METHOD FOR ACTUATING AN ELECTRIC MOTOR AND CONFIGURATION FOR EXERTING OSCILLATORY ROTATION OF A DRIVESHAFT

Abstract

A method for actuating an electric motor for a rheometer includes transferring drive energy to a sample. A time profile for deflection is periodic, a value for deflection is a measured variable, the motor is actuated by a manipulated variable, the measured and manipulated variables are mutually nonlinear, an approximation function for the time profile is a weighted sum of base functions, weights for base functions are a parameter vector, the manipulated variable is a weighted sum of base functions. The measured variable is sampled, sampled values are used within a time window, an approximation function for sampled values is a weighted sum of base functions, the weights are an actual parameter vector, a difference between intended and actual parameter vectors is subtracted from the manipulated parameter vector, the manipulated variable is a weighted sum of base functions and values of a new manipulated parameter vector are used as weights.

Claims

1. A method for actuating an electric motor for an oscillatory rotation of a driveshaft of the electric motor or a driveshaft of a rheometer, the method comprising the following steps: a) using the electric motor to transfer drive energy of the electric motor to a sample resisting oscillation of the electric motor; b) predetermining an intended time profile to be achieved for a deflection or for a sample torque and providing the intended time profile with a periodic predetermined form; c) continuously establishing an actual value for the deflection or for the sample torque as a measured variable; d) actuating the electric motor by predetermining a manipulated variable in a form of a voltage applied to the electric motor or a current flowing through electric motor; e) the measured variable and the manipulated variable behaving nonlinearly with respect to one another, at least within a region between a maximum and a minimum of the predetermined periodic intended time profile; f) establishing an approximation function for the intended time profile as a weighted sum of a number of predetermined periodic base functions with a time offset where necessary, and establishing weights being used for individual base functions as an intended parameter vector; g) predetermining the manipulated variable as a sum of the base functions weighted by manipulated parameters of a manipulated parameter vector, and initially predetermining the intended parameter vector multiplied by a predetermined factor as manipulated parameter vector; subsequently carrying out the following steps h) to k) continuously and repeatedly in accordance with a regulating process, as follows: h) continuously sampling the measured variable and using last established sampled values for the measured variable within a predetermined time window; i) establishing an approximation function for the sampled values of the measured variable within the time window as a weighted sum of the base functions, and establishing weights being used for the individual base functions as an actual parameter vector; j) establishing a difference between the intended parameter vector and the actual parameter vector and subtracting the difference, possibly weighted by a further predetermined factor, from the manipulated parameter vector; and k) predetermining the subsequently used manipulated variable as a weighted sum of the base functions, and using the values of newly generated manipulated parameter vector as weights in steps h) to j).

2. The method according to claim 1, which further comprises at least one of: using sine and cosine oscillations as the base functions, or providing a first base function (f.sub.1(t)) with a predetermined basic shape and compressing each of further base functions (f.sub.2(t), . . . ) in relation to the first base function f.sub.1(t) by a predetermined integer value n, in such a way that f.sub.n(t)=f.sub.1(n*t); or setting a number of base functions to be less than 5.

3. The method according to claim 1, which further comprises: predetermining the base functions as periodic functions; and selecting the sampling in such a way that more than 100 samples are taken during a period duration of a base function with a longest period.

4. The method according to claim 1, which further comprises: predetermining the base functions periodically; and providing the time window, within which samples are used, with a duration of between 25% and 100% of a period duration of the base function with a longest period.

5. The method according to claim 1, which further comprises: predetermining the base functions as periodic functions; and periodically repeating adaptation of steps h) to k), wherein a time period of between 25% and 100% of a period duration of the base function with a longest period lies between two adaptations in each case.

6. A configuration for exerting oscillatory rotation of a driveshaft of a motor or a driveshaft of a rheometer for measuring viscosity of a sample, the configuration comprising: a) an electric motor including a driveshaft for transferring drive energy of said electric motor to the sample; b) a motor regulator having a periodic intended time profile to be achieved, being predetermined in advance for a deflection or for a sample torque; c) a measuring device continuously establishing an actual value for the deflection or for the sample torque as a measured variable and reporting said measured variable to said regulator; d) said regulator actuating said electric motor by predetermining a manipulated variable in a form of a voltage applied to said electric motor or a current flowing through said electric motor; e) said measured variable and said manipulated variable behaving nonlinearly with respect to one another, at least within a region between a maximum and a minimum of said predetermined periodic intended time profile; f) said regulator establishing an approximation function for said intended time profile as a weighted sum of a number of predetermined periodic base functions, with a time offset where necessary, and establishing weights being used for individual base functions as an intended parameter vector; g) said regulator predetermining said manipulated variable as a sum of said base functions weighted by manipulated parameters of a manipulated parameter vector, and initially predetermining said intended parameter vector multiplied by a predetermined factor as a manipulated parameter vector; and said regulator subsequently performing the following functions h) to k) in accordance with a regulating process in a continuous and repeated manner, as follows: h) said regulator continuously sampling said measured variable from said measuring device and using last established sampled values for said measured variable within a predetermined time window; i) said regulator establishing an approximation function for said sampled values of said measured variable within said time window as a weighted sum of said base functions, and establishing weights being used for said individual base functions as an actual parameter vector; j) said regulator establishing a difference between said intended parameter vector and said actual parameter vector and said regulator subtracting said difference, possibly weighted by a further predetermined factor, from said manipulated parameter vector; and k) said regulator predetermining a subsequently used manipulated variable as a weighted sum of said base functions, and said regulator using values of a newly generated manipulated parameter vector as weights in said functions h) to j).

