BR102021006285A2 - DIGITAL TRANSIT TIME FLOW METERS USING REVERSE TIME ACOUSTIC - Google Patents

Abstract

A microcontroller and a method for determining a flow velocity with an electronic processing unit of an ultrasonic travel time flowmeter with arbitrary waveform signals. The electronic processing unit comprises receiver and transmitter terminals, a signal processing unit and a signal generation unit which is configured to generate an oscillating electrical output signal having a time-dependent amplitude, wherein the time-dependent amplitude varies according to the stored signal parameters.

Description

DIGITAL TRANSIT TIME FLOW METERS USING REVERSE TIME ACOUSTICS

[001] The current order refers to flowmeters, and in particular to ultrasound transit time flowmeters.

[002] Various types of flowmeters are currently in use to measure the volume flow of a fluid, such as a liquid or a gas, through a pipe. Ultrasonic flowmeters are either Doppler flowmeters, which make use of the acoustic Doppler effect, or transit time flowmeters, sometimes also called transmission flowmeters, which use a propagation time difference caused by motion. relative to source and medium. Travel time is also known as flight time or transit time.

[003] An ultrasonic travel time flowmeter evaluates the difference in the propagation time of ultrasonic pulses that propagate in and against the flow direction. Ultrasonic flowmeters are supplied as in-line flowmeters, also known as intrusive or wet flowmeters, or as clamp-on flowmeters, also known as non-intrusive flowmeters. Other forms of flowmeters include Venturi channels, overflow sills, radar flowmeters, Coriolis flowmeters, differential pressure flowmeters, inductive magnetic flowmeters, and other types of flowmeters.

[004] When there are irregular flow profiles or open channels, more than one propagation path may be required to determine the average flow velocity. Among others, multipath procedures are described in hydrometric standards such as IEC 41 or EN ISO 6416. As an additional application, ultrasonic flowmeters are also used to measure flow profiles, for example with an acoustic Doppler current profiler (ADCP ). The ADCP is also suitable for measuring water velocity and flow in rivers and open water.

[005] Zhaohong Zhang, "Matched-Filter Ultrasonic Sensing: Theory and Implementation", White Paper SLAA814-December 2017, discloses a theory of operation of a compatible filter-based ultrasonic sensing technology and a single-chip implementation platform using the microcontroller (TI) MSP430FR6047.

[006] It is an object of the present invention to provide an improved transit time flowmeter and a corresponding computer-implemented method for measuring an average flow velocity or a flow profile of a fluid in general and in particular for liquids such as water and/or for gases.

[007] In a flow measurement device according to the present specification, sound transducers in the form of piezoelectric elements, also known as piezoelectric transducers, are used to generate and receive a measurement signal.

[008] Alternative sound transmitters comprise lasers that excite a metal membrane or other light-absorbing surface to vibrations, or coil-driven loudspeakers. According to other embodiments, the flowmeter produces pressure waves in other ways, such as through a MEMS device, using a piezoelectric membrane, and so on. The receiver side can also be represented by other means that are different from piezoelectric transducers but detect ultrasonic waves

[009] Although the term "piezoelectric transducer" is used frequently in the present description, it also means other sound wave transducers that produce or detect ultrasonic waves.

[010] A measurement signal according to the present specification can be modeled by a matched filter. When the word "signal" is used with reference to a signal manipulation step, it may, in particular, refer to a representation of a signal in computer memory.

[011] In particular, a signal representation can be defined by pairs of digitized amplitude values and associated discrete times.

[012] An ultrasonic flow meter according to the present specification can provide desired properties of a focusing property using a signal in an arbitrary way, for an ultrasonic flow meter to obtain a signal with the desired properties in a receiving transducer or through calculations after receiving ADC

[013] By way of example, the frequency of sound waves that are used in a flow meter according to the invention may be between 20 kHz and 2 MHz, which corresponds to an oscillation period of 0.5 microseconds (μs ), but can even be as high as 800 MHz.

[014] In one aspect, the present specification discloses a microcontroller for determining a flow velocity with an ultrasonic travel time flowmeter using a time-inverted digitized signal. In particular, the flowmeter can be "fully digital" in the sense that the signal evaluation is done by digital signal processing and the electrical signals applied to the ultrasonic transducers are step signals, which can also be considered digital signals. In other words, electrical signals have a sequence of discrete voltage levels that are constant over predetermined sampling times. Thus, there is no need for a DAC with an analog part and the DAC can be replaced with a pulse generator.

[015] The microcontroller is capable of generating arbitrary pulse signals. In particular, the microcontroller can be configured to generate pulse trains of arbitrary frequency and length.

[016] The microcontroller according to the present specification may comprise multiple buffered output terminals, for example, four or eight output terminals, instead of just one. For this purpose, the microcontroller may comprise several buffers, for example four or eight buffers, in a parallel structure, which allows a time-multiplexed output of all signals to all respective channels or buffers.

[017] In other words, a transmitter path or a signal generation path on a chip or FGPA, or a path connected to such a chip or FPGA is operative to generate arbitrary pulse trains with arbitrarily chosen pulse length. Here "arbitrary" means arbitrary within a predetermined range and accuracy, e.g. within a predetermined minimum and maximum frequency range and within a predetermined accuracy

[018] The microcontroller comprises a first receiver terminal for connecting a first ultrasonic transducer, a first transmitter terminal for connecting the first ultrasonic transducer, a second receiver terminal for connecting a second ultrasonic transducer, a second transmitter terminal for connecting the second ultrasonic transducer. The receiver terminals and the transmitter terminals are also called "connectors".

