EP2019962A1

EP2019962A1 - Method for processing measurement signals from a vortex velocity flow sensor

Info

Publication number: EP2019962A1
Application number: EP07728910A
Authority: EP
Inventors: Ole Koudal; Saso Jezernik; Höcker RAINER
Original assignee: Endress and Hauser Flowtec AG; Flowtec AG
Current assignee: Endress and Hauser Flowtec AG
Priority date: 2006-05-12
Filing date: 2007-05-08
Publication date: 2009-02-04
Also published as: JP4951060B2; US20080011099A1; JP2009537010A; WO2007131918A1; US8024136B2; DE102006022635A1

Abstract

The invention describes a method for processing measurement signals (US) from a vortex velocity flow sensor for measuring a flow of a medium through a measuring tube (1), said flow sensor having an accumulator body (3) which is arranged in the measuring tube (1) and a sensor (5) for detecting pressure fluctuations which occur in the region of the accumulator body (3) and for converting these pressure fluctuations into an electrical measurement signal (US), with a small demand for computing power and storage space, in which at least part of the measurement signal (US) is sampled and digitized, an autocorrelation (AK(T)) of the digitized measurement signal (US) is calculated, and the flow is derived using at least one property of the autocorrelation (AK(T)).

Description

description

Method for signal processing for measuring signals of a vortex flow sensor

The invention relates to a method for signal processing for measuring signals of a vortex flow sensor.

Vortex flow transducers are used in industrial measurement technology for the measurement of volume flows.

They work on the principle of Karman 'vortex street. In this case, a medium whose volumetric flow rate is to be measured is passed through a measuring tube in which a bluff body is inserted. Behind the bluff body alternately formed on both sides vortex with opposite direction of rotation. These vortices each generate a local negative pressure. The pressure fluctuations are detected by the sensor and converted into electrical measuring signals. The vortices develop very regularly within the permissible operating limits of the sensor. The number of vortices generated per unit time is proportional to the volume flow.

For detecting the pressure fluctuations, e.g. Capacitive sensors, such as those described in European Patent EP-Bl 0229 933.

The electrical measuring signals derived from the sensor have a sinusoidal profile in the case of heal. From this, the number of vortices produced per unit of time is determined on the basis of zero crossings of the electrical signal. In doing so, e.g. the time from one zero crossing to the next is measured, or e.g. in a time interval determines the time from the first to the last zero crossing and the number of zero crossings occurring in the time interval.

Under real measuring conditions, the case may occur that the signal sinusoidal in the Healfall interferences are superimposed, which can lead to an additional zero crossing occurs, or that a zero crossing disappears. Such measurement errors are very difficult to recognize.

Another disadvantage of such methods is that additional information present in the signal, e.g. whose amplitude or harmonics remain unused.

This information can be used by, for example by means of a fast Fourier transform (FFT) frequency determination is performed. However, such methods require that the measurement signal be at a high sampling rate is digitized and a large number of samples is stored at least temporarily. In addition, this form of frequency determination requires a large number of mathematical operations. Vortex flowmeters are already available today as so-called 2-wire devices. Such 2-wire devices have two connecting lines, via which both the power supply of the device and the transmission of measurement results. In the industry, a standard has been established for these 2-wire devices, according to which the power supply via a 24 volt power source and the 2-wire devices flowing through the connecting lines current depending on the transmitted measurement results to values between 4 mA and 20 niA rules. Accordingly, only a small amount of electrical power is available to these sensors. As a result, the achievable computing power is limited.

It is an object of the invention to provide a method for signal processing for measuring signals of a vortex flow sensor with low demand for computing power and storage space. For this purpose, the invention in a method for signal processing for

Measuring signals of a vortex flow sensor for measuring a flow of a medium through a measuring tube,

[0013] - having a baffle body arranged in the measuring tube, and [0013] a sensor for detecting pressure fluctuations occurring in the region of the baffle body and for converting these pressure fluctuations into [0015] an electrical measuring signal, in which

[0016] at least part of the measurement signal is sampled and digitized, [0017] an autocorrelation of the digitized measurement signal is calculated, and [0018] the flow is derived from at least one property of the autocorrelation.

According to an embodiment of the method, the characteristic is a frequency or a period of autocorrelation, and the frequency or period of the autocorrelation is set equal to a frequency or a period of the measurement signal.