7. The configuration according to claim 6, wherein at least one of: sine and cosine oscillations are used as said base functions, or a first base function (f.sub.1(t)) has a predetermined basic shape and further base functions (f.sub.2(t), . . . ) are each compressed in relation to said first base function f.sub.1(t) by a predetermined integer value n, in such a way that f.sub.n(t)=f.sub.1(n*t), or said base functions have a number less than 5.

8. The configuration according to claim 6, wherein: said base functions are predetermined as periodic functions; and said sampling is selected in such a way that more than 100 samples are taken during a period duration of said base function with a longest period.

9. The configuration according to claim 6, wherein: said base functions are periodic; and said time window, within which said samples are used, has a duration of between 25% and 100% of a period duration of said base function with a longest period.

10. The configuration according to claim 6, wherein: said base functions are periodic; and said regulator periodically repeats an adaptation of functions h) to k), and a time period of between 25% and 100% of a period duration of said base function with a longest period lies between two adaptations in each case.

Description

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING

[0036] FIG. 1 is a block diagram of a particularly preferred embodiment of the invention showing a motor to which a predetermined voltage profile or current profile is applied by a regulator by way of a voltage source as well as a sample to which drive energy is transferred;

[0037] FIG. 2 is a diagram showing an advantageous example of base functions;

[0038] FIG. 3 is a diagram showing a measured variable; and

[0039] FIG. 4 is a graph of an intended parameter vector against deflection.

DETAILED DESCRIPTION OF THE INVENTION

[0040] Referring now to the figures of the drawings in detail and first, particularly, to FIG. 1 thereof, there is seen a motor 1 to which a predetermined voltage profile U.sub.M or current profile I.sub.M is applied by a regulator 3 by way of a voltage source. In a manner dependent on a predetermined intended time profile for a deflection w of the motor or for a sample torque M, the regulator 3 accordingly sets a current time profile or a voltage time profile as a manipulated variable u(t). The electric motor 1 is actuated for an oscillatory rotation of the driveshaft thereof. The electric motor 1 transfers a drive energy thereof onto a sample 2 through a motor shaft. The sample 2 is situated between two plates, of which at least one is rotated counter to the sample 2 in such a way that, overall, the sample 2 is subjected to a shearing or rotational movement. Different torques arise on the motor shaft depending on the deflection of the driveshaft of the electric motor 1 due to the specific viscosity of the sample 2. These established or set deflections w and torques M can be related to one another, as a result of which the specific viscoelastic behavior of the sample 2 to be examined can be established.

[0041] It is either the sample torque M or the deflection w which is predetermined in advance in the form of an intended variable e(t) so that such a measurement can be undertaken overall. In this case, the intended time profile e(t) has a periodic, predetermined form and is predetermined for the regulator 3.

[0042] The configuration in FIG. 1 contains a measuring device 4, which continuously determines either the actual value of the deflection w or the actual value of the sample torque M. Ultimately, this measuring device 4 supplies actual values for the deflection w or the sample torque M as measured variable y(t) and transfers the latter to the regulator 3.

[0043] The assumption is made within the scope of the invention that the sample 2 exhibits nonlinear behavior. If the driveshaft of the motor 1 is only moved within a small deflection range about a work point, the sample 2 usually has a linear behavior around the relevant work point. However, if the deflection w is increased, this has as a consequence in the case of a nonlinear sample 2 that the measured variables y(t) and the manipulated variable u(t) behave nonlinearly in relation to one another, at least within a range between the maximum and minimum of the predetermined, periodic intended time profile e(t). Due to this nonlinear behavior, it is not possible either to already estimate or establish a manipulated variable u(t), which ultimately obtains the desired intended time profile e(t), in advance. Moreover, the problem of a sample 2 changing during the measurement, in particular having a behavior exhibiting hysteresis, may also arise, and so setting a manipulated variable u(t) in advance for the purposes of reaching a predetermined intended time profile e(t) is not possible. It is for this reason that the invention uses the iterative method described in more detail below, in which the predetermined intended time profile e(t) for the deflection w or the sample torque M is ultimately achieved in a simple manner.

[0044] Initially, that is to stay still before the iterative adjustment, an approximation function e(t) is established for the intended time profile e(t), which approximation function is established as weighted sum of a number of predetermined, periodic base functions f.sub.1(t), f.sub.2(t), . . . which may be offset in time when necessary.

