Method of operating a magnetic inductive flowmeter and corresponding magnetic inductive flowmeter

Abstract

A method for operating a magnetic-inductive flowmeter is described, the magnetic-inductive flowmeter having a measuring tube for conducting a medium, a magnetic-field generating device for generating a magnetic field which penetrates the measuring tube perpendicularly to the flow direction of the medium, an electrode pair for measuring a voltage induced in the medium as a noisy raw measurement signal, wherein the noisy raw measurement signal is forwarded to a signal processing device and at least a plurality of recorded noisy raw measurement signals are processed by the signal processing device into de-noised flow measurement values. The reliability of the determined measurement values can be determined by determining a plurality of de-noised comparative flow measurement values, determining a comparative frequency spectrum for the de-noised comparative flow measurement values, calculating a current de-noised flow measurement value, determining a current measurement frequency spectrum, comparing the current measurement frequency spectrum with one of the comparative frequency spectra and determining a deviation value, and the current de-noised flow measurement value being marked as unreliable and / or as reliable.

Description

Technical Field

[0001] This invention relates to a method for operating a magnetic induction flow meter, the flow meter comprising: a measuring tube for guiding a medium; a magnetic field generating device for generating a magnetic field that penetrates the measuring tube perpendicular to the flow direction of the medium; and an electrode pair for measuring a voltage induced in the medium within the measuring tube as a noisy raw measurement signal, wherein the noisy raw measurement signal is forwarded to a signal processing device, and the signal processing device processes at least a plurality of recorded noisy raw measurement signals into a denoised flow measurement value. Furthermore, this invention also relates to such a magnetic induction flow meter. Background Technology

[0002] The flow meters mentioned earlier have been known for decades, based on the principle of magnetic induction measurement. Therefore, the methods used to operate such flow meters, as described previously, have also been known for decades. The magnetic induction measurement principle is based on the fact that forces act on charge carriers that move perpendicular to or have a component of motion perpendicular to the relevant magnetic field. For flow measurement to be performed using this principle, the medium guided in the measuring tube must have a certain degree of conductivity. The faster the medium moves through the measuring tube and thereby through the magnetic field generated by the magnetic field generating device, the stronger the separation of charge carriers in the corresponding section of the measuring tube, and the stronger the electric field caused by charge separation becomes. This electric field forms between the electrodes of the measuring tube and can be considered as a voltage between these electrodes. The measured voltage between these electrodes is proportional to the flow velocity, at least for a period of time during which the magnetic field is constant and the conductivity or charge carrier concentration in the medium remains constant.

[0003] Even though the basic principles of magnetic induction measurement are quite clear, several obstacles remain to be overcome in practical measurement techniques until a flow meter that reliably provides accurate flow information is available. One of these obstacles is that the voltage induced in the medium serves as a significantly noisy raw measurement signal attached to the electrodes of the electrode pair. The signal-to-noise ratio of this noisy raw measurement signal is very unfavorable, making it impossible to directly extract reliable and stable flow information from the noisy raw measurement data. The cause of this noise is essentially the electrochemical processes at these electrodes.

[0004] To reduce noise caused by electrochemical processes, it is known in the prior art, for example, that the direction of the magnetic field is continuously changed, causing the voltage induced in the medium to also change direction. For this purpose, the energizing direction of the coil in the magnetic field generating device is regularly changed. Because the magnetic field direction cannot be changed immediately due to inductance, and there is always a transition region where the magnetic field is not constant, it must be ensured that only the original measurement signal detected under a constant magnetic field is used to determine the flow rate. While the described switching of the magnetic field can reduce some electrochemical effects, the original measurement signal at these electrodes still has significant noise. To develop the noisy original measurement signal into a suitable, denoised flow measurement, the recorded noisy original measurement signal is denoised in a signal processing device, for example, by averaging multiple recorded noisy original measurement signals. The denoised flow measurement value thus obtained is then typically displayed and / or output via a valid signal interface, typically via a 4-20 mA interface (often using a superimposed digital HART protocol).

[0005] For users of flow meters, it is not only the accuracy of the measurements provided by the flow meter when it is fault-free that is of interest. Depending on the application of the flow meter, it may be very important to identify whether there are any faults in the flow meter, especially those that affect the measurement values. Such faults are sometimes potentially dangerous because they do not occur spontaneously and clearly, but rather occur in a subtle process and are therefore particularly difficult to detect. Summary of the Invention

[0006] Therefore, the object of the present invention is to describe a method for operating a magnetic induction flow meter and a magnetic induction flow meter operated using this method, wherein the reliability of the determined measurement value can be determined by the method and the magnetic induction flow meter, and wherein the method and the magnetic induction flow meter can thus warn the user of the flow meter of unreliable measurement values.

