25 seconds) and also not longer than a certain maximum duration (for example 70 seconds).

[0052] In steps 5 to 8, operations for the measurement with respect to rumination are there only continued with acceleration data which has been recognized as being associated precisely with one chewing phase in step 4. In addition to the detection of the start and end, this also involved the fact that it was detected that there was no interruption between the start and the end and that the interval between the start and the end is a time period which is realistic for the chewing phases during rumination.

[0053] Step 5 “1 to N”: three acceleration phases were respectively measured at the individual measuring times in step 1, wherein the individual values represent the acceleration in one coordinate direction in each case. The value triple of a measurement of an individual point in time therefore signifies a vector in the space, the direction of which depends on the ratios between the magnitudes of the three individual acceleration measurements, and the absolute value of which is obtained from geometric addition of the individual components. The respective vector which is defined by the three measured acceleration components, in terms of direction and absolute value, changes from one measuring time to the next. In step 5, it is calculated how large the component of the individual measured acceleration vectors occurring in the respective direction is for a multiplicity of directions which are assumed in the space in a “hedgehog-like fashion”, for example a hundred directions. For this purpose, the scalar product of the unit vector (vector with a length of 1) in the respective direction can be calculated with the respective acceleration vector.

[0054] Over all the acceleration measurements of a chewing phase this means geometrically that a multiplicity of path profiles is derived from the single curve, running in the three-dimensional space and formed by connecting the tips of the acceleration vectors to one another in a chronologically ordered fashion, wherein each individual path profile occurs on precisely only one direction line in the space and precisely only one path profile occurs on each direction line. (A “direction line” in the space in this sense is a straight line through the origin of the assumed coordinate system).

[0055] In step 6 “FFT” (A Fourier transformation) is carried out for each of the many path profiles which were calculated in step 5 and which each occur only along one direction vector in the space. That is to say the profile of a scalar variable (absolute value of the acceleration) over time is no longer represented as a sequence of value pairs of point in time/absolute value of the acceleration but rather as a sum of sinusoidal oscillations which are determined individually by frequency, amplitude and phase, wherein in this case the amplitude symbolizes an acceleration. In a specialist field, in particular what is referred to as an FFT (Fast Fourier Transformation) is best known and has been introduced as an algorithm for Fourier transformation for digital data processing, for which reason more details will not be given on this here. The result of the step 6 is therefore a separate Fourier representation for each of the individual acceleration profiles plotted against the time—which acceleration profiles are each geometrically oriented along an individual direction vector of the direction vectors.

[0056] In step 7 “N to 1”, that Fourier representation which most clearly shows a frequency distribution from which the chewing rhythm of the currently ruminating animal can be read off is filtered out from the individual Fourier representations (each individual representation of which respectively stands for a direction in the space). This is then the case if, in that frequency range in which the chewing frequency has been found empirically to occur, either a single oscillation relating to the amplitude dominates clearly over all the other oscillations in this frequency range or if in this frequency range only very few oscillations, associated locally with respect to their frequency, dominate clearly in this frequency range with respect to the amplitude over all the other oscillations. In the first case, the frequency of the individual oscillation is the searched-for chewing frequency. In the second case, the chewing frequency can be determined by extending a parabola, the axis of symmetry of which is parallel to the ordinate, in the Fourier representation, through those points through which the three most dominant oscillations are defined (the abscissa corresponds to frequency, the ordinate corresponds to the amplitude). The abscissa value of the location of the apex of the parabola is the searched-for chewing frequency.

[0057] In the case of cattle, the chewing frequency in the case of rumination is in the region from approximately 40 to approximately 85 chewing actions per minute (0.67 to 1.42 Hz), typically in the range from 60 to 70 chewing actions per minute (1 to 1.17 Hz).

[0058] In step 8 “Count Sum” counters can be reset and obtained if result can be output. The outputting of a result relating to the instantaneous value of a variable describing rumination can be the chewing frequency or still better the number of chewing actions per bolus. In order to be able to specify the number of chewing actions per bolus, the chewing frequency is to be multiplied by the duration of the chewing phase determined in step 4.

[0059] A counter counts the chewing actions taking place during rumination. At the value which is already located in this counter, the number of chewing actions determined in the last processing cycle is added. This number is obtained as the calculated chewing frequency multiplied by the duration of the respective chewing phase which is bounded by two swallowing-regurgitation phases, the detection of which took place in step 4.

[0060] A second counter counts the swallowing processes taking place during the rumination. At the value already located in this counter, the value 1 is added per swallowing-regurgitation phase.

