EP2516973A2

EP2516973A2 - Method for determining and monitoring the level of a medium in a container according to a runtime measurement method

Info

Publication number: EP2516973A2
Application number: EP10776682A
Authority: EP
Inventors: Peter KLÖFER; Winfried Mayer; Dietmar Spanke
Original assignee: Endress and Hauser SE and Co KG
Current assignee: Endress and Hauser SE and Co KG
Priority date: 2009-12-23
Filing date: 2010-11-11
Publication date: 2012-10-31
Also published as: WO2011076478A3; CN102812337A; DE102009055262A1; CN102812337B; US20120265486A1; WO2011076478A2

Abstract

The invention relates to a method for determining and monitoring the level of a medium in a container by means of a field device according to a runtime measurement method, wherein corresponding application and device related test signals and response signals to be expected from a fill level surface are determined in a teaching phase and application and device related comparison signals are determined therefrom, wherein test signals are output in the direction of the medium and application and device related response signals are received in an operating phase and the comparison signals are compared to the response signals by means of a comparison algorithm and a value for a probability of matching (w) is determined, wherein the fill level is determined and output as a measurement value upon the determined value of the matching probability exceeding a predefined threshold and/or a new test signal is output for once again determining a response signal upon falling short of the predefined threshold.

Description

Method for determining and monitoring the level of a medium in a container according to a transit time measurement method

The present invention relates to a method for determining and monitoring the level of a medium in a container according to a transit time measuring method according to claim 1.

Corresponding methods for determining and monitoring the filling level in a container are frequently used in the measuring devices of automation and process control technology. For example, the Applicant produces and distributes such level gauges under the names Prosonic, Levelflex and Micropilot, which operate according to the transit time measurement method and serve to determine and / or monitor a level of a medium in a container. These level gauges transmit a periodic transmission signal in the microwave or ultrasound range by means of a transmitting / receiving element in the direction of the surface of a medium and receive the reflected echo signals after a distance-dependent transit time. Commercially available fill level gauges working with microwaves can basically be divided into two classes. A first class, in which the microwaves are sent by means of an antenna in the direction of the medium, reflected on the product surface and then after a

distance-dependent run time are received again. And a second class, in which the microwaves are guided along a waveguide in the direction of the filling material, at the

Füllgutoberfläche be reflected due to the existing impedance jump there and the reflected waves along the waveguide back again.

As a rule, an echo function representing the echo amplitudes as a function of the transit time is formed from the received echo signals, each value of this echo function corresponding to the amplitude of an echo reflected by the transmitting element at a specific distance.

In this determined echo function, a useful echo is determined which corresponds to the reflection of the transmission signal at the product surface. From the duration of the useful echo results in a known propagation speed of the transmission signals directly the distance between the

Füllgutoberfläche and the transmitting element.

In order to simplify the echo curve evaluation, the received raw signal of the pulse sequences are not used, but the envelope, the so-called envelope, is determined. The envelope is obtained, for example, by the raw signal of the pulse trains

rectified and then filtered via a low-pass filter.

Today, in fill level sensors operating according to the transit time principle, the following processing steps are carried out when determining the distance: The response in response to the transmitted

Test signal received analog response signal (intermediate frequency signal) is depending on the sensor principle in filtered analog or after previous A / D conversion in digital stages, possibly transformed from time to frequency domain, rectified and logarithmized. The result of this

Processing chain is the so-called envelope, in soft then by means of various

Algorithms for the level echo is sought. The algorithm is selected according to more or less complex rules; in the simplest case, the algorithm only looks for the global maximum of the envelope.

In this type of evaluation, the information content of the response signal before the step of the level echo search is greatly reduced and essentially limited to the amplitude information.

There are a variety of different methods for determining the useful echo in an envelope, which can be divided into two basic methods. Either the static detection methods with static echo search algorithms and / or the dynamic detection methods with dynamic echo search algorithms, for example using history information.

According to a first method according to a static echo search method, the useful echo, which has a greater amplitude than the remaining echoes, is selected by a static echo search algorithm. Thus, the echo in the envelope with the largest amplitude is determined as the true echo.

According to a second method according to a static echo search method, it is assumed by a static echo search algorithm that the wanted echo is the first incoming echo in the envelope after the transmit pulse. Thus, the first echo in the envelope is selected as the true echo.