[0007] The previously proposed and described task, in the case of the method for operating a magnetic induction flow meter described at the beginning, and in the case of the magnetic induction flow meter described, is primarily and substantially solved by: in the calibration step, determining multiple denoised comparative flow measurements; and determining comparative spectra for each of these comparative flow measurements based on at least a portion of the multiple recorded noisy original measurement signals upon which the calculation of these denoised comparative flow measurements is based. This process can be performed at the manufacturer's facility, which has the advantage that the volumetric flow rate can be accurately specified as the comparative flow rate. However, the process can also be performed in the installation state. This has the advantage that the effects related to individual installation conditions are also taken into account. This calibration step does not need to be repeated continuously during the normal operation of the magnetic induction flow meter; rather, performing it once is sufficient. It is also reasonable to perform the calibration process again, for example, after the inspection of the magnetic induction flow meter. In any case, it should be ensured that this step is performed when the magnetic induction flow meter is in a fault-free state.

[0008] The method according to the invention further includes, during measurement operation, calculating a current denoised flow measurement value based on a plurality of recorded noisy raw measurement signals; this calculation is a fairly "normal" determination of the current measurement value. Additionally, a current measurement spectrum is determined based on at least a portion of the plurality of recorded noisy raw measurement signals on which the calculation of the current denoised flow measurement value is based.

[0009] Finally, in the comparison step, the current measurement spectrum is compared with one of these comparison spectra, and a deviation value is determined. Based on this deviation value, the current denoised flow measurement is marked as unreliable and / or marked as reliable. That is, either only the unreliable current denoised flow measurement is marked as unreliable (not marked as "reliable" qualitatively), or only the reliable current denoised flow measurement is marked as reliable (not marked as "unreliable" qualitatively), or the corresponding qualitative marking occurs in both cases; a specific current denoised flow measurement certainly cannot be marked as both "reliable" and "unreliable" simultaneously.

[0010] Specifically, it is stipulated that if the deviation value deviates from the expected value, the current denoised flow measurement value shall be marked as unreliable.

[0011] The present invention is based on the understanding that these noisy raw measurement signals contain not only information about the induced voltage, i.e., flow information, but more precisely, information that has not been used to date, but can still have high added value for the user. It has been recognized that the spectrum of the noisy raw measurement signal, which is used as the basis for the denoised flow measurement by averaging, changes systematically with the flow rate. Therefore, the spectrum obtained from the noisy raw measurement signal is characteristic for the determined denoised flow measurement. If, under fault-free conditions, i.e., in the calibration steps mentioned herein, the corresponding comparison spectrum is determined for a series of comparative flow measurements, it is therefore possible to check the reliability of the current denoised flow measurement by determining the current measurement spectrum based on the noisy raw measurement signal on which the current denoised flow measurement is based and comparing that measurement spectrum with a comparison spectrum that belongs to a comparative flow measurement as similar as possible to the current denoised flow measurement.

[0012] An extended scheme of this method specifies that the amplitude spectrum, i.e., the amplitude spectrum of the comparison spectrum and the amplitude spectrum of the current measurement spectrum, is used as the spectrum, which is obtained, in particular, by Fast Fourier Analysis (FFT) of the recorded noisy original measurement signal. That is, the recorded noisy original measurement signal is here in digitized form and exists in the sampling time grid. The general result of frequency analysis is usually a complex spectrum, which can be divided into a frequency-dependent amplitude spectrum and a frequency-dependent phase spectrum according to amplitude and phase. It has been shown that sufficiently good comparison results are achieved even when only the corresponding amplitude spectra are used, in comparing the current measurement spectrum with one of the comparison spectra.

[0013] According to a preferred design of this method, in the comparison step, the area between the current measured spectrum and the comparison spectrum used is calculated as the deviation value. Since the most significant changes in these spectra exist at low frequencies, it is particularly specified that the area between these spectra—especially between these amplitude spectra—is calculated in a frequency range up to 1 kHz, preferably up to 500 Hz, and most preferably up to 300 Hz.

[0014] An extension of this method is characterized by smoothing at least one spectrum, particularly by applying a moving average, preferably by applying an equally weighted moving average, or particularly by applying a Savitzky-Golay filter, especially when using a 2nd or 3rd order polynomial and especially when using a range width between 10 and 16, particularly preferably using a range width between 11 and 15. The spectra obtained from these recorded, noisy raw measurement signals are also relatively noisy, i.e., they exhibit significant differences between adjacent amplitude values ​​(and, if necessary, phase values, if used for evaluation). The proposed smoothing of these spectra has proven advantageous in order to facilitate the evaluation of the spectra to be compared and to make them less susceptible to fluctuations. The range of values ​​specified for the range width and the polynomial order have proven advantageous in order to avoid making the computation of the smoothed spectrum too expensive.