[019] In one embodiment, the first receiver terminal and the first transmitter terminal and the second receiver terminal and the second transmitter terminal coincide. This can be especially useful when you don't have overlapping sending and receiving.

[020] The first receiver terminal and the second receiver terminal are connected to a signal processing unit or signal processing means, which is provided for evaluating the signals received from the transducers. The signal processing unit comprises an analogue-to-digital converter (ADC) and an evaluation circuit. Unlike a simple timer circuit, the ADC can provide several options for calculating the shape of a signal.

[021] A predicted signal shape can be used to reduce the sampling rate and/or amplitude resolution of the ADC. In particular, power saving features may make use of a signal shape in accordance with the present specification. The sampling rate and/or the accuracy range of the ADC amplitude scanning can also be non-uniform and time dependent to further reduce power consumption and this time dependence can depend on a predicted signal shape. Furthermore, the ADC can be turned on only after a specified time after sending the measurement signal and turned off again after the response signal has been received.

[022] The first transmitter terminal and the second transmitter terminal are connected to a signal generating unit. The signal generating unit comprises a memory for storing signal parameters and a buffer.

[023] The signal generating unit and the signal processing unit are configured to send the oscillating electrical output signal to the first transmitting terminal and to receive a response signal at the second receiving terminal and to derive a flow rate from a fluid of at least the received reply signal

[024] In one embodiment, the signal generating unit buffer is connected to a pulse width modulation (PWM) unit, which is connected to a low-pass filter. Pulse-width modulation can provide a simple and robust way to generate an analog electrical signal from a digital signal, which is then applied to an ultrasonic transducer such as a piezoelement, piezomembrane, or speaker membrane. , a microelectromechanical element, or another type of transducer. By using power electronics, the output signal of a pulse-width modulation can be made large enough that, in many cases, no additional signal amplification is needed. PWM can provide sufficient quality of the resulting analog signal.

[025] The microcontroller PWM can also be used with a low or weak pass filtering after the buffer output or the PWM to generate an output signal with a step shape or a step shape. To this end, the low-pass filter can be provided as an adjustable or switchable low-pass filter.

[026] PWM signals store amplitude information in the signal's timing information. The granularity of the time resolution has an impact on the resolution of the PWM signal. For this reason, the PWM signal generator can be designed to deliver a higher frequency than a comparable ADC, for example 10 times higher. The amplitude of the PWM signal generator can be the same or similar to a pulse generator.

[027] In another embodiment, the microcontroller signal generating unit comprises an array of field-programmable gates (FPGA), which is connected to a digital input / output connector. The signal generating unit is operative to derive the oscillating signal from an FPGA output signal, wherein the FPGA output signal is modified by providing programming instructions through the digital input/output connector.

[028] In particular, the FPGA may contain a logic or circuit arrangement for deriving a desired output signal from the given signal parameters. Among others, the signal parameters can be read from a memory, precomputed or computed in real time. In one embodiment, calculating the signal parameters from which the output signal is derived depends on a received signal.

[029] In another embodiment, the signal generating unit is configured to generate a signal train, the signal train comprising oscillating signal portions that are separated by predetermined periods of silence in which an output voltage is constant, in particular output voltage may be zero during quiet period. In one embodiment, the oscillating signal portions are repeated at regular intervals. In another embodiment, the duration of the oscillating signal portions and the time intervals between them is determined by a calculation. In particular, the calculation of the time intervals may comprise a random component or jitter within a predetermined magnitude range, wherein the oscillating signal portions have a time-varying distance from each other.

[030] According to another embodiment, the signal generation of the oscillating signal comprises applying a pulse-width modulation and applying a low-pass filter to the pulse-width modulation output signal.

[031] In yet another embodiment, the signal generation of the oscillating signal comprises generating a signal train, the signal train comprising oscillating signal portions that are separated by predetermined silent periods in which an output voltage is constant, in that constant voltage level can be zero

[032] During a measurement of a flow velocity, the microcontroller applies the second measurement signal to one of the transducers, receives a response signal for the second measurement signal on the other of the transducers, and processes the response signal to derive a velocity flow.

[033] According to one aspect, the current specification discloses a computer-implemented method for determining a flow velocity of a fluid in a conduit or fluid channel, in particular in a pipe or tube, using an ultrasonic flowmeter of transmission time. In a preferred embodiment, "computer-implemented" refers to an implementation on small-scale electronic components, such as microprocessors, ASICs, FPGAs, and the like, that can be used in portable or compact stationary digital signal processing devices that they are usually smaller in size than workstations or mainframe computers and can be placed where needed along a fluid pipe.

[034] Below, the terms “channel”, “conduit”, “passage”, etc. are used as synonyms. The subject of the order can be applied to all types of fluid conduits, regardless of their respective shape and regardless of whether they are open or closed. The subject of the order can also apply to all types of fluids or gases, whether they are gases or liquids, or a mixture of both.

[035] Throughout the application, the term “computer” is frequently used. Although a computer includes devices such as a laptop or desktop computer, signal transmission and reception can also be done by microcontrollers, application-specific integrated circuits (ASICs), FPGAs, etc.

[036] In addition, a connecting line between the transducers can be offset relative to a center of the fluid conduit in order to obtain a flow velocity in a predetermined layer, and there can be more than one pair of transducers. Furthermore, the measurement signal can be provided by more than one transducer and/or the response signal to the measurement signal can be measured by more than one transducer.

[037] In addition, the current specification discloses a flow measurement device with a first piezoelectric transducer that is connected to the first connector, and with a second ultrasonic transducer, such as a piezoelectric transducer, that is connected to the second connector. In particular, ultrasonic transducers, such as piezoelectric transducers, can be provided with attachment regions, such as an attachment mechanism for attaching them to a pipe.