It is possible to combine both methods in a static echo search algorithm, e.g. a so-called first echo factor is defined. The firstchofactor is a given factor by which an echo must exceed a certain amplitude in order to be recognized as useful echo. Alternatively, a delay-dependent echo threshold may be defined which must exceed an echo in order to be recognized as a true echo.

According to a third method, the level measuring device is notified once the current level. The level gauge can identify the associated echo as a true echo based on the predetermined level and, for. through a suitable dynamic

Track echo search algorithm. Such methods are called echo tracking. In this case, maxima of the echo signal or the echo function are determined, for example, in each measuring cycle and, on the basis of the knowledge of the fill level determined in the preceding measuring cycle and an application-specific maximum rate of change of the filling level Useful echo determined. From a running time of the current useful echo thus determined, the new fill level results.

A fourth method is described in DE 102 60 962 A1. There, the useful echo is determined based on previously stored in a memory data. In this case, echo functions are derived from received echo signals which reproduce the amplitudes of the echo signals as a function of their transit time. The echo functions are stored in a table, each column serving to record one echo function each. The echo functions are stored in the columns in an order that correspond to fill levels associated with the respective echo functions. In operation, the useful echo and the associated fill level are determined by the echo function of the current one

Transmission signal determined using the table.

In DE 103 60 710 A1, a fifth method is described in which periodically transmission signals are sent in the direction of the contents, their echo signals recorded and in a

Echo function is determined, at least one echo property of the echo function is determined, and based on the echo properties of at least one previous measurement, a prediction for the expected echo properties in the current measurement is derived. The echo properties of the current measurement are determined using the prediction, and the actual fill level is determined on the basis of the echo properties. This method comes close to echo tracking in the broadest sense.

In DE 10 2004 052 1 10 A1, a sixth method is described which achieves the improvement of useful echo recognition by echoing and classifying the echoes in the envelope.

In WO 02065066 A1, a method for highly accurate measurement of the level

described by the intermediate frequency signal stored digitally and thus both the amplitude and the phase information is kept available. Using the digitized intermediate frequency signal, the fill level can be determined with millimeter precision.

DE4308373C2 describes a method which extracts the echoes and their echo features from the envelope. The echo features are the form factor, position and time and amplitude of the echo. The form factor feature is determined as the ratio between a 6dB leading width and 6dB total width of the respective echoes. For example, for a symmetric s echo, this value is 1/2. Using fuzzy logic, the probability for each echo is calculated to be a false echo, multiple echo, or true echo. The one with the highest useful echo probability is selected as the useful echo. From EP 0 459 336 a method for processing ultrasonic echo signals is also known in which the received signal is digitally sampled and stored in a memory, the received signal being the envelope of the echoes. After the recording of the received signal by the signal processing, the echoes by means of a suitable method, for. B.

Optimal filter and a threshold detection extracted and detected all occurring within a measurement echoes.

Also known are methods for suppressing unwanted echoes contained in the received signal, for example, due to disturbing objects that are in addition to the measurement object in the detection range of the sensor. If the interfering objects are spatially fixed and at the same time the range of motion of the measuring object is limited, then a sufficient suppression of false echoes can be achieved by a suitable choice of the evaluation time window.

From DE 33 37 690 a method is known in that disturbing object echoes can be suppressed by first detecting all disturbing object echoes and storing them in a memory in a teaching phase in which the measuring object is not in the detection range of the sensor. During measurement operation, the currently detected echoes are compared with the learned echoes. If there is a sufficient match, the echo is classified as a noise echo and accordingly suppressed, while the remaining echoes are assigned to DUTs.

The documents DE 33 37 690 and EP 0 459 336 also describe methods which hide false echoes caused by multiple reflections between the sensor and an object in that the maximum runtime to be evaluated is limited so that echoes occurring outside this runtime are ignored. In the solution shown in EP 0 459 336, the echo amplitude can additionally be evaluated as a criterion for the multi-echo cancellation.

From DE 38 21 577 a method for the suppression of false echoes on the basis of

Plausibility checks known. Since the gradient with which the measurement situation can change is limited due to the finite speed of movement of objects, echoes are only evaluated if their temporal position and amplitude are sufficiently plausible on the basis of previous measurement situations. In this way, especially stochastically occurring interference signals can be safely suppressed.