[0015] This comparison step can be implemented in different ways in specific implementation schemes. According to the first variant, the denoised comparison flow measurement value and its associated comparison spectrum calculated in the calibration step are at least partially stored. In the comparison step, the measured spectrum is compared with the comparison spectrum that is least different from the current denoised flow measurement value, and a corresponding deviation value is determined. That is, in this variant, a relatively large number of comparison flow measurements, and especially their associated comparison spectra, must be recorded and stored to ensure good distinguishing accuracy between reliable or credible current denoised flow measurements and unreliable or untrustworthy current denoised flow measurements. Since this comparison spectrum, which must be very similar to the current measurement spectrum under fault-free conditions, is always used, it is specifically stipulated that the expected value for the deviation value is limited to a range near zero.

[0016] According to the second variant, in the calibration step, one of the multiple comparison spectra is selected and stored as the reference spectrum—that is, selected and stored as the reference spectrum for all comparisons. For the other comparison spectra, reference deviation values ​​from the reference spectrum are determined individually, as in the comparison step. These reference deviation values ​​must be stored for later comparisons; however, it is not necessary to store comparison spectra different from the reference spectrum. Next, in the comparison step, the current measured spectrum is compared with the reference spectrum, and the deviation value is determined. A suitable reference deviation value or a value derived from the reference deviation value is used as the expected value. In this variant of the method, although a series of comparison spectra with their respective comparison flow measurements must also be recorded first, this is only for the calibration step.

[0017] In the above-described method variant, it is also preferable to specify the comparison step as follows: from among multiple reference deviation values, the reference deviation value to which the comparison flow measurement value is closest to the current measured flow measurement value whose reliability is to be checked is selected as the appropriate reference deviation value.

[0018] In a particularly preferred extension of this method, it is possible to achieve significantly higher resolution, i.e., to establish a more accurate boundary between reliable current flow measurements and impossible current flow measurements, at very low cost. This can be achieved by deriving a functional relationship between the deviation value and the comparative flow measurement value, based on a pair of values ​​consisting of the deviation value between the comparison spectrum and the reference spectrum and the comparative flow measurement value belonging to the comparison spectrum. This functional relationship enables the determination of a reference deviation value—even if estimated—for each current flow measurement value. That is, the comparison is no longer limited to using multiple reference deviation values ​​determined in the calibration step, but in practice can utilize any number of—approximate—reference deviation values.

[0019] An extension of the above method is characterized by the fact that the functional relationship is based on an approximate function consisting of the obtained value pairs of the deviation between the comparison spectrum and the reference spectrum and the comparison flow measurement values ​​belonging to the comparison spectrum, especially based on an interpolation function or compensation function, and especially based on a linear or quadratic interpolation function or compensation function.

[0020] This method does not necessarily need to be run entirely on the corresponding magnetic induction flowmeter involved in its operation. Specifically, the detected noisy raw measurement signal can be sent to an external computing unit via the interface of the magnetic induction flowmeter, where all other calculations can be performed. In particular, the current denoised flow measurement value and the associated detected raw measurement signal can also be sent via the interface, and the calculation of the current measurement spectrum can be performed using the external computing unit. All method steps of the calibration and comparison steps can also be performed outside the magnetic induction flowmeter. Importantly, there is ultimately information regarding whether the current denoised flow measurement value is reliable or unreliable.

[0021] The proposed task is also addressed in the case of the magnetic induction flowmeter mentioned at the beginning, more specifically by storing at least one comparison spectrum in memory, which is calculated based on multiple recorded noisy raw measurement signals for comparing flow measurements. The calibration process can, but does not necessarily, be performed entirely through the magnetic induction flowmeter. Of course, these noisy raw measurement signals must be detected in the magnetic induction flowmeter, but the comparison spectrum does not necessarily need to be calculated at the magnetic induction flowmeter. During measurement operation, the current denoised flow measurement value is calculated based on the multiple recorded noisy raw measurement signals, and the current measurement spectrum is determined based on at least a portion of the multiple recorded noisy raw measurement signals on which the calculation of the current denoised flow measurement value is based. In the comparison step, the current measurement spectrum is compared with one of these comparison spectra, and a deviation value (D) is determined. If the deviation value deviates from the expected value, the current denoised flow measurement value is marked as unreliable.

[0022] Preferably, in order to indicate the unreliability of the current denoised flow measurement value, a corresponding flag is set in the memory, or a corresponding signal is displayed on the display device of the magnetic induction flow meter, or a corresponding message is sent via the communication interface of the magnetic induction flow meter, for example, the message is sent to the process control system or to the connected operating equipment.

[0023] In a preferred design of the magnetic induction flow meter, it is specified that the signal processing device and / or another computing unit are designed to enable the magnetic induction flow meter to execute the method steps described in detail before operation.