[038] In addition, the current specification discloses a flow measuring device with a tube portion. The first ultrasonic transducer such as the piezoelectric transducer is mounted on the pipe portion at a first location and the second ultrasonic transducer such as the piezoelectric transducer is mounted on the pipe portion at a second location. In particular, the transducers can be attached to the pipe part. Providing the device with a tube portion can provide benefits when the device is pre-calibrated with respect to the tube portion.

[039] The device can be compact and portable. A portable device according to the present specification that is equipped with surface mountable transducers such as clamp transducers can be used to check a pipe in any accessible location. In general, the device can be stationary or portable. Preferably, the device is compact enough to be placed in a desired location and sufficiently protected from environmental conditions such as moisture, heat and corrosive substances.

[040] In particular, the application-specific electronics may be provided by electronics comprising the aforementioned computer-readable memory. According to other embodiments, the application-specific electronics are provided by components with a hard-wired or configurable circuit, such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA).

[041] A learning procedure to learn a signal can be performed as follows. Here, the signal that is generated through the learning procedure may refer to a digital filter or another discrete time sequence. The signal can be used to generate a sound signal or it can be used for computing purposes to process a digitized version of a received sound signal.

[042] Learning is done by capturing the response to the channel step under controlled condition, for example, at zero flow. This can be done using one or more pulses as a broadcast signal, which is transmitted through the channel. Here, "channel" refers to a communication channel, which comprises the fluid.

[043] The step response is then processed as follows:

[044] In the case of the signal generated by PWM, the step response is time-reversed and digitized so that the PWM transmission signal corresponds to the time-reversed step response.

[045] In the case of a pulse train input, the channel step response is stored in a digital receive filter after analog to digital conversion.

[046] A measurement phase can then be performed as follows:

[047] In case a PWM signal is used, the time reversed channel impulse response derived from the learning process is used as a broadcast signal, providing the time reversed digital signal as input to a PWM device and connecting an output from the PWM-device to an ultrasonic transducer.

[048] In case a pulse train is used, the same signal used during the learning phase can be used as a transmission signal or measurement signal in the measurement phase. For example, a step signal or a single pulse or pulse train can be used as a transmission signal or as a measurement signal. In one example, the pulse train has 10 to 30 pulses. The measurement signal can alternatively be longer or shorter compared to the learning phase.

[049] In case a PWM signal is used as a broadcast signal, the broadcast signal bends with the impulse response of the channel and a distinct peak is received at the input of the ADC. In other words, in the approximation of an LTI system (Linear Time Invariant-system), the received response signal is a convolution of the transmission signal with the previously determined impulse response of the channel.

[050] The received signal is digitized and the peak of the signal is, for example, used to derive travel time information. This can be done through a high resolution sampling process or, alternatively, by interpolation of the received signal. The peak detection of the mentioned signal belongs to the PWM, the peaks only appear right after the ADC, if the correlation process takes place in the channel.

[051] The peak can also be calculated after the ADC if the stored inverse filter is correlated with the receive filter.

[052] In case a pulse train is used as a transmission signal, we explain that signal processing can be performed by a DTRAF or DTRAC method according to the present invention. The DTRAC method can provide improved energy efficiency compared to the DTRAF method. One reason for the improved efficiency is that the DTRAF method filters all incoming signals with FIR filters. The DTRAC method, on the other hand, does not do this power-consuming operation, instead it correlates the digitized measurement signals directly.

[053] In the case of DTRAF, the received signal is filtered with the inverse of the previously stored channel response, usually in a finite impulse response filter, after the learning phase. As a result of this filtering operation, a distinct peak is obtained. This operation can be done for both upstream and downstream signals. The resulting signals are then correlated and interpolated.

[054] Second, in the DTRAC case, the ADC signals from the measurement process are correlated with the training signals and then interpolated. This can again be done for upstream and downstream signals, so that a travel time difference ΔT between the upstream and downstream signals can be calculated.

[055] The subject of the present specification is now explained in more detail with respect to several figures, in which:

[056] Fig. 1 shows a first flowmeter arrangement with two piezoelectric elements;

[057] Fig. 2 shows the flowmeter arrangement of Fig. 1, one direct signal and two scattered signals;

[058] Fig. 3 shows the flowmeter arrangement of Fig. 1 when viewed in the flow direction;

[059] Fig. 4 shows a second flowmeter arrangement with four piezoelectric elements and four direct signals;

[060] Fig. 5 shows the flowmeter arrangement of Fig. 4 when viewed in the flow direction

[061] Fig. 6 shows a many-to-one sensor arrangement for a flow measurement in accordance with the present invention;

[062] Fig. 7 shows a one-to-many sensor arrangement for a flow measurement in accordance with the present invention;

[063] Fig. 8 shows a one-to-one sensor arrangement for a flow measurement in a layer according to the present invention;

[064] Fig. 9 shows an arrangement of multiple sensors for measuring flow in multiple layers in accordance with the present invention;

[065] Fig. 10 shows a device for measuring a flow rate according to the present invention;

[066] Fig. 11 illustrates an iterative procedure for deriving a waveform for use in the flowmeter of Fig. 10;

[067] Fig. 12 illustrates another procedure for deriving a waveform for use in the flowmeter of Fig. 10;

[068] Fig. 13 illustrates another procedure for deriving a waveform for use in the flowmeter of Fig. 10;