According to a first development, a zero or a minimum of the autocorrelation is determined, and the property of the autocorrelation is determined based on the location of the zero or the minimum. According to a development, the location of the minimum by an adjustment a parabola is determined at points of autocorrelation. According to another embodiment, the property of the autocorrelation is the

Frequency or period of the same and is adjusted by a cosine

Function to the autocorrelation determined. According to one embodiment, the measuring signal passes through before the formation of the

Autocorrelation an adaptive filter. According to a development of the latter embodiment, the adaptive filter is an adaptive bandpass filter. According to a development of the latter embodiment, a frequency of the autocorrelation is determined, and a frequency range in which the bandpass filter is permeable is set based on the frequency of the autocorrelation. According to another embodiment of the latter embodiment, an adaptive line enhancer is used as the adaptive bandpass filter. According to another embodiment of the latter embodiment, the adaptive filter is a notch filter. According to a development of the latter development leads the adaptive

Notch filter is a filter function that provides a residual signal that is used to optimize the

Filter serves, and a complementary filter function, which provides the measurement signal, on the basis of which the autocorrelation is determined. Furthermore, the invention consists in a vortex flow sensor for

Measuring a flow of a medium through a measuring tube, with [0031] a baffle arranged in the measuring tube and [0032] a sensor for detecting pressure fluctuations occurring in the region of the baffle [0033] and for converting these pressure fluctuations into an electrical measuring signal [0035] an analog-to-digital converter serving to sample and digitize at least part of the measurement signal, and [0037] signal processing serving to autocorrelate the digitized measurement signal and derive the flow rate on the basis of at least one property of the autocorrelation.

An advantage of the invention is that not only individual measuring points of the measuring signal are received by the autocorrelation, but course and amplitudes of the

Be taken into account. As a result, a higher measurement accuracy is achieved. Another advantage is the white signal superimposed on the measurement signal Noise in the autocorrelation only at the correlation time has zero, and thus can be easily hidden.

The invention and further advantages will now be explained in more detail with reference to the figures of the drawing, in which five embodiments are shown; like parts are provided in the figures with the same reference numerals.

Fig. 1 shows a vortex flowmeter;

FIG. 2 shows a sensor of a vortex flowmeter; FIG.

Fig. 3 shows schematically a formation of vertebrae behind a

Bluff body;

Fig. 4 shows a measurement signal as a function of time;

Fig. 5 shows an autocorrelation of the measurement signal of Fig. 4 as a function of

Correlation time;

FIG. 6 shows an autocorrelation of a sinusoidal measurement signal. FIG

Superimposed exclusively white noise;

Fig. 7 shows a minimum portion of the autocorrelation

And one adapted to points in the vicinity of the minimum

Parabola;

Fig. 8 shows a connected to a sensor of the sensor

Circuit with an adaptive filter; and

Fig. 9 shows a connected to a sensor of the sensor

Circuit with a notch filter.

Fig. 1 shows an example of a vortex flowmeter, and Fig. 2 shows an example of an associated sensor. This works on the principle of the Karman 'vortex street. A medium whose volumetric flow rate is to be measured is passed through a measuring tube 1. In a flow path of the medium represented by an arrow in FIG. 1, a bluff body 3 is introduced. Behind the bluff body 3 are alternately formed on both sides vortex with opposite direction of rotation. This is shown schematically in FIG. The vortices each produce a local negative pressure. The pressure fluctuations are detected by the sensor and converted into electrical measuring signals. The vortices develop very regularly within the permissible operating limits of the sensor. The number of vortices generated per unit time is proportional to the volume flow. The frequency f of the measuring signals is proportional to the flow rate of the medium in the measuring tube 1, which in turn is proportional to the volume flow.

It is a sensor 5 for detecting in the region of the bluff body 3, for example on or behind the bluff body 3, occurring pressure fluctuations and to convert these pressure fluctuations in an electrical measurement signal U _s provided. For this purpose, pressure sensors, such as the capacitive sensors mentioned above are suitable. The sensor 5 is arranged in the flow direction behind the bluff body 3 in the measuring tube 1. The electrical measuring signals U _s derived from the sensor 5 have a sinusoidal profile in the case of heal. 4 shows such a measurement signal U _s as a function of time t. The frequency f of the measuring signal U _s is proportional to the flow velocity of the medium in the measuring tube 1 and thus to the flow.