[0024] Of course, each of the previously described features, which may be used in conjunction with the method for operating a magnetic induction flow meter, can also be objectively applied to a magnetic induction flow meter designed to implement the described method. Conversely, features that may have been described previously only in the context of a magnetic induction flow meter can also be understood as features of the method for operating that magnetic induction flow meter as described herein, or more precisely, features that can be understood individually or in combination with other features as features of the method for operating that magnetic induction flow meter as described herein. Attached Figure Description

[0025] The invention provides detailed possibilities for designing and extending methods for operating magnetic induction flow meters according to the invention, and for corresponding magnetic induction flow meters according to the invention. For this purpose, reference is made, on the one hand, to the dependent claims of the independent claims, and on the other hand, to the following description of embodiments taken in conjunction with the accompanying drawings. In the drawings:

[0026] Figure 1A magnetic induction flow meter according to the present invention and a corresponding method for operating the magnetic induction flow meter are illustrated schematically.

[0027] Figure 2 The comparison spectrum (amplitude spectrum) is shown, calculated from the recorded noisy raw measurement signal.

[0028] Figure 3 It shows the low-frequency case Figure 2 Details in the text;

[0029] Figure 4 The described method for operating a magnetic induction flow meter is illustrated as a flowchart.

[0030] Figure 5a , 5b Reference spectra with comparison spectra are shown respectively;

[0031] Figure 6 The reference deviation value, which depends on the flow rate, is shown.

[0032] Figure 7 This illustrates the functional relationship between the reference deviation value and the comparative flow measurement value, with tolerance bands for the expected value in the case of a measuring tube with a reduced circular diameter; and

[0033] Figure 8 The functional relationship between the reference deviation value and the comparative flow measurement value, with tolerance bands for the expected value in the case of a measuring tube with a reduced angular diameter, is shown. Detailed Implementation

[0034] exist Figures 1 to 8 The text describes a method 1 for operating a magnetic induction flow meter 2, with different emphases. Figure 1 The structure of the magnetic induction flow meter 2 is shown in the figure.

[0035] The magnetic induction flow meter 2 includes: a measuring tube 3 for guiding the medium; a magnetic field generating device 4 for generating a magnetic field that penetrates the measuring tube 3 perpendicular to the flow direction of the medium; and an electrode pair 5 for measuring the voltage induced in the medium in the measuring tube as a noisy raw measurement signal 6. This arrangement is often referred to as the "sensor" of the magnetic induction flow meter 2.

[0036] The noisy raw measurement signal 6 is processed through signal path 7 (see Figure 1The signal recording device 8 records the noisy raw measurement signal 9 from the electrode pair 5 with high impedance and forwards it to the signal processing device 10 as the recorded noisy raw measurement signal 9. The signal processing device 10 processes the recorded noisy raw measurement signal 9 into at least a denoised flow measurement value 11, and the denoised flow measurement value 11 is then output via a valid signal 12 in the illustrated embodiment. In the illustrated embodiment, the valid signal interface 12 is a two-wire power interface with superimposed HART protocol, as is widely used in the process industry. However, the fact that a denoised flow measurement value 11 is output is not important in the present case.

[0037] exist Figure 1 In the diagram, the first signal path 7 is enclosed by a dashed box. These components of the magnetic induction flow meter 2 typically form what is known as a transmitter. In the signal processing device 10, in order to denoise the recorded noisy raw measurement signal 9, multiple individual values ​​are averaged and processed into a usable denoised flow measurement value 11.

[0038] The method 1 for operating the magnetic induction flow meter 2 shown in these figures, and therefore also in Figure 1 The magnetic induction flowmeter 2 shown in the figure is characterized by (as) Figure 4 (Flowchart shown in the diagram): In calibration step 15, a plurality of denoised comparative flow measurements CQ1, CQ2, and CQ3 are determined, i.e., the plurality of denoised comparative flow measurements are determined by averaging a plurality of recorded noisy original measurement signals 9. Based on at least a portion of the plurality of recorded noisy original measurement signals 9 upon which the calculations of these denoised comparative flow measurements CQ1, CQ2, and CQ3 are based, comparative spectra CS1, CS2, and CS3 are determined for these denoised comparative flow measurements CQ1, CQ2, and CQ3, respectively. Figure 2 and Figure 3 The diagram shows the comparison spectra CS1, CS2, and CS3 determined for the comparison flow measurements CQ1, CQ2, and CQ3. To clarify that the comparison spectra CS1, CS2, and CS3 are calculated using the recorded, noisy raw measurement data 9, which also forms the basis for determining the corresponding comparison flow measurements CQ1, CQ2, and CQ3, the comparison spectra CS1, CS2, and CS3 are represented by the symbols CS1(CQ1), CS2(CQ2), and CS3(CQ3).

[0039] The letter indicators used here for different method objects are intended to better identify the parameters of the presented method object. Strictly speaking, these letter indicators are not formula symbols, but rather more like "speaking" reference symbols. Figure 4In this process, the relationship between the comparative flow measurement value CQi and the corresponding comparative spectrum CSi is illustrated by the allocation arrows in calibration step 15. Importantly, the comparative spectra CS1, CS2, and CS3 calculated for the denoised comparative flow measurements CQ1, CQ2, and CQ3 are characterizing for the comparative flow measurements CQ1, CQ2, and CQ3, respectively; the method used to determine the reliability, reasonableness, and credibility of the current flow measurement value is based on this understanding.