[069] Fig. 14 shows a comparison between a PWM signal from a pulse width modulator, which approximates a time-inverted TRA signal, and the time-inverted signal;

[070] Fig. 15 shows a comparison between a response signal to the PWM signal of Fig. 15 and a response signal to the time-inverted signal of Fig. 15;

[071] Fig. 16 shows the simulation results for a PWM-based correlated TRA;

[072] Fig. 17 shows simulation results for signals generated with the digital TRA correlation method shown in Fig. 21;

[073] Fig. 18 shows simulation results for signals generated with the digital TRA filtering method shown in Fig. 19;

[074] Fig. 19 shows a visualization of the DTRAF method as a block diagram;

[075] Fig. 20 shows an illustration of a general peak interpolation for determining a maximum of a correlation function;

[076] Fig. 21 shows a visualization of the DTRAC method as a block diagram, and

[077] Fig. 22 shows another embodiment of a proposed flow rate measuring device.

[078] In the following description, details are provided to describe the embodiments of the present invention. It should be apparent to one skilled in the art, however, that the embodiments can be practiced without such details.

[079] Some parts of the embodiments, which are shown in the Figures, have similar parts. Like parts have the same names or numbers as like parts with a prime symbol or an alphabetic symbol. The description of such similar parts also applies by reference to other similar parts, where appropriate, thereby reducing repetition of text without limiting disclosure.

[080] Fig. 1 shows a first flowmeter arrangement (10). In the flowmeter arrangement, a first piezoelectric element (11) is placed on an outer wall of a tube (12), which is also referred to as a tube (12). A second piezoelectric element (13) is placed on an opposite side of the tube (12) such that a straight line between the piezoelectric element (11) and the downstream piezoelectric element (13) is oriented at an angle β with the direction ( 14) of the mean flow, which is at the same time also the direction of the tube (12) axis of symmetry. The angle β is chosen to be approximately 45 degrees in the example of Fig. 1, but it can also be steeper, like 60 degrees, or shallower, like 30 degrees.

[081] A piezoelectric element, such as the piezoelectric elements (11), (13) of Fig. 1 can, in general, be operated as an acoustic transmitter and as an acoustic sensor. An acoustic transmitter and an acoustic sensor can be provided by the same piezoelectric element or by different regions of the same piezoelectric element. In this case, a piezoelectric element or transducer is also referred to as a piezoelectric transmitter when it is operated as a sound transmitter or source and is also referred to as an acoustic sensor or receiver when it is operated as an acoustic sensor.

[082] When a flow direction is as shown in Fig. 1, the first piezoelectric element (11) is also referred to as the "upstream" piezoelectric element and the second piezoelectric element (13) is also referred to as the "downstream" piezoelectric element. A flow meter according to the present specification works for both flow directions in essentially the same way and the flow direction of Fig. 1 is provided as an example only.

[083] Fig. 1 shows a flow of electrical signals from Fig. 1 for a configuration in which the upstream piezoelectric element (11) is operated as a piezoelectric transducer and the downstream piezoelectric element (13) is operated as an acoustic sensor. For the sake of clarity, the application works both upstream and downstream, i.e. the positions of the piezoelectric elements can be interchanged.

[084] A first computing unit (15) is connected to the upstream piezoelectric element (11) and a second computing unit 16 is connected to the downstream piezoelectric element (13). The first computing unit (15) comprises a first digital signal processor, a first digital buffer and a first digital to analog converter (ADC). Likewise, the second computing unit (16) comprises a second digital signal processor, a second digital buffer and a second digital to analog converter (ADC). The first computing unit (15) is connected to the second computing unit (16).

[085] The arrangement with two computing units (15), (16) shown in Fig. 1 is provided as an example only. Other embodiments may have different numbers and arrangements of computation units. For example, there may be only one central computing unit, or there may be two ADC or buffer units and one central computing unit, or there may be two small-scale computing units on the transducers and a larger central computing unit.

[086] A computing unit or computing units can be provided by microcontrollers or application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs), for example.

[087] A method for performing a measurement process according to the present invention comprises the following steps:

[088] A predetermined digital measurement signal is generated by synthesizing an electrical signal with the digital signal processor of the first computing unit (15). The electrical signal is sent from the first computing unit (15) to the piezoelectric transducer (11) along the signal path (17). The piezoelectric transducer (11) generates a corresponding ultrasonic test signal. Units (15) and (16) can also be supplied in a single unit.

[089] The measurement signal is provided as an arbitrary waveform. For example, the arbitrary waveform can be provided by a pulse width modulated wobble with a base frequency in the MHz range, such as a 1 MHz wobble. The arbitrary waveform can also represent a unit pulse

[090] The ultrasonic test signal travels through the fluid (eg liquid) in the tube (12) to the piezoelectric sensor (13). In Fig. 1, a direct signal path of the ultrasound signal is indicated by an arrow (18). Likewise, a direct signal path from the ultrasonic signal in the reverse direction is indicated by an arrow (19). A response signal is picked up by the piezoelectric sensor (13), sent to the second computing unit (16) along the signal path (20) and digitized by the second computing unit (16).

[091] In an additional step, a digital measurement signal is derived from the digitized response signal through signal processing. According to other embodiments, deriving the measurement signal comprises further processing steps.

[092] In a flow meter according to an embodiment of the present specification, the same measurement signal is used for both directions (18), (19), the downstream and upstream direction, providing a simple and efficient arrangement . According to other embodiments, different measurement signals are used for both directions. In particular, the measurement signal can be applied to the original test signal receiver. Such arrangements can provide benefits for asymmetric conditions and tube shapes.