[0061] In step 9 “Control direct”, a first subordinate control unit defines when a measuring attempt is to be started again by the device a carried on the animal, and possibly also how long, if appropriate, measurement is to take place. The rules according to which this takes place in step 9 are defined by a superordinate logic process 11 “Control superior” in accordance with the outlined example.

[0062] For the monitoring of the state of an animal by determining the number of bites per bolus during rumination it is not necessary to perform measurement continuously. For example, it may be sufficient if the number of chewing actions per bolus during rumination is determined successfully three or four times a day.

[0063] If the aim is to find the instantaneous value of a variable which describes the rumination, ideally a logic is stored which starts measuring processes more frequently on the day (for example every half hour) until the number of successful measuring processes at which rumination was actually detected and analyzed has reached the amount which is necessary for one day. Of course, this logic can also be influenced by further processes such as, for example, the age of the animal, indications of it being on heat or chronological proximity to a birth event or suspicion of illnesses. There is ideally also the possibility of this logic being edited by a person using an input device. The described method is therefore highly efficiently in terms of energy consumption in the device a located on the animal because the measurements do not have to be carried out continuously but rather only a couple of times per day so that the operating time of the device a per day can be for example in the range of 1% of the daily duration. It is therefore possible to ensure that the battery in the device a can last for a very long time.

[0064] In step 10, a radio link is set up between the (fixed) device b and the device a located on the animal, and the determinations relating to the device a are communicated to the device a. In so far as measurement is not carried out continuously in any case, this relates at least to the time of the next measuring cycle.

[0065] The described evaluation of the measurement data according to steps 4 to 8 appears to be costly. However, in fact it is easy to carry out since it can be carried out on a fixed device, for example a PC, and since computing capacity, energy and time can be sufficiently made available very easily therein. As a result of the highly developed evaluation, the risk of incorrect measurements is very low and the measuring accuracy which is achieved is very good even though few, relatively short measuring cycles, at an extremely low measuring frequency (typically 10 Hz) are sufficient on the animal.

[0066] The described method for finding out the instantaneous value of a quantifiable variable which describes the rumination can be modified and/or expanded in many respects within the inventive concept. In this respect the following is to be noted:

[0067] For example during the evaluation of the data it would be possible to dispense with the detours via the projection of the acceleration vectors to a large number of individual direction vectors (step 5), the Fourier transformations of the accelerations per individual direction vector and the search for the Fourier transformation which is most informative, and instead to measure accelerations with a relatively high measuring frequency in device a (approximately 100 measurements per second instead of 10 measurements) and to find out the times of extreme values through a comparison of the magnitude of successive measurement results and to infer from their intervals the frequency of oscillations and ultimately the chewing frequency or the duration of individual chewing actions. However, owing to the relatively high measuring frequency more energy would be used in device a, and owing to the lower reliability of the acquired measurement results it would be necessary to carry out measurements very much more often and to attempt to achieve sufficient measuring accuracy and reliability of the direction through the formation of averages and through further logic filters. In order to achieve equally good accuracy and reliability the energy consumption per day in the device a located on the animal would therefore be dramatically higher; presumably the hardware in the device a would have to be constructed in a more costly fashion so that more acceleration measurements can be carried out over time and that more measurement results can be buffered. However, a smaller consumption of energy would nevertheless presumably be found to be sufficient than when simply measuring the duration of the rumination per day in a general way.

[0068] For example, before a measuring cycle which is to be carried out on the device a in order to record the accelerations during the rumination of at least one bolus, a significantly shorter measuring cycle can be carried out, by means of which it is determined whether there are actually sufficient accelerations for rumination to be possibly occurring. The evaluation in this respect can comprise, for example, the fact that measured values of the accelerations are summed up over a short duration and the result is compared with a threshold value, and can also already be carried out on the device a located on the animal. If in this context it is determined that rumination is in fact definitely not occurring, the pending measurement relating to rumination can be postponed by a defined duration, for example half an hour.

[0069] The transmission of data between the device a located on the animal and the evaluation station b does not necessarily have to occur in a wireless fashion by radio. It can, for example, also occur by means of electrical conduction using the animal's body as an electrical conductor as soon as the animal touches an electrode which is in contact with the evaluation station, or at least comes so dose to the electrode that capacitive signal transmission is possible.

[0070] It is also possible to carry out Fourier transformation on the raw data of the acceleration measurements or on fewer data items from the acceleration measurements to be processed according to FIG. 1. The frequency of the chewing actions can, if appropriate, also be read out to a certain extent from the relevant results, but the result is less accurate and less unambiguous than according to the sequence described with reference to FIG. 1.