[0040] exist Figure 2 and Figure 3 It can be seen that the amplitude of the comparison spectrum CSi changes systematically with the changing flow measurement value CQi, especially at low frequencies. This relationship is used in method 1 according to the invention. Here, the comparison flow measurement values ​​CQi are selected such that these comparison flow measurement values ​​cover as much of the entire measurement range used as possible.

[0041] Calibration step 15 is typically performed once when the magnetic induction flow meter 2 is in a trouble-free state, such as as a factory calibration or immediately after the magnetic induction flow meter 2 is installed during the user's process.

[0042] In the subsequent measurement run 16, the current denoised flow measurement value MQ is calculated based on a plurality of recorded noisy raw measurement signals 9. The current measurement spectrum MS is determined based on at least a portion of the plurality of recorded noisy raw measurement signals 9 on which the calculation of the current denoised flow measurement value MQ is based. For clarity, this measurement spectrum MS is not shown in these figures; it is considered as another spectrum within the range of the comparison spectrum CSi.

[0043] Finally, in comparison step 17, the current measured spectrum MS is compared with one of these comparison spectra CSi and the deviation value D is determined. Figure 4 In this context, the determination of the deviation value D is symbolically described by the functional diff(MS,CSi). Based on the deviation value D, the current denoised flow measurement value MQ is marked as unreliable. In the embodiment, it is specified that the deviation of the deviation value D from the expected value DE indicates that the current denoised flow measurement value MQ has been classified as unreliable (invalid), otherwise the flow measurement value MQ is reliable (valid).

[0044] In the illustrated embodiment, method 1 is implemented such that the amplitude spectrum, i.e., the amplitude spectrum of the comparison spectrum CS and the amplitude spectrum of the current measurement spectrum MS, are used as the spectra CS and MS, respectively. Since the recorded noisy raw measurement signal 9 is already a digitized value (A / D conversion) in the time-sampling system, all these spectra are obtained here through fast Fourier analysis of the recorded noisy raw measurement signal 9. Figure 1 It can be seen that the recorded noisy raw measurement signal 9 is transmitted to a second signal processing device 14 via another signal path 13, where frequency analysis is performed. Calculations may also be performed in the signal processing device 10. Here, in order not to jeopardize the existing certification of the first signal path 7 (in this case, regarding the Safety-Integrity-Level SIL), the shown path of a separate signal path 13 has been followed.

[0045] In all design schemes of Method 1, the process in the current case is as follows: In comparison step 17, the area between the current measured spectrum MS and the used comparison spectrum CS is calculated as the deviation value D. Specifically, the area calculation (not shown in detail) is performed over a frequency range of up to 500 Hz. Especially... Figure 3 It can be seen that within this range, there are particularly large variations in these spectra under different traffic conditions, making it particularly easy to identify the differences between the spectra to be compared.

[0046] in accordance with Figure 3 As shown in Figure 5, the spectrum CS obtained from the noisy original measurement signal 9 is also very noisy, i.e., it has a large amplitude difference even when the amplitudes are adjacent at frequencies. Consequently, the area calculation between the two spectra described earlier is also very unstable, even for spectra determined for closely spaced flows. To achieve characteristics with smaller fluctuations, the spectrum CS and MS are smoothed, and in the current case—even if not explicitly presented—the frequencies are smoothed by applying an equally weighted moving average using 20 adjacent spectral values.

[0047] In the illustrated embodiment, the permissible deviation of the deviation value D from the expected value DE is described by a tolerance band 18 around the expected value DE or around the curve of the expected value DE. If the deviation value D is within the tolerance band 18, it is considered that D=DE is satisfied and a reliable measurement exists (see [link to relevant documentation]). Figure 7 and Figure 8 ).

[0048] There are different possibilities for implementing comparison step 17. Two variant schemes are then described.

[0049] The first variation of method 1 specifies that, for example, like Figure 2 and Figure 3 As shown, the denoised comparative flow measurement value CQi and its corresponding comparative spectrum CSi calculated in calibration step 15 are at least partially stored, and in comparison step 17, the measured spectrum MS is compared with the comparative spectrum CSi and a deviation value D is determined, which comparative spectrum CSi belongs to the denoised comparative flow measurement value CQi with the smallest difference relative to the current denoised flow measurement value MQ. Figure 3 The diagram shows different comparison spectra CSi, i.e., comparison spectra CS1, CS2, and CS3, to which the comparison flow measurement values ​​CQi, i.e., CQ1, CQ2, and CQ3, belong. The specific flow values ​​for these are 0 m³ / h, 18 m³ / h, and 29.8 m³ / h. The measurement spectrum MS is not shown. The above variant scheme can still be based on... Figure 3 To be explained.