[093] The operation of the flow meter is now explained in more detail in relation to Fig 1.

[094] The ultrasound measurement signal travels through the liquid in the tube (12) to the piezoelectric sensor (13). A response signal is picked up by the piezoelectric sensor (13), sent to the second computing unit (16) along the signal path (20) and digitized by the second computing unit (16)

em que L é o comprimento do caminho direto entre os elementos piezelétricos (11), (13), β é o ângulo de inclinação do caminho direto entre os elementos piezelétricos (11), (13) e a direção do fluxo médio, e c é a velocidade do som em o líquido sob as condições de pressão e temperatura dadas.[095] A similar process is performed for a signal traveling in the reverse direction (19), that is, the aforementioned measurement signal is applied to the downstream piezoelectric element (13) and a response signal is measured by the upstream piezoelectric element (11) to obtain a TOFup upstream flight time in the reverse direction (19). The first computing unit (15) determines a flow rate, for example according to the formula

where L is the length of the direct path between the piezoelectric elements (11), (13), β is the angle of inclination of the direct path between the piezoelectric elements (11), (13) and the mean flow direction, and c is the speed of sound in the liquid under the given pressure and temperature conditions.

[096] The upstream flight time is given by

[097] Using this formula, it is not necessary to determine the temperature or pressure, which in turn determine the density of the fluid and the speed of sound, or measure the speed of sound or the density of the fluid directly. In contrast, the first order error does not cancel for only one measurement direction.

[098] Instead of using a factor of L times cos (beta), a flow-dependent value can be derived from calibration measurements with known flow velocity. These calibration values take into account other effects such as flow profiles and contributions from sound waves that were scattered and did not travel along a straight line.

[099] According to yet another embodiment, the measurement signal, which must be provided to a piezoelectric transmission element, is synthesized using an arbitrary signal form.

em que t e τ são variáveis de tempo, SigFlow representa um sinal a montante ou a jusante sob condições de medição, quando há um fluxo de fluido através do tubo, e em que SigNoFlow representa um sinal sob condições de calibração em fluxo zero. Os limites de soma infinita representam uma janela de tempo suficientemente grande [-T1, + T2] do tempo T1 ao tempo T2. Em palavras mais gerais, -T1 e + T2 não precisam ser iguais e, por razões práticas, isso pode ser vantajoso para o medidor de vazão.[0100] According to another embodiment of the present specification, the time of flight of a signal is evaluated using a cross-correlation technique. In particular, the respective time offsets can be evaluated by cross-correlating the signal received downstream or upstream with the signal received at zero flow rate according to the formula:

where t and τ are time variables, SigFlow represents an upstream or downstream signal under measurement conditions when there is fluid flow through the pipe, and where SigNoFlow represents a signal under calibration conditions at zero flow. Infinite sum bounds represent a sufficiently large time window [-T1, + T2] from time T1 to time T2. In more general words, -T1 and +T2 need not be the same, and for practical reasons this can be advantageous for the flowmeter.

[0101] The TOFup - TOFdown time offset is then obtained by comparing the upstream correlation function maximum time with the downstream correlation function maximum time. The correlation function envelope can be used to more accurately determine the location of the maximum.

[0102] In another embodiment, a separate evaluation unit is provided between the first computing unit (15) and the second computing unit (16), which performs the calculation of signal arrival times and flow velocity.

[0103] In general, the measured signal from the acoustic sensor results from a superposition of scattered signals and a direct signal. Scattered signals are reflected from the inner and outer walls of the tube one or more times, including additional scattering processes within the tube wall. This is shown, by way of example, in Fig. two.

onde t é uma variável de tempo, A é um fator de atenuação, s é o sinal emitido que é deslocado no tempo pelo tempo desconhecido de chegada TOA e n (t) é um termo de ruído. O tempo de chegada TOA é então derivado correlacionando o sinal recebido x (t) com o sinal de medição deslocado no tempo s (t) de acordo com

em que -inf, inf são limites de integração menos infinito e mais infinito e a convolução x (τ) s (t- τ) é uma função que é integrada em relação à variável de tempo τ[0104] The arrival time can also be determined using a blended filter technique. According to a simple model, which is based on the assumptions of direct signal propagation and mirror-like reflections on the conduit walls, the received response signal can be modeled as

where t is a time variable, A is an attenuation factor, s is the emitted signal that is time-shifted by the unknown arrival time TOA, and n(t) is a noise term. The arrival time TOA is then derived by correlating the received signal x(t) with the time-shifted measurement signal s(t) according to

where -inf, inf are limits of integration minus infinity and plus infinity and the convolution x (τ) s (t- τ) is a function that is integrated with respect to the time variable τ

[0105] Arrival time is the time value or function argument for which the correlation becomes a maximum

[0106] The aforementioned correlation can also be expressed in terms of a convolution with a "matched filter" h(t) that takes the form

[0107] where a is a normalization factor and s (t) is the measurement signal. This procedure can be generalized to the case of multiple receivers as follows. A set of phase shifts is determined which maximizes the sum of the response signal amplitudes of the individual receiving transducers. An arrival time is determined by applying a filter adapted to the sum function of the individual received response signals shifted by the predetermined set of phase shifts.

[0108] The transducer configuration of Fig. 1 is a direct line. Other arrangements, which make use of reflections on the opposite side of the tube, are also possible, such as "V" and "W" configurations. The V and W configuration work based on reflections from the tube wall, which induce more scattering than the straight-line configuration. The application subject will benefit from these settings as long as the paths are understood correctly.