These methods, described above, each work properly in a variety of applications. However, problems always occur when the echo originating from the fill level can not be unequivocally identified by the method and the useful echo signal jumps due to process conditions. If an echo other than the level echo is inadvertently classified as a useful echo, there is a risk that an incorrect level will be output without this being noticed. Depending on the application, this can lead to overfilling of containers, empty pumps or other events that are sometimes associated with considerable dangers.

Due to the probing probes described above, a false or turbulent measurement of the level of the medium in the tank can occur. In the worst case, there is a so-called echo loss, in which the useful echo signal can no longer be identified or found.

The invention has for its object to provide a reliable and fast method for

Identification of useful echo signals in response signals of the level measuring devices operating according to the transit time measuring principle.

This object of the invention is achieved by the method features mentioned in claim 1.

Advantageous developments of the invention are specified in the dependent claims of the method.

Further details, features and advantages of the subject matter of the invention will become apparent from the following description with the accompanying drawings, in which preferred

Embodiments of the invention are shown. Shown in the figures

Embodiments of the invention are for better clarity and simplicity

Elements that correspond in their structure and / or function with the same

Provided with reference numerals. Show it:

Fig. 1 shows an embodiment of a measuring device for determining the level with a

corresponding envelope,

Fig. 2 shows an inventive embodiment of a measuring device for determining the

Filling level by means of the method according to the invention for identifying the useful echo signal in the response signals,

3 shows an example of intermediate frequency signals of a flat level surface and of a noise element, and

Fig. 4 shows a method according to the invention, for the differentiation of false echoes and

Use echoes of the flat level surface can be used. FIG. 1 shows a measuring device 1 operating according to the transit time measuring method for determining the fill level F of a medium 7. The measuring device 1 is mounted on a container 5 on a nozzle. The measuring device 1 shown is a transmitting / receiving element 6 radiating freely into the process space with a measuring transducer 9. The measuring transducer 9 has at least one transmitting / receiving unit 3, which carries out the generation and the reception of the measuring signals, a control / Evaluation unit 2, which enables the signal processing of the measurement signals and for controlling the measuring device 1, and also a communication unit 4, which controls and regulates the communication via a bus system and the power supply of the measuring device 1 on. In the control / evaluation unit 2, for example, a memory element is integrated, in which the measurement parameters and echo parameters are stored and are stored in the measurement factors and echo factors. The transmitting / receiving element 6 is embodied in this embodiment, for example, as a horn antenna, but can be configured as a transmitting / receiving element 6 any known antenna form, such as rod or planar antenna. In the transmitting / receiving unit 3, a measuring signal is generated for example in the form of a high-frequency transmission signal S and emitted via the transmitting / receiving element 6 in a predetermined emission characteristic in the direction of medium 7. After a travel time t dependent on the distance x traveled, the transmission signals S reflected at the boundary surface 8 of the medium 7 are received again as a reflection signal R by the transmission / reception element 6 and the transmission / reception unit 3. The downstream control / evaluation unit 2 determines from the reflection signals R an echo function 10 which determines the amplitudes of the echo signals of these reflection signals R as a function of the

represented distance x or the corresponding transit time t. By a

Analog / digital conversion and filtering of the analogue echo function or the echo curve 10, a digitized envelope 1 1 is generated.

An envelope 11 depicting the measurement situation in the container 5 is shown as being proportional to the travel distance x of the transmission signal S. For better understanding, reference lines are assigned to the corresponding echo signals in the envelope 1 1, so that the cause-and-effect principle can be detected at a glance. In the beginning of the envelope 1 1 is the Abklingverhalten or the so-called ringing to see, due to multiple reflections or by

Formation can arise in the transmitting / receiving element 6 or the neck. Furthermore, in the initial region of the envelope 1 1, an echo signal 14 is shown, which is caused by the false echo K of the inflow or filling flow of the medium 7. There are also false echoes K caused by the formation of voids in solid applications, but the cavities are not explicitly shown in this figure.

According to the current state of the art, there are different approaches to the exact position of the

Use echo signal in the determined echo function 10 or the digital envelope 11 to determine. From the exact determination of the measurement position of the level F in the envelope 1 1 depends, which measurement accuracy can be achieved with this echo measurement principle under the given measurement conditions.