According to the invention, at least part of the measurement signal U _{s is} sampled and digitized. In this case, it suffices to scan and digitize a signal train which contains only a small number of oscillations, eg 10 oscillations, with a sufficient sampling rate, eg 10 points per oscillation. The individual sampling points are shown in FIG. 4 and, following the digitization, are available as measuring points [UsCt ₁ ); tj available, where U U of the received value ₈ (I ₁₎ ζ at the sampling time of the measuring signal _s and i is an integer running index with i = 1 .. M. Between two consecutive sampling times ζ and t _{1 + 1 there} is in each case an identical period of time Δt predetermined by the sampling rate.

According to the invention, an autocorrelation AK (T) as a function of

Correlation time T of the digitized measurement signal U _s (t) is calculated, and the flow derived on the basis of a property of the autocorrelation AK (T).

The autocorrelation AK (T) can, for example, according to the following calculation rule:

[0064]

AK (T) = Σ ^N U ₈ (I ₁ ) U ₈ (I ₁ + T)

I = 1

Be determined. The measuring points U ₈ (I ₁ ) were sampled at discrete times ζ. Accordingly, the autocorrelation AK (T) can be calculated for discrete correlation times T _k , where T _k = k Δt where k = 0, 1, 2 ... K and the condition N + K <M is satisfied.

The autocorrelation function of a periodic signal is also periodic and has the same frequency f and the same period P as the output signal, fct the output signal sinusoidal, for example in the form U (t) = U ₀ sin (2 π ft) , the corresponding autocorrelation function is a cosine function AK u (T) = A ₀ cos (2 π f T) with the same frequency f.

The autocorrelation function of an uncorrelated signal S, for example of white Noise has a nonzero value only for the correlation time T = 0. For all other correlation times T ≠ 0, the associated autocorrelation AKs (T) = 0.

In the case of the sinusoidal measuring signal U _s (t) shown in FIG. 4, the following applies:

[0071]

AK (T) = E ^N U ₈ Ct ₁ ) UsCt ₁ + T) = C cos (2π f T)

I = 1

Where C is a constant and

F is equal to the frequency of the measurement signal U _s (t).

The associated autocorrelation function AK (T) is shown in FIG.

If the measurement signal consists of a sinusoidal signal to which only white noise is superimposed, then the latter relationship applies only to correlation times T ≠ 0. At T = O, the autocorrelation AK (T) has a value C ₀ greater than C is. A corresponding autocorrelation function is shown in FIG. White noise only affects the value at T = O in the autocorrelation. All other areas of autocorrelation are noise free.

A property of the autocorrelation AK (T) is determined and the flow is derived from this property. Preferably, the property of the autocorrelation AK (T) is a frequency f or a period P of the autocorrelation AK (T). As explained above, the frequency f of the autocorrelation or its period P is substantially equal to the frequency f or the period P of the measurement signal U _s . The latter is a measure of the flow and, as described above, can be used to determine the flow. Preferably, the frequency f or the period P of the autocorrelation AK (T) is set equal to the frequency f or the period P of the measurement signal U _s . This offers the advantage that frequency f or period P of the measuring signal U _s can be determined much more accurately using the autocorrelation AK (T) than by evaluating the original measuring signal U _s> eg by detecting the zero crossings of the measuring signal U _s it is possible. Upon detection of the zero crossings of the measurement signal U _s , each disturbance can affect the number of zero crossings. On the other hand, in the autocorrelation, white noise affects locally only at the correlation time T = 0. A further advantage of the autocorrelation AK (T) is that its accuracy does not depend on the accuracy of a single measuring point, but instead takes measuring points of a signal train, and thus information about the amplitude and course of the utilized signal train is utilized. Correspondingly, the autocorrelation AK (T) for determining the frequency f or the period P of the measurement signal U _s much better suited than the measurement signal U _s itself.

Assuming that the measurement signal U _s (t) is a sinusoidal signal to which no or exclusively white noise is superimposed, the frequency f of the autocorrelation function AK (T) and thus the frequency of the sensor signal could already be based on a single value of the autocorrelation for T ≠ 0. However, this approach has not provided highly accurate, reproducible results with real measurement signals. One reason for this is that real measurement signals contain additional interference signals in addition to a sinusoidal signal and a white noise. According to the invention, therefore, a property of the autocorrelation AK (T) is used to determine the flow. It is sufficient to limit the calculation of the essentially cosinusoidal autocorrelation to distinct sections of the same. Significant sections are for example in the range of zero crossings, as well as minima or maxima of the autocorrelation.