[0109] In a V configuration, the two transducers are mounted on the same side of the pipe. To register a 45 degree reflection, they are placed about a pipe diameter apart in the direction of flow. The W configuration makes use of three reflections. Similar to the V configuration, the two transducers are mounted on the same side of the pipe. To record a signal after two 45 degree reflections, they are placed two pipe diameters apart in the direction of flow.

[0110] Fig. 2 shows, by way of example, a first acoustic signal "1", which travels directly from the piezoelectric element (11) to the piezoelectric element (13), and a second acoustic signal "2", which is scattered twice on the periphery of the tube (12).

[0111] For the sake of simplicity, scattering events are shown as reflections in Fig. 2 to 5, but the actual spreading process can be more complicated. In particular, the most relevant scattering normally occurs on the tube wall or on the material that is mounted in front of the piezoelectric transducers. Fig. 3 shows a view of Fig. 2 in the flow direction in the A-A viewing direction.

[0112] Figs. 4 and 5 show a second sensor arrangement in which another piezoelectric element (22) is positioned at a 45 degree angle to the piezoelectric element (11) and another piezoelectric element (23) is positioned at a 45 degree angle with respect to the piezoelectric element (13).

[0113] In addition, Figs. 4 and 5 show direct or straight line acoustic signal paths for a situation where the piezoelectric elements (11), (22) are operated as piezoelectric transducers and the piezoelectric elements (13), (23) are operated as acoustic sensors. The piezoelectric element (23), which is at the rear of the tube (12) in the view of Fig. 4, is shown by a dashed line in Fig. 4.

[0114] Figs. 6 to 9 show, by way of example, different arrangements of clamp-on piezoelectric transducers for which a flow measurement according to the present specification can be used. By supplying multiple sending transducers with signals based on arbitrary waveforms, it is possible to obtain an improved signal at a receiving transducer, for example, signal improved beamshaping properties. By providing multiple receiving transducers, a received measurement signal can be evaluated more efficiently and/or greater design freedom for arbitrarily shaped measurement signals can be achieved.

[0115] Figs. 6 to 9 are aligned so that the force of gravity on a liquid in tube (12) points downwards. However, arrays that are rotated with respect to the arrays in Fig. 6 to 9 can also be used. The viewing direction of Figs. 6 to 9 is along the longitudinal axis of the tube (12). An upstream or downstream position of a transducer is not indicated in Figs. 6 to 9.

[0116] In the arrangement of Fig. 6, an array of five piezoelectric elements (31-35) is provided at a first location and another piezoelectric element (36) is placed upstream or downstream of the first location. The array of piezoelectric elements (31-35) can be used to obtain a predetermined wavefront and to achieve improved focusing of an acoustic wave in a predetermined direction when the array of five elements (31-35) is used as a transmitter and the other element (36) is used as a receiver.

[0117] In the arrangement of Fig. 7, a single piezoelectric element (37) is provided at a first location and an array of five piezoelectric elements (38-42) is placed upstream or downstream of the first location. The array of piezoelectric elements (38-42) can be used to obtain an improved wavefront recording of the response signal. The improved recording can then be used to obtain an improved flow measurement signal, which is then applied to the single piezoelectric element (37).

[0118] Fig. 8 shows an arrangement of two piezoelectric elements (43), (44) in which one element is placed downstream with respect to the other. The distance d from the connecting line between the piezoelectric elements (43), (44) to the axis of symmetry of the tube (12) is about half the radius of the tube (12), so that a layer of flux at a distance d from the central axis of the tube (12) can be measured.

[0119] Especially for clamp-on transducers, such as the piezoelectric elements (43), (44) shown in Fig. 8, the flow measurement according to the present specification provides an enhanced signal at the receiver piezoelectric element (44), (43) through beamforming.

[0120] Fig. 9 shows an arrangement of eight piezoelectric elements (45 – 52), which are spaced 45 degrees apart. Various arrangements are possible with respect to upstream-downstream collocations

[0121] In an arrangement, sensor locations alternate between upstream and downstream along the perimeter, for example, (45, 47, 49, 51) upstream and (46, 48, 50, 52) downstream.

[0122] In another arrangement, the first four consecutive elements, such as (45-48), along the perimeter are placed upstream or downstream in relation to the other four elements, such as (49-52). In another arrangement with 16 piezoelectric elements, all piezoelectric elements (45 - 52) of Fig. 9 are placed in a plane and the arrangement of Fig. 9 is repeated in the upstream or downstream direction.

[0123] In the arrangements of Figs. 6-9, a receiving transducer is displaced with respect to a sending transducer with respect to a longitudinal direction of a conduit or a direction of flow. Specifically, the transducer (36) of Fig. 6 is offset with respect to the transducers (31 to 35), the transducer (37) of Fig. 7 is offset with respect to the transducers (38 to 42), the transducer (44) of Fig. 8 is offset with respect to the transducer (43) and in Fig. 9 the opposite transducers are displaced relative to each other in the longitudinal direction of the duct. For example, transducer (51) is offset relative to transducer (47) and transducer (52) is offset relative to transducer (46).

[0124] Fig. 10 shows, by way of example, a flow measuring device (60) for measuring a flow in the arrangement of Fig. 1 or other arrangements according to the present invention. In the arrangement of Fig. 1, the flow measuring device (60) is provided by the first and second computing units (15) and (16) (not shown in Fig. 10).

[0125] A first connector (61) of the flow measuring device (60) is connected to a first piezoelectric transducer (11) in a fluid conduit (12) and a second connector (62) of the flow measuring device is connected to a second piezoelectric transducer (13) in the fluid conduit (12).