[0050] If we assume that the current denoised flow rate value MQ is determined to be MQ = 17 m^3 / h and that the measurement is error-free, then the corresponding measured spectrum MS must be compared with the comparison spectrum CS2, because the comparison flow rate measurement CQ2 belonging to the comparison spectrum CS2 is closest to the current denoised flow rate measurement value MQ. Since there is no error, the measured spectrum MS will practically match the comparison spectrum CS2. If the area difference between the two spectra MS and CS2 is calculated as the deviation value D, then this area difference is close to zero. This is always the case in the fault-free condition, so the expected value E defines a range of values ​​around zero. In the fault condition, these spectra are separated further by an area difference that is more or less significantly different from zero, so the current denoised flow rate measurement to be tested is identified as unreliable and correspondingly marked as unreliable.

[0051] Correspondingly, the second variation of Method 1 is characterized in that, in calibration step 15, one of the multiple comparison spectra CSi is selected and stored as the reference spectrum CSref. For the other comparison spectra CSi, i.e., in principle as in comparison step 17, reference deviation values ​​Di,ref from the reference spectrum CSref are determined respectively. In comparison step 17, the current measured spectrum MS is compared with the reference spectrum CSref and the deviation value D is determined. The suitable reference deviation value Di,ref, or the value Dref derived from the reference deviation value Di,ref, is used as the expected value DE.

[0052] Method 1 can be based on Figure 5 to... Figure 7 This is to be explained, even though the current measurement spectrum MS is not shown here. Figure 5a and Figure 5b The spectrum in corresponds to in Figure 2 and Figure 3 The spectrum in the comparison spectrum. The comparison spectrum CS1 is selected and stored as the reference spectrum CSref. For each other comparison spectrum, i.e., for comparison spectra CS2 and CS3, reference deviation values ​​D2,ref and D3,ref are determined (in Figure 5a and Figure 5b The difference in area between the lines in the diagram). That is, in Figure 4 In the notation, comparison step 17 must be: D2,ref=diff(CS2, CSref) and D3,ref=diff(CS3, CSref). Only these values ​​must be stored, not the comparison spectra CS2 and CS3. If the current denoised flow measurement value MQ and its current measurement spectrum MS exist, then the measurement spectrum MS is compared with the reference spectrum CSref and the corresponding deviation value D is obtained, i.e.: D=diff(MS,CSref).

[0053] If we again assume that the current denoised flow measurement has been determined to be MQ = 17 m³ / h and that the measurement is error-free, then the appropriate reference deviation value Di,ref as the expected value DE is the reference deviation value D2,ref, because this reference deviation value D2,ref is the one among multiple reference deviation values ​​Di,ref to which its corresponding comparative flow measurement value CQ2 is closest to the current measured flow measurement value MQ. Therefore, assuming no error in determining the current denoised flow measurement value MQ, D = D2,ref is practically applicable, meaning D is close to D2,ref, especially within the tolerance range of 18 around D2,ref.

[0054] Therefore, the example produced works well because it thoughtfully accepts the current denoised flow measurement value MQ, which is very close to the available comparative flow measurement value, CQ2. To achieve good coverage of the measurement range of the magnetic induction flowmeter 2 using method 1, it is necessary to work with a larger number of comparative flow measurements CQ, which... Figure 6 The values ​​are outlined in the diagram. Here, eight reference deviation values ​​Di,ref have been determined for either the eight comparative flow measurements CQ or the comparative spectra assigned to these eight comparative flow measurements. The tolerance range 18 is also described.

[0055] According to Figure 6As can be seen, the reference deviation values ​​Di,ref are arranged such that the functional relationship Dref(CQ) between the reference deviation value Dref and the comparative flow measurement value CQ can be derived from the obtained value pair (Di,ref, CQi) consisting of the deviation value Di,ref between the comparison spectrum CSi and the reference spectrum CSref and the comparative flow measurement value CQi belonging to the comparison spectrum CSi. In particular, this functional relationship is based on an approximate function of the obtained value pair (Di,ref, CQi) consisting of the deviation value Di,ref between the comparison spectrum CSi and the reference spectrum CSref and the comparative flow measurement value CQi belonging to the comparison spectrum CSi, where in Figure 7 The quadratic compensation function was calculated.

[0056] Interestingly, this relationship Dref(CQ) clearly depends on the geometry of measuring tube 3. Figure 7 The measuring tube 3 in the middle has a reduced circular diameter (to increase the flow velocity in the measuring section), while... Figure 8 The measuring tube 3 in the middle has a reduced rectangular diameter. In the first case, the relationship Dref(CQ) can be simulated very well by means of a quadratic compensation function, and in the second case, even a linear compensation function is sufficient.