In Fig. 2, a pulse radar level indicators 1 is shown, which shows the distance by direct

Measurement of the transit time of the microwave pulse as a transmission signal S, radiated from the transmitting element 6 and reflected from the surface 8 of the medium to be measured 7, determined. Pulse radar level gauges 1 operate in the time domain and therefore do not require fast Fourier analysis, which is characteristic of single frequency modulated continuous wave (FMCW) radar. The transit time t of the microwave pulses is in the nanosecond range for a distance of a few meters. For this reason, as already mentioned, a special time-transformation method is needed to accurately measure the very short difference times between two pulses. It requires a slow motion recording of the microwave pulses with a stretched time axis. The pulse radar level gauge 1 uses a uniform periodic transmission signal S having a high pulse repetition frequency. By a sequential sampling method for time stretching the time axis of the received signals or response signal A, the extremely fast and uniform signals into a usable, stretched time signal, the so-called

Intermediate frequency signal ZF to be transformed. This periodic response signal A consists of the actual transmission signal S, at least one useful echo R and at least one false echo or multiple echo K. The intermediate frequency signal ZF hereby resembles an ultrasonic signal. The microwave pulse, of, for example, 6.3 GHz, is transformed by means of the sequential sampling / sampling method into an intermediate frequency ZF of, for example, 76 kHz, and the pulse repetition frequency of, for example, 3.5 MHz is thus reduced to a frequency of 40 Hz.

The echoes of a pulse radar are individually separated and separated in time. This means that the pulse radar is better suited for handling multiple echoes and false echoes, which often occur in process and bulk containers.

The frequencies used in radar level gauges 1 have been reviewed by manufacturers for reasons of licensing, licensing, and availability

Microwave components and selected due to expected technical advantages. The

different transmission frequencies of the antennas 6 of level measuring devices 1 are used tailored to the application and the measurement situation. The achievable accuracy of a pulse radar level gauge 1 depends on the application, on the antenna design, on the qualities of the HF electronics or evaluation electronics and on the one used

Signal processing software.

The inventive approach to determine the level F is shown in FIG. The solution proposed by the invention makes use of the approach of an envelope-free evaluation, by directly using the intermediate frequency signal IF to search for the useful echo R. The use of the intermediate frequency signal ZF to search the useful echo R of

Level F has the advantages that the measurement signal information not only, as with the

Use Envelope 1 1 to restrict amplitude information.

In order to be able to use as much information as possible of the response signal A, this must first be recorded unprocessed. For the subsequent direct evaluation, one chooses the general approach that for each sensor principle the analog response signal A results from the previously transmitted test signal T by means of a model parameter MP.

From this the following equation can be derived:

R = MP ^* T

For the task of level measurement, the model parameter MP can be expressed in a quasi-static environment as a linear time-invariant system.

However, the model parameter MP is dependent on all reflections of the test signal T or transmission signal in the container 5, which are located in the field of view of the sensor 6. The received

Response signals A result from the geometry of the container 5, the level F and

various parasitic effects. Furthermore, the response signals A differ in size and shape. The level surface z. In contrast, approach at the edge, an impeller or general internals as punctiform or arcuate reflectors form as a noise signal K in the ideal case.

This situation is shown in Fig. 3. An even surface supplies as the response signal A

True echo R that is a uniform, sinusoidal pulse packet. In contrast, supplies

False echo K as the response signal A a non-uniform pulse packet.

For example, a nozzle edge 18 represented an annular reflector. Thus, it should be possible to differentiate on the basis of the different response signals A, whether it is the reflection signal or useful echo R reflected by the flat filling surface 8 or

False echo signals or false echoes K of a noise element 12, 13,14,15,16, 18 acts. This principle of differentiation is intended to be proposed here for the selection and identification of the wanted echo R of the filling level F.

The method for determining the wanted echo signal R is shown in FIG. 4 and becomes

For example, as described in the following process steps implemented: Sampling and recording of the response signal A for in comparison to selected test signal S or derived comparison parameter MP.

Comparison of the response signal A or the comparison parameters derived therefrom AMP with a series of comparison signals V or the comparison parameters derived therefrom VMP. The comparison signals V are expected response signals R to the selected test signal S for a useful echo R generated by level surfaces.