[0126] Within the flow measurement device (60), the first connector (61) is connected to an analog-to-digital converter (64) through a multiplexer (63) and a first amplifier (74). The second connector (62) is connected to a digital buffer (67) through a second amplifier (75) and a demultiplexer (66).

[0127] The buffer (67) is connected to a waveform generator (69), which is connected to a waveform database (70). The waveform database (70) is connected to the ADC (64) through a corresponding module (68), wherein the corresponding module 68 is provided to match parameters with specific circumstances of the conduit (12) under test. The ADC (64) is further connected to a speed computing unit (71), which is connected to a result memory (72).

[0128] During a signal generation phase, the waveform generator (69) retrieves waveform parameters from the waveform database (70), derives an electrical signal from the retrieved waveform parameters, and sends the signal to the digital buffer (67).

[0129] More specifically, the flow measurement device (60) can be provided with an electronic processing unit (53) in the form of a microcontroller or an FPGA comprising - among others - several ADCs (64) with a high resolution , a digital correlator (not shown in Fig. 10) and a module that supports signal envelope detection. In the transmission direction, the flow measuring device (60) generates digital signals with variable frequency. In the case of a square wave signal, the duty cycle of the signal can be changed by modulating the pulse width of the square wave signal, which is also known as PWM.

[0130] The ADC (64), the corresponding module (68), the waveform database (70), the waveform generator (69), the buffer (67) are provided, and in the electronic processing (53), and a microcontroller or a microprocessor. Among others, the electronic processing unit (53) has a connector pin (54) for connecting the first amplifier (74), a connector pin (55) for connecting the second amplifier (75), a connector pin (not shown) for connect a power battery (not shown), and a connector pin (not shown) to connect a ground potential (not shown).

[0131] Other components that are not shown in Fig. 10, as an electrical energy converter, can be provided between the supply battery and a connector pin.

[0132] In an alternative embodiment, it is possible to measure a "current fingerprint" of the electronic processing unit (53), when executing the proposed methods. It has been found that a portion of the methods calculation is not in the peek (which would be a distinct current fingerprint), but rather in flat regions after the peek.

[0133] The flow measuring device (60) can transmit signals with a predetermined frequency and amplitude. These signals are output with the digital buffer (67). Alternatively, the output can be provided by a component that converts the predetermined output signal into a PWM signal.

[0134] The device in Fig. 10 is shown by way of illustration. A device for performing a flow measurement according to the present specification may comprise more or less components than shown in Fig. 10. In particular, digital signals according to the DTRAF and DTRAC methods mentioned below can be generated by an adjustable pulse generator unit (76) and do not require a corresponding module (68) and/or a data shape database. waveform (70) and/or a waveform generator (69).

[0135] In the case of a time-inverted signal generated by PWM, functionality is provided to invert a received response signal, or a part of the received response signal, with respect to time and to send the inverted signal as a signal of input for the pulse width modulator.

[0136] The procedures in the following Figs. 11 to 13 can be used to generate digital output signals, especially adapted for use in a measurement according to the invention. Among others, the phase and amplitude of the digital signal can be varied.

[0137] Fig. 11 illustrates an iterative procedure for generating an ultrasound output signal that meets predefined criteria.

[0138] In a first step (80), a measurement signal is generated with a pulse generator. In a second step (81), the measurement signal is applied to a first transducer (11). A response signal is measured at the second transducer (13) in step (82)

[0140] In an additional step (83), the response signal is evaluated according to predetermined criteria. For example, the response signal can be combined with a predetermined waveform. If it is determined in step (84) that the response signal meets the predetermined criteria, the measurement signal parameters are determined and stored in step (86), preferably in the waveform database (70) for use later.

[0141] Otherwise, the frequency, amplitude and/or phase of the measurement signal or also other signal parameters are adjusted in step (85) and the procedure returns to the first step (80) in which a predetermined measurement signal is generated.

[0142] This iterative method can also be applied to a multiple transducer array. In the case of multiple send transducers, the individual frequencies, amplitudes or phases of the respective measurement signals are adjusted. In the case of multiple receiver transducers, the criteria are applied to the response signals that are received at the receiver transducers.

[0143] Fig. 12 shows another method of deriving an arbitrary waveform signal, comprising the following steps:

[0144] In a step (90), a transducer bandwidth is measured. In one example, the transducer bandwidth is approximately 300 kHz around a transducer center frequency of approximately 1 MHz. An example of a signal that makes good use of this transducer bandwidth is a signal, which has a rectangular bandwidth of 300 kHz in the frequency domain.

[0145] A band-limited function, such as a rectangular signal, is generated in the frequency domain. A corresponding signal or function in the time domain is obtained in step (92) by applying the reverse Fourier transform, which produces a function similar to sinc.

[0146] The sync-like function is then truncated to a suitable signal length, which does not have much signal strength but enough information in step (93). The signal is then used as an input signal to a pulse generator or pulse width modulation in step (94) in order to generate a pulse train.

[0147] The resulting signal is used as a measurement signal in the upstream and downstream directions in steps (95 and 96).

[0148] Fig. 13 shows another method for deriving a measurement signal. The method in Fig. 13 is a variation of the method of Fig. 12. For the sake of brevity, similar steps are not explained again. According to the method in Fig. 13, the function or waveform is adjusted so that there is no displacement at zero fluid flow.

[0149] In a step (105), it is tested whether a signal time deviation is below a predetermined threshold. If the offset is below the predetermined threshold, the signal parameters are stored in a step (107). Otherwise, the procedure returns to step (101).