[0057] Figure Labels

[0058] 1. Method for operating a magnetic induction flow meter

[0059] 2. Magnetic induction flow meter

[0060] 3 Measuring tube

[0061] 4. Magnetic field generating device

[0062] 5 electrode pairs

[0063] 6. Noisy raw measurement signal

[0064] 7 signaling pathways

[0065] 8. Signal recording device

[0066] 9. Recorded noisy raw measurement signals

[0067] 10. Signal Processing Device

[0068] 11. Noise-reduced flow measurement values

[0069] 12 valid signal interfaces

[0070] 13 Another signaling pathway

[0071] 14 Second Signal Processing Device

[0072] 15 Calibration Procedure

[0073] 16 Measurement Operation

[0074] 17 Comparison Steps

[0075] 18. Tolerance zone around the expected value or Dref(CQ)

[0076] Comparison of CQ and CQi noise-reduced flow measurement values

[0077] Comparison of CS and CSi spectra

[0078] DE Expected value for deviation

[0079] CSref reference spectrum

[0080] MQ's current denoised traffic measurement

[0081] MS measurement spectrum

[0082] D, Di deviation values

[0083] Dref; Di,ref reference deviation value

[0084] The Dref(CQ) function relationship.

Claims

1. A method (1) for operating a magnetic induction flow meter (2), the magnetic induction flow meter comprising: a measuring tube (3) for guiding a medium; a magnetic field generating device (4) for generating a magnetic field that penetrates the measuring tube (3) perpendicular to the flow direction of the medium; and an electrode pair (5) for measuring a voltage induced in the medium in the measuring tube (3) as a noisy raw measurement signal (6), wherein the noisy raw measurement signal (6) is forwarded to a signal processing device (10) as a recorded noisy raw measurement signal (9) and the signal processing device (10) processes at least a plurality of recorded noisy raw measurement signals (9) into a denoised flow measurement value (11). Its features are, In calibration step (15), multiple denoised comparative flow measurements (CQ1, CQ2, CQ3) are determined. Based on at least a portion of the plurality of recorded noisy original measurement signals (9) on which the calculation of the denoised comparison flow measurements (CQ1, CQ2, CQ3) is based, comparison spectra (CS1, CS2, CS3) are determined for the denoised comparison flow measurements (CQ1, CQ2, CQ3). During measurement operation (16), the current denoised flow measurement (MQ) is calculated based on multiple recorded noisy raw measurement signals (9). The current measurement spectrum (MS) is determined based on at least a portion of the plurality of recorded noisy original measurement signals (9) on which the calculation of the current denoised flow measurement (MQ) is based. In the comparison step (17), the current measurement spectrum (MS) is compared with one of the comparison spectra (CSi) and a deviation value (D) is determined, wherein the deviation value (D) is calculated as the area between the current measurement spectrum (MS) and the comparison spectrum (CS) used, and based on the deviation value (D), the current denoised flow measurement value (MQ) is marked as unreliable and / or marked as reliable.

2. The method (1) according to claim 1, characterized in that, If the deviation value (D) deviates from the expected value (DE), the current denoised flow measurement value (MQ) is marked as unreliable.

3. The method (1) according to claim 1 or 2, characterized in that, The amplitude spectrum, namely the amplitude spectrum of the comparison spectrum (CS) and the amplitude spectrum of the current measurement spectrum (MS), are used as the spectrum (CS, MS), wherein the spectrum is obtained by fast Fourier analysis of the recorded noisy original measurement signal (9).

4. The method (1) according to claim 1 or 2, characterized in that, In the comparison step (17), the area between the current measurement spectrum (MS) and the comparison spectrum (CS) used is calculated as the deviation value (D) in a frequency range up to 1 kHz.

5. The method (1) according to claim 4, characterized in that, In the comparison step (17), the area between the current measurement spectrum (MS) and the comparison spectrum (CS) used is calculated as the deviation value (D) in a frequency range up to 500 Hz.

6. The method (1) according to claim 4, characterized in that, In the comparison step (17), the area between the current measurement spectrum (MS) and the comparison spectrum (CS) used is calculated as the deviation value (D) in a frequency range up to 300 Hz.

7. The method (1) according to claim 1 or 2, characterized in that, Smooth at least one spectrum (CS, MS).

8. The method (1) according to claim 7, characterized in that, The at least one spectrum is smoothed by applying a moving average.

9. The method (1) according to claim 8, characterized in that, The at least one spectrum is smoothed by applying an equal-weighted moving average.

10. The method (1) according to claim 7, characterized in that, The at least one spectrum is smoothed by applying a Savitsky-Gore smoothing filter.

11. The method (1) according to claim 10, characterized in that, The at least one spectrum is smoothed by applying a Savitsky-Gore smoothing filter, using a second- or third-order polynomial.