[0045] Advantageously, sine or cosine oscillations f.sub.1(t)=sin(a.sub.0t), f.sub.2(t)=sin(2a.sub.0t), . . . are used as base functions f.sub.1(t), f.sub.2(t), . . . , where a.sub.0 represents a base frequency of in particular 1 Hz, and the first base function f.sub.1(t) has a predetermined basic shape and the further base functions are in each case compressed in relation to the first base function by a predetermined integer value in such a way that f.sub.n(t)=f.sub.1(n*t). Preferably, use is only made of a few base functions in total. The present exemplary embodiment uses only three base functions in total.

[0046] By way of example, an advantageous example for base functions is depicted in more detail in FIG. 2. If the intended time profile e(t) is intended to be represented by an approximation function e(t), it is necessary to establish the individual weights, by using which the base functions f.sub.1(t), f.sub.2(t), . . . are intended to be weighted, in order to ultimately arrive at a time profile which corresponds to the intended time profile e(t) to the best possible extent e(t)e(t)=e.sub.1f.sub.1(t)+e.sub.2f.sub.2(t)+ . . . . The weights e.sub.1, e.sub.2, . . . established thus are established as an intended parameter vector E=[e.sub.1, e.sub.2, . . . ] and kept available for the further procedure. To the extent that use is made of sine and cosine oscillations, the values of the intended parameter vector E may be established e.g. by using a discrete Fourier transform or a Fast Fourier Transform (FFT).

[0047] For the purposes of initially setting the manipulated variable u(t), a manipulated parameter vector U=[u.sub.1, u.sub.2, . . . ] is predetermined, the individual elements of which represent weights whichmultiplied by the base functionsapproximately reproduce the manipulated variable u(t) as a weighted sum.

u(t)u(t)=u.sub.1f.sub.1(t)+u.sub.2f.sub.2(t)+ . . .

[0048] The intended parameter vector E, multiplied by a predetermined factor x, is predetermined as an initial value for the manipulated parameter vector U. The predetermined factor x is set in advance as follows: 0.5 if M is predetermined and 0.5*J*(2*pi*f.sub.n).sup.2 if w is predetermined (J: inertia of the measurement drive).

[0049] An iterative method is now presented below, by using which the regulator 3 continuously adapts the manipulated variable u(t) in order to generate a deflection w or a sample torque M in accordance with the predetermined intended time profile e(t). As is depicted in FIG. 3, the measured variable y(t)either the deflection w or the sample torque Mis sampled to this end. Advantageously, sampling takes place at very short intervals, wherein, in relation to the period duration of the base function f.sub.1(t) with the respective longest period, more than 100 samples are taken during such a period duration. In the case of a period duration of the base function f.sub.1(t) of 1000 ms, the sampling rate is preferably 512 Hz. Preferably, between 256 and 512 sampled values, in particular 256 or 512 sampled values, are recorded per oscillation. The sampled values within a time window W, which immediately precedes the respectively current time, are used. The time window W, within which the samples are used, is e.g. set to a duration of between 25% and 100% of the period duration of the base function f.sub.1(t) with the longest period.

[0050] Subsequently, the sampled values of the measured variable y(t) within the time window W are also subjected to the same analysis as the intended time profile. An approximation function y(t) is established as a weighted sum of the base functions; the individual weights, thus established, for the individual base functions are combined to form an actual parameter vector Y.

y(t)y(t)=y.sub.1f.sub.1(t)+y.sub.2f.sub.2(t)+ . . . ; Y=[y.sub.1, y.sub.2, . . . ]

[0051] A difference D between the intended parameter vector E and the actual parameter vector Y is established in a further step. This difference D seen in FIG. 4 is weighted by a predetermined factor v, which, in particular, lies between 0.2 and 0.5. This difference D is subtracted from the manipulated parameter U.sub.n and the manipulated parameter U.sub.n+1 for the next iteration step is thus formed.

U.sub.n+1:=U.sub.nD=U.sub.n(EY)*v

[0052] In a last step, the manipulated variable u(t) for the next iteration step is established as a weighted sum of the base functions on the basis of the newly established manipulated parameter vector U.sub.n+1.u(t)=u.sub.1f.sub.1(t)+u.sub.2f.sub.2(t). Subsequently, sampling is once again carried out within a subsequent time window W, an actual parameter vector Y is once again established, the difference D is established between the intended parameter vector E and the actual parameter vector Y and that difference is subtracted from the manipulated parameter vector U, and the manipulated parameter vector U is once again used for generating the manipulated variable u(t). This process is undertaken continuously by the regulator 3 in order to achieve appropriate adaptation to the measured variable, i.e. the deflection w or the sample torque M.

[0053] The adaptation can be repeated as often as desired. There is a time period between two respectively adaptations in each case of between 25 and 100% of the period duration of the base function f.sub.1(t) with the longest period.