Determining the match probabilities W of the recorded one

Response signal A with the test signals based on the signals themselves or derived therefrom comparison parameters AMP.

When a predetermined probability value W is exceeded, the associated fill level F is determined and output as the measured value.

If the set probability value W is not reached for a comparison signal V, the measurement is repeated.

The test signals S can arbitrary amplitude and angle modulated baseband or

Be bandpass signals. Preferably, ramp-shaped frequency-modulated signals, so-called chips, baseband pulses or pulse-shaped modulated, monofrequency high-frequency signals are used.

As a comparison signals V can be achieved by automated parametric analyzes z. B. by EM simulations or by systematic test measurements and possibly their interpolation can be obtained. You can z. B. stored in a large database with the associated test signals and cataloged by applications.

The learning possibility is not limited to the learning phase E but can lead to new insights from test measurements and simulations by means of a continuous, systematic extension and improvement of the database content.

The match probability W indicates with what probability that

Response signal A comes as a pulse packet in the intermediate frequency range ZF from a flat reflector or the surface 8 of the medium 7. In FIG. 2, the comparison algorithm has calculated the probability values of 97% for the pulse packet of the wanted echo R in the above image and 6% for the pulse packet of the false echo K of the stirring blade 15 in the lower image.

A direct comparison of the response signal A of a measurement with a series of comparison signals V can be very complicated and inefficient, depending on the sensor design 6. Long response signals A with many samples would require a lot of storage space for comparison signals V and response signal A as well as computation-intensive comparison algorithms. By modeling the test signals S or comparison signals V, their essential content can be found in a few comparison model parameters VMP be summarized, requiring significantly less storage space in the sensor 6. If the modeling for the response signal A is repeated after each measurement, instead of the interference signals K and reflection signals R, their response model parameters AMP can be compared with the stored comparison model parameters VMP. It can be expected that the reasonably simple geometry of an infinite surface 8 of the filling level F of a medium 7 can be described using simple models, and thus less total computing power is required for modeling and comparison of the model parameters MP, VMP, AMP, as for the comparison longer response signals A and comparison signals V.

The modeling corresponds to an estimate of the answer operator. For this purpose, for example, the modeling methods are used or can be derived accordingly from:

- Parametric method

- Neural Networks

- Subspace procedure eg B MUSIC

- Method of adaptive beamforming

The parametric methods define a specific form of the distribution function of the

Based on probability density and then optimize their parameters.

The subspace algorithm MUSIC (Multiple Signal Classification) uses the ideally occurring orthogonality between the eigenvectors of the noise subspace and the spatial group response, which is associated with sought directions of arrival. The MUSIC algorithm makes no special demands on the shape of the spatial impulse response of the group. In special cases, e.g. a linear antenna array, can be dispensed with the calculation of the full spectrum.

In the subspace methods, first an order estimate, i. an estimate of the number of goals. In the ideal case, the eigenvalues that can be assigned to the reflection signals R can be clearly separated from the eigenvalues that belong to the noise: the

Noise values are all the same, while the eigenvalues of the reflection signal are larger. This fact can be used to estimate the number of received reflection signals. With the help of the subspaces thus separated, the goals can be resolved, despite small differences between false echoes K and useful echoes R. Compared to maximum likelihood methods, the subspace methods require less computing power, which is particularly favorable for use in process automation, whose field device 1 is operated on the basis of the required intrinsic safety with low consumption power and which therefore has little energy available. The echo separation capability also makes it possible to achieve greater robustness to disturbances, for example, through container installations 12, material deposits 13 and / or agitators 14 in the container 5, since the reflections of the false echoes K of the Nutzechos R originating reflections can be separated. The large dynamic range of the radar signals and the ultrasonic signals and the sensor inaccuracies additionally complicate an accurate subspace separation. As a result, further signal processing techniques to increase the robustness of the angular separation, such as calibration and decorrelation, must be investigated.

In the case of the neural networks as a modeling method, the causal relationships between the input signal or test signal and the corresponding, determined response signals A or output variables are stored in the form of at least one transfer function or model parameter.

The method according to the invention also has the advantages that the basic knowledge about the measuring procedure remains in the company and is not disclosed, since the comparison signals supplied in the teaching-in phase always represent only a small part of the database content. Furthermore, updates and thus measures to increase the performance of devices already in operation 1 by exchanging or supplementing the locally stored comparison signals V and modeling methods or their model parameters MP feasible.