12. The method (1) according to claim 10, characterized in that, The at least one spectrum is smoothed by applying a Savitsky-Gore smoothing filter, using a range width between 10 and 16.

13. The method (1) according to claim 10, characterized in that, The at least one spectrum is smoothed by applying a Savitsky-Gore smoothing filter, using a range width between 11 and 15.

14. The method (1) according to claim 2, characterized in that, The permissible deviation of the deviation value (D) from the expected value (DE) is described by a tolerance zone (18) around the expected value (DE) or around the curve of the expected value (DE).

15. The method (1) according to claim 1, 2 or 14, characterized in that, The denoised comparative flow measurement (CQ) and its associated comparative spectrum (CS) calculated in calibration step (15) are at least partially stored, and in the comparison step, the measured spectrum (MS) is compared with a comparative spectrum (CSi) and a deviation value (D) is determined, which is the denoised comparative flow measurement (CQi) with the smallest difference relative to the current denoised flow measurement (MQ), wherein the expected value (E) defines a range of values ​​around zero.

16. The method (1) according to claim 1, 2 or 14, characterized in that, In calibration step (15), one of the multiple comparison spectra (CS) is selected and stored as a reference spectrum (CSref); for the other comparison spectra (CSi), a reference deviation value (Di,ref; Dref) is determined from the reference spectrum (CSref) as in comparison step (17); in comparison step (17), the current measurement spectrum (MS) is compared with the reference spectrum (CSref) and the deviation value (D) is determined, wherein a suitable reference deviation value (Di,ref) or a value (Dref) derived from the reference deviation value (Di,ref) is used as the expected value (DE).

17. The method (1) according to claim 16, characterized in that, The following reference deviation value is selected from a plurality of reference deviation values ​​(Di,ref) as the appropriate reference deviation value (Di,ref), wherein the comparative flow measurement value (CQi) to which the reference deviation value belongs is the closest to the current measured flow measurement value (MQ) among the plurality of reference deviation values ​​(Di,ref).

18. The method (1) according to claim 16, characterized in that, The functional relationship (Dref(CQ)) between the reference deviation value (Dref) and the comparative flow measurement value (CQ) is derived based on the obtained value pair (Di,ref, CQi) consisting of the deviation value (Di,ref) between the comparison spectrum (CSi) and the reference spectrum (CSref) and the comparison flow measurement value (CQi) belonging to the comparison spectrum (CSi).

19. The method (1) according to claim 18, characterized in that, The functional relationship is based on an approximate function of the obtained value pair (Di,ref,CQi) consisting of the deviation (Di,ref) between the comparison spectrum (CSi) and the reference spectrum (CSref) and the comparison flow measurement (CQi) belonging to the comparison spectrum (CSi).

20. The method (1) according to claim 19, characterized in that, The functional relationship is based on interpolation functions or compensation functions.

21. The method (1) according to claim 19, characterized in that, The functional relationship is based on linear or quadratic interpolation functions or compensation functions.

22. A magnetic induction flow meter (2), the magnetic induction flow meter comprising: a measuring tube (3) for guiding a medium; a magnetic field generating device (4) for generating a magnetic field that penetrates the measuring tube (3) perpendicular to the flow direction of the medium; and an electrode pair (5) for measuring the voltage induced in the medium in the measuring tube (3) as a noisy raw measurement signal (6), wherein the noisy raw measurement signal (6) is forwarded to a signal processing device (10) as a recorded noisy raw measurement signal (9) and the signal processing device (10) processes at least a plurality of recorded noisy raw measurement signals (9) into a denoised flow measurement value (11). Its features are, At least one comparison spectrum (CS) is stored in the memory, which is calculated based on multiple recorded noisy raw measurement signals in terms of comparing flow measurement values ​​(CQ); During measurement operation, the current denoised flow measurement value (MQ) is calculated based on multiple recorded noisy raw measurement signals. The current measurement spectrum (MS) is determined based on at least a portion of the plurality of recorded noisy original measurement signals on which the calculation of the current denoised flow measurement (MQ) is based. In the comparison step (17), the current measurement spectrum (MS) is compared with one of the comparison spectra (CSi) and a deviation value (D) is determined, wherein the deviation value (D) is calculated as the area between the current measurement spectrum (MS) and the comparison spectrum (CS) used, and based on the deviation value (D), the current denoised flow measurement value (MQ) is marked as unreliable and / or marked as reliable.

23. The magnetic induction flow meter according to claim 22, characterized in that, To indicate the unreliability of the current denoised flow measurement (MW), a corresponding flag is set in the memory; or a corresponding signal is displayed on the display device of the magnetic induction flow meter; or a corresponding message is sent via the communication interface.

24. The magnetic induction flow meter according to claim 22 or 23, characterized in that, The signal processing device and / or another computing unit are designed to cause the magnetic induction flowmeter to perform the method steps of the characteristic portion of any one of claims 2 to 18 during operation.

Images