According to a first variant, the course of the autocorrelation AK (T) in

Range of the first zero crossing used. For this purpose, the autocorrelation AK (T) is calculated, for example, for increasing correlation times T, with T = k Δt where k = 0, 1, 2 .., until at least a first negative value occurs, i. AK (x Δt) <0, where x is a natural number between 1 and K. On the basis of these values, a first zero point NS1 of the autocorrelation AK (T) is determined. The position of the first zero point NS 1 can be determined, for example, by a simple interpolation in which the point (AK (x Δt) <0; x Δt) and the preceding point (AK ((xl) Δt)> 0; (x -1) .DELTA.t), in which the autocorrelation has a positive value, a line is placed whose zero crossing is then set equal to the first zero NS 1. The first zero point NS 1 is given below according to the following calculation rule:

NS 1 = (x-AK (x Δt) / [AK (x Δt) -AK ((x-1) Δt]) Δt

The first zero point NS 1 corresponds to a quarter period of the substantially cosine autocorrelation AK (T). A full period P of the autocorrelation AK (T) accordingly has a duration equal to four times the first zero NS 1. The following applies: P = 4 NS 1. Accordingly, a frequency f results for the autocorrelation AK (T) with f = 1 / (4 NSl). This frequency f, reproduces the frequency f of the measurement signal U _s (t) very accurately. The frequency f of the measurement signal U _s (t) is thus set equal to the frequency f of the autocorrelation AK (T) based on the position the first zero point NS1 was determined. The same applies accordingly for the period P.

The sought flow rate is proportional to the frequency f of the measurement signal U _s and can thus be determined already on the basis of the first zero point NS 1 of the autocorrelation AK (T). Analogously, the flow can of course be determined on the basis of the associated period P.

An advantage of this method is that only a very small number of autocorrelation values are needed and the required arithmetic operation is very simple operations requiring only a few computational steps. Accordingly, only very small storage space and low processing power are required for this method.

According to a second variant, the course of the autocorrelation AK (T) in

Range of a minimum, preferably the first minimum Ml used. The first minimum M1 can be determined in various ways. For example, the autocorrelation AK (T) is calculated for increasing values of T, with T = k Δt with k = 0, 1, 2 .., until at least one value AK (y Δt) of the autocorrelation AK (T) has been determined, which is greater than an immediately preceding value AK ((yl) Δt). This gives the approximate location of the minimum Ml. The exact position of the minimum Ml, ie the correlation time T _Min , at which the minimum Ml occurs, can then be determined on the basis of a derivative of the autocorrelation AK (T) in this range. This exploits that a derivative of a function, where the function has a minimum, is equal to zero. Consequently, a zero of the derivative can be determined and set equal to the sought first minimum Ml.

Alternatively, for example, in the manner described above, the first zero point NS1 be determined and the position of the minimum Ml be estimated based on the location of the first zero point NS1. It is exploited that the minimum Ml is to be expected at a correlation time T _Mm which is twice the correlation time of the first zero point NS1. In addition, a derivative of the autocorrelation is preferably calculated in the range in which the minimum Ml is to be expected and determines a zero crossing of the derivative. The zero of the derivative corresponds to the position of the first minimum Ml.

Preferably, the exact location of the minimum Ml is determined by fitting a parabola p (T) to autocorrelation points in the range of the minimum. This can, as shown in FIG. 7, take place on the basis of at least three points of the autocorrelation AK (T). As points are, for example, the above [AK (yΔt); yΔt], [AK ((yl) Δt); (yl) Δt] and the preceding point [AK ((y-2) Δt); (Y-2) .DELTA.t]. The parabola p (T) is shown in FIG. 7 as a dashed line.

Likewise, the zero crossing of the derivative of the autocorrelation AK (T) can be used in the selection of the points. In this case, for example, that point of the autocorrelation which has the smallest distance to the zero point of the derivative and the two points directly to the right and left thereof are used.

At these three points, as shown in Fig. 7, a parabola is adjusted. For this purpose, the following approach can be used:

P (T) = P ₀ (T-T _M111 ) ² + _Pl

Where p ₀ , Pi and T _{Mm are} coefficients which are calculated on the basis of the three points of

Autocorrelation AK (T) can be determined. The correlation time T _min corresponds to the position of the minimum of the parabola p (T) and accurately reflects the correlation time of the minimum of the autocorrelation AK (T). Accordingly, it is set equal to the correlation time T _{Min of} the minimum Ml of the autocorrelation AK (T). In this way, the minimum can be determined very accurately even if only a few points of the autocorrelation are available.