The inventive method is not only alone, as explicitly shown in Fig. 1 and 2, implemented in free-radiating microwave measuring devices 1, but an application of the method according to the invention can be carried out in other transit time measurement systems, such as TDR measuring devices or ultrasonic measuring devices. When using ultrasonic measuring devices can be dispensed with the generation of the intermediate frequency signal, since the frequencies of the

Ultrasonic signal is located in a frequency working range of the electronics of the signal processing.

LIST OF REFERENCE NUMBERS

1 field device, measuring device

2 control / evaluation unit

3 transceiver unit

4 communication unit

5 containers

6 transmitting / receiving element, sensor

7 medium

8 boundary layer, surface

9 transducers

10 echo function, echo curve

1 1 Envelope

12 container installations

13 material deposit

14 agitator

15 material inflows

16 Restless surface

7 communication / supply line

18 nozzle edge

Amp amplitude value

S transmission signal, test signal

A response signal

R reflection signal, useful echo

K interference signal, false echo

N noise

V comparison signals

IF intermediate frequency signal

AZF response intermediate frequency signal

VZF comparison intermediate frequency signal

W match probability,

probability value

G predetermined limit

x way, walk

t time, running time

F level

B operating phase

E learning phase MP model parameters, comparison parameters

VMP comparison model parameter

MPS model parameter set

RMP response model parameter

Claims

claims

1. A method for determining and monitoring the level (F) of a medium (7) in a container (5) by means of a field device (1) according to a transit time measurement method,

in a learning phase (E), corresponding application and device-related test signals (S) and response signals (R) to be expected from a fill level surface are determined,

- where from the application and device related test signals (S) and from the

Fill level surface expected response signals (R) in the learning phase (E) application and device-related comparison signals (V) are determined

wherein test signals (S) are emitted in the direction of the medium (7) in an operating phase (B) and application-related and device-related response signals (A) are received,

- In an operating phase (B) by means of a comparison algorithm, the application and device-related comparison signals (V) with the application and device-related

Response (A) and a value for a

Match probability (w) is determined

- In the operating phase (B) when exceeding the determined value of the

Match probability (w) over a predetermined limit (G) of the level (F) is determined and output as a measured value and / or falls below the predetermined limit (G) for re-determination of an application and device-related

Response signal (A) a new test signal (S) is sent out.

2. The method according to claim 1,

wherein in the learning phase (E) from the application and device-related comparison signals (V) by means of a modeling method corresponding comparison model parameters (VMP) are derived and stored in the level gauge (1) as a model parameter set (MPS).

3. The method according to claim 2,

wherein in the operating phase (E) from the application and device-related response signals (A) by means of the modeling method corresponding actual response model parameters (RMP) are derived and the match probability (W) by means of a

Comparison algorithm that uses the current response model parameters (RMP) with those in the

Model parameter set (MPS) stored comparison model parameter (VMP) compares, is determined.

4. The method according to claim 1,

wherein the response signals and / or comparison signals (V) are converted into low frequency response intermediate frequency signals (AZF) and / or comparison intermediate frequency signals (VZF) by means of sequential sampling and wherein these intermediate frequency signals (AZF.VZF) are digitized by means of an analogue to digital conversion become.

5. The method according to claim 1,

where as modeling method parametric analyzes, for. B. by EM simulations, or systematic test measurements in the learning phase (E) are performed.

6. The method according to at least one of claims 1 or 5,

where as modeling method parametric analyzes, for. B. by EM simulations, or systematic test measurements in the learning phase (E) steadily and systematically in the

Operating phase (B) are performed.

7. The method according to at least one of claims 1, 5 or 6,

wherein parametric methods, neural networks, subspace methods and / or adaptive beamforming methods are used to derive the model parameters by means of modeling methods.

8. The method according to at least one of claims 1, 2, 5 or 6,

wherein the determined comparison signals (V) in a database application-specific and / or device-specific cataloged and associated with the associated test signals (S) are stored.

9. The method according to at least one of claims 1 and 8,

wherein amplitude-and / or angle-modulated baseband signals, ramp-shaped, frequency-modulated signals, baseband pulses or pulse-shaped modulated monofrequency high-frequency signals are used as test signals (S).