[0071] For the quantifiable variable which was obtained by evaluating the sensor data and which, under certain circumstances, indicates something about an instantaneous state of the rumination process, in addition to the already described variables “number of the chewing actions per bolus” and “chewing frequency”, for example the following variables are also possible within the inventive concept length of the chewing phase, length of the regurgitation phase, length of the swallowing phase, ratio of said phase lengths with respect to one another, averaged magnitudes of accelerations during individual phases, ratios of magnitudes of accelerations in different phases.

[0072] The method which is described in detail for the measurement of the rumination can be applied in a largely analogous fashion for the measurement of the consumption of feed, that is to say eating. The consumption of feed also occurs in repeating sequences of, in each case, a plurality of chewing actions and a swallowing process. Differences with respect to rumination relate to the frequency of the chewing actions (this is higher than in the case of rumination) and the duration between two swallowing processes (this is shorter than in the case of rumination), as well as the fact that a swallowing process takes up less time during the consumption of feed than a swallowing-regurgitation process during rumination.

[0073] In the case of the consumption of feed, a lower number of chewing actions between two swallowing processes is an indication of feed which is rich in protein; a larger number of chewing actions between two swallowing processes is, on the other hand, an indication of feed with a relatively high portion of long fibers.

[0074] Instead of counting the chewing actions during the consumption of feed, it is also possible to measure the duration between successive swallowing processes. The information about this duration relating to the feed and the state of the animal can be tendentially compared with the information about the numerical ratio from the chewing actions and swallowing processes because a relatively long duration between two swallowing processes goes hand in hand with a relatively high number of chewing actions between two swallowing processes. However, according to previous observations the number of chewing actions per swallowing process appears to be more informative as a basis for comparison between different animals than the duration between two swallowing processes. The chewing frequency from animal to animal appears to fluctuate more greatly than the quantity of feed moved during a swallowing process.

[0075] Drinking can be detected extremely well from the measured acceleration data by means of a variance analysis in the form of comparison of the variance within two adjoining time windows, as described above for the example of rumination. Given relatively low acceleration values there is a sequence, which repeats with a slow rhythm, composed of a relatively long suction phase during which virtually no acceleration takes place and a relatively short swallowing phase with a relatively high acceleration. The period duration of a cycle which is composed of sucking and swallowing is typically 5 to 10 seconds in cows.

[0076] The superordinate logic process 11 “Control superior” is a superordinate control process. For example, in said process further influencing variables other than only time and acceleration values can be taken into account. Typically events of a localization are taken into account and processed logically as well as prescriptions which can be defined on a user interface in such a way that they can be edited. By virtue of the process 11, further process steps, by which the measured acceleration data is examined and evaluated to determine whether, and if so how, consumption of feed or drinking takes place, can be controlled in parallel with or as an alternative to the process steps 4 to 8.

[0077] Likewise, by means of the superordinate process 11 it is possible to detect—typically on the basis of acceleration data—whether rumination, consumption of feed or drinking is at all occurring. If none of these occurs for a relatively long time, the recording and detailed analysis of acceleration measurement data can be restricted, that is to say can be carried out on a “sample basis” oral in periods which are relatively widely spaced apart. Therefore, in particular in the sleeping phases it is possible to save energy in the device a. The device a is, of course, carried in a mobile fashion on the animal and therefore requires a battery or an accumulator.

[0078] In the process 11, the counting results of step a are also stored and, if appropriate, processed further to form further information, if necessary alarms etc. Likewise, if appropriate results relating to the consumption of feed or drinking can be stored and processed further.

[0079] By means of the further processing, for example the following information is generated, and can also be output: [0080] Number of swallowing processes per day [0081] Number of swallowing processes during the consumption of feed per day [0082] Ratio of numbers of chewing actions per swallowing process during the rumination [0083] Ratio of numbers of chewing actions per swallowing process during consumption of feed [0084] Number of swallowing processes while drinking [0085] Deviations from the average value of the individual numbers or ratio of numbers with respect to a standardized value (or the respective average value for the herd) [0086] Deviations from the average value of the individual numbers or ratio of numbers with respect to the average value for the same animal in preceding (selectable) time periods [0087] Information on the extent to which the above deviations are an indication of, for example, the animal being on heat yes/no, predicted optimum insemination time period, too much or too little protein . . . [0088] Miscellaneous information on the state: healthy normal state, supposed milk yield . . .

[0089] The method according to the invention has mainly been explained and described until now with reference to the application for milk cows.

[0090] Within the scope of the activity of a person skilled in the art, said method can also be adapted for application to other animals. For this adaptation, essentially the characterizing acceleration values, variances and repetition frequencies of processes are to be identified and the evaluation methods correspondingly adjusted.