Of course, the parabola can also be adapted to more than three points of autocorrelation, e.g. using the method of minimizing the distance squares. This can in particular bring a gain in accuracy if the measurement signal is digitized with a high sampling rate and thus significantly more than the above 10 points per oscillation are present.

The correlation time T _{Min of} the first minimum Ml corresponds to half a period Vi P. The frequency f of the autocorrelation AK (T) and thus the measurement signal U _s is thus f = 1 / (2 T _Min ).

Both methods described above have a very low number of autocorrelation points. The number can be reduced even further if an estimated value for the frequency f or for the period P is determined in advance. Such an estimate may e.g. be a meter specific value or be a value determined in a previous measurement for one of these quantities. If such an estimated value is present, then the range in which the first zero point NS1 or the first minimum M1 is to be expected can be limited and the calculation of the autocorrelation can be limited to these ranges.

The accuracy with which the flow can be determined can be further increased, by fitting a cosine function K (T) = cos (2 π f _c T) to the autocorrelation AK (T). For this purpose, preferably at least one full period of the autocorrelation AK (T) is calculated and the autocorrelation obtained is normalized. In this case, the value of the autocorrelation in the autocorrelation time T = 0 is preferably excluded, since the white noise is reflected in this value. The adaptation can be done, for example, by minimizing a sum of the distance squares J between the values of the cosine function and the associated normalized values ak (TJ of the autocorrelation, the frequency f _{c of} the cosine function serving as an adaptation parameter.

The sum of the distance squares can be determined, for example, according to the following calculation rule:

J (f _c ) = Σ ^L (ak (l Δt) - cos (2 π f _c 1 Δt)) ²

1 = 1

[0099] where 1 is an integer index running from 1 to L,

[0100] ak (l Δt) the normalized values of the autocorrelation, and

[0101] f _{c are} the adaptation parameters.

Subsequently, that frequency f _c at which an optimal adaptation is present is set equal to the frequency f of the autocorrelation AK (T) and equal to the frequency f of the measuring signal U _s , on the basis of which the flow is determined on the basis of the existing proportionalities. In the example mentioned, the optimum adaptation corresponds to the minimum of the sum of the distance squares J (f _c ).

In real measurement situations, the measurement signal can, in addition to the substantially sinusoidal wanted signal and any white noise present, generate additional interference signals, e.g. non-stationary noise, be superimposed, and may exhibit amplitude variations. Even in these situations, the described methods show very accurate measurement results.

The measurement accuracy can be further improved by the measurement signal U _s before the formation of the autocorrelation passes through a filter and the autocorrelation AK (T) is determined based on the filtered measurement signal.

Fig. 8 shows a corresponding circuit comprising the sensor 5 and a filter 7 connected thereto. The filtered measurement signals are fed to an analog-to-digital converter 9, which supplies the digital filtered measurement signals to a signal processing unit 11, for example a microprocessor. The signal processing unit 11 determines in the manner described above the autocorrelation and their property, on the basis of which subsequently the flow is derived. This can also be done by means of the signal processing unit 11.

The filter 7 which the measurement signal U _s undergoes before the formation of the autocorrelation AK (T) is, for example, an adaptive bandpass filter. Preferably, the frequency f of the auto-correlation AK (T) is determined, and a frequency range in which the band-pass filter is transmissive is set based on the frequency f of the autocorrelation. In this case, use is made of the fact that the frequency f of the autocorrelation AK (T) is essentially equal to the frequency f of the measurement signal U _s . The filter 7 is thereby continuously adapted to the frequency f of the measurement signal U _s by permitting the permeability of only one frequency range in the immediate vicinity of the frequency f of the measurement signal U _s . Accordingly, interference signals of other frequencies contained in the measurement signal are filtered out immediately.

Alternatively, the passband of the bandpass filter can be adjusted on the basis of a frequency of the measurement signal derived directly from the measurement signal. For this purpose, for example, counted in a conventional manner, the zero crossings of the measuring signal and based on their number per unit time, the frequency can be determined.

Instead of or in addition to the illustrated analog filter 7, a digital filter 13 may also be used. This is then, as shown in dashed lines in Fig. 7, to arrange behind the analog digital filter 9. Here are suitable, for example digital adaptive band-pass filters, which are used analogously to what has been said above in connection with the analog adaptive band-pass filter.

Another preferred embodiment for a digital filter 13 are so-called adaptive line enhancers (ALE). Such filters automatically adjust to the frequency of the measurement signal and cause a damping of the background noise.

Alternatively or additionally, in particular at high sampling rates, an average value filter 15 can be used. This is also shown in FIG. Mean value filters smooth the measuring signal by averaging over several consecutive measured values.

FIG. 9 shows a further circuit by means of which the measurement signal is subjected to filtering before the autocorrelation is formed. As filter, an adaptive notch filter 17 is provided here. Notch filters are highly selective filters for a narrow stopband. A notch filter 17 described by a filter function H (z) supplies at the output thereof a residual signal R which contains all the portions of the input signal lying outside the stop band. Usually, this residual signal R becomes Optimization of Notch filter 17 is used. The optimization of the notch filter 17 can be effected, for example, as shown schematically in FIG. 9, by adapting the filter 17 to the effect that the residual signal R is minimized.

According to the invention, the notch filter 17, in addition to the filter function H (z) complementary filter function 1 - H (z), which filters out the actual measurement signal U _s from the incoming sensor signal. This filtered-out measurement signal U _s is then further processed, as described above, by supplying it to the signal processing unit 11, whose autocorrelation AK (T) is determined and the flow is determined on the basis of at least one property of the autocorrelation AK (T).

[0113]

Table 1

1 measuring tube

3 bluff bodies

5 sensor

7 filters

9 analog digital converters

11 signal processing unit

13 digital filters

15 mean value filter

17 notch filters

[0114]

Claims

claims

1. A method for signal processing for measuring signals (U _s ) of a

Vortex flow sensor for measuring a flow of a medium through a measuring tube (1), - the one in the measuring tube (1) arranged bluff body (3) and - a sensor (5) for detecting in the region of the bluff body (3) occurring pressure fluctuations and for the conversion of these Pressure fluctuations in an electrical measurement signal (U _s ), in which - at least a portion of the measurement signal (U _s ) is sampled and digitized, - an autocorrelation (AK (T)) of the digitized measurement signal (U _s ) is calculated, and - Flow is derived from at least one property of autocorrelation (AK (T)).

2. The method of claim 1, wherein - the property is a frequency (f) or a period (P) of the autocorrelation (AK (T)), and - the frequency (f) or the period (P) of Autocorrelation (AK (T)) equal to a frequency (f) or a period (P) of the measuring signal (U _s ) is set.

3. The method of claim 1 or 2, wherein - a zero point (NSL) or a

Minimum (Ml) of the autocorrelation (AK (T)) is determined, and - the property of the autocorrelation (AK (T)) is determined on the basis of the position of the zero (NSl) or the minimum (Ml).

4. The method according to claim 3, wherein the position of the minimum (Ml) by a

Adjustment of a parabola (p (T)) to points of autocorrelation (AK (T)) is determined.

5. The method according to claim 1 or 2, wherein - the property of

Autocorrelation (AK (T)) is the frequency (f) or period (P) of the same, and - this is determined by fitting a cosine function to the autocorrelation (AK (T)).

6. The method of claim 1, wherein the measuring signal (U _s ) before the formation of the

Autocorrelation (AK (T)) an adaptive filter (7, 13, 17) passes through.

7. The method of claim 1, wherein the adaptive filter (7, 13) an adaptive

Bandpass filter is.

8. The method of claim 7, wherein -a frequency (f) of the autocorrelation

(AK (T)) is determined, and - a frequency range in which the bandpass filter is transmissive is set on the basis of the frequency (f) of the autocorrelation (AK (T)). 9. The method of claim 7, wherein the adaptive band-pass filter is an adaptive

Line Enhancer is used.

10. The method of claim 6, wherein the adaptive filter (17) is a notch filter.

11. The method of claim 10, wherein the adaptive notch filter - a

Performs filter function (H (z)) which supplies a residual signal (R) which serves to optimize the filter (17), and - performs a complementary filter function (lH (z)) which delivers the measurement signal (U _s ), on the basis of which the autocorrelation (AK (T)) is determined.

12. vortex flow sensor for measuring a flow of a

Medium through a measuring tube (1), with - in the measuring tube (1) arranged bluff body (3) and - a sensor (5) for detecting occurring in the region of the bluff body (3) pressure fluctuations and for converting these pressure fluctuations into an electrical measurement signal ( U _s ), an analog-to-digital converter (9), which serves to scan and digitize at least a part of the measurement signal (Us), and a signal processing (11), which serves to generate an autocorrelation (AK T)) of the digitized measurement signal (U _s ) and derive the flow on the basis of at least one property of the autocorrelation (AK (T)).