• 1) Uma diferença de tempo de voo ΔT é obtida a partir da diferença das medições a montante e a jusante.
• 2) A frequência do sinal é ajustada. Depois disso, o procedimento volta para 1) e a amplitude e a fase do sinal no domínio do tempo são alteradas até que haja deslocamento de zero, o que implica que uma diferença de tempo de zero seja medida para o fluxo de fluido zero.
• 3) Alternativamente, o sinal em 1) é gerado com uma etapa de correção 2). Uma pré-distorção é aplicada ao sinal de acordo com critérios predeterminados. A pré-distorção pode ser escolhida, de modo que o receptor possa ser projetado de uma forma que seja adequada para um sinal de resposta recebido específico. Por exemplo, os sinais em 1) e 2) podem ser prédistorcidos de modo que os cruzamentos de zero no receptor apareçam em ntervalos de tempo equidistantes. Assim, um receptor de largura de banda estreita pode ser usado. Isso se aplica ao tempo de vôo e medições TRA.

[0150] Another method (not shown in the figures) is explained below:

• 1) A time-of-flight difference ΔT is obtained from the difference of upstream and downstream measurements.
• 2) The signal frequency is adjusted. After that, the procedure goes back to 1) and the amplitude and phase of the signal in the time domain are changed until there is zero offset, which implies that a time difference of zero is measured for zero fluid flow.
• 3) Alternatively, the signal in 1) is generated with a correction step 2). Pre-distortion is applied to the signal according to predetermined criteria. Pre-distortion can be chosen so that the receiver can be designed in a way that is suitable for a specific received response signal. For example, the signals in 1) and 2) can be predistorted so that zero crossings at the receiver appear at equidistant time intervals. Thus, a narrow-bandwidth receiver can be used. This applies to flight time and TRA measurements.

[0151] In steps 1-3, an arbitrary digital waveform is used instead of just an oscillation with a rectangular envelope. The aforementioned procedure and the procedures of Figs. 11 to 13 can be used for all digital time-of-flight and time inversion systems.

[0152] In the following, flow measurements using a DTRAF method, using a DTRAC method, and using a time-inverted signal that is applied to a pulse-width modulation are explained in more detail.

[0153] Digital TRA filtering (DTRAF) is a digital version of the well-known TRA method. A basic concept of DTRAF is to not send a time-reversal signal and instead replace it with a digital filter operation, but it is also possible to send the time-reversal signal together with DTRAF (or DTRAC) without change the underlying principles. Consequently, the DTRAF training process produces a digital FIR filter, where the TRA method is performed digitally. This simplifies the electronics in terms of removing the TDC and lower specifications on the DAC, for example with regard to accuracy and electrical power consumption. In contrast, the ADC is an important component. In the end, interpolation of the processed signals increases the accuracy of the sampling period to the picosecond region.

[0154] The proposed DTRAF method is illustrated in Fig. 19 in block diagram form.

[0155] In the arrangement of Fig. 19, the digital buffer (67) of the arrangement of Fig. 10 is replaced by an adjustable pulse generator unit (APGU) 76. The method is now explained in more detail:

com Ts como o período de amostragem. Este período de amostragem deve coincidir com o período de amostragem das medições posteriores. Caso um ADC mais rápido tenha sido usado, o filtro digital precisa ser reamostrado para o período de amostragem do modo de medição.[0156] First, the TRA training process for the flow measuring device (60) is done with high precision or with the on-board ADC. The training output is a digital FIR filter. The system gets excited with an input signal x[n] in a training mode:

with Ts as the sampling period. This sampling period must coincide with the sampling period for subsequent measurements. If a faster ADC was used, the digital filter needs to be resampled for the measurement mode sampling period.

y'12 e y'21[n] são os sinais de medição digitalizados. Aqui e no seguinte, a variável entre colchetes denota um índice de tempo discreto, que corresponde a n vezes o tempo de amostragem Ts no caso de uma amostragem uniforme que começa no tempo t = 0.[0157] In a measurement mode, the system is animated again with the input signal x[n]:

y'12 and y'21[n] are the digitized measurement signals. Here and in the following, the variable in square brackets denotes a discrete time index, which corresponds to n times the sampling time Ts in the case of uniform sampling starting at time t = 0.

[0158] In the next step, these digitized measurement signals are involved with the trained digital filters with time inversion:

[0159] This additional filter operation increases computational time due to the many multiplication and addition operations. Note here that it can be beneficial when instead of two different trained digital filter units (120, 121) the same digital filter units in both convolutions are used. Time inversion can be performed in memory, storing in reverse order or also reading in reverse order. Subsequently, these two filtered measurement signals y12[n] and y21[n] are correlated (or combined by any pattern matching) with each other in a cross-correlation unit 122:

em que n é um índice de tempo e r é a função de correlação entre os dois sinais de medição y12[n] , y21[n] .[0160] It is important to note here that this cross-correlation with the cross-correlation unit (122) does not need to be fully calculated. In particular, the correlation function need not be sampled at such a high frequency that the maximum is always close to a sampling time. Cross-correlation is an estimator for time delays and its maximum argument corresponds to the time delay between the two measurement signals:

where n is a time index and r is the correlation function between the two measurement signals y12[n] , y21[n] .

[0161] The time delay between the upstream and downstream signal is twice the time delay between the signal at zero flow and a signal at current fluid flow.

Com baixas taxas de amostragem, isso é com precisão suficiente para estimar n ^ *:

onde n ̂^ * denota uma estimativa de n ^ *. Em termos geométricos, n ̂^ * é o atraso em amostras inteiras entre os dois sinais de medição filtrados y12[n] e y21[n] .[0162] However, to calculate not all cross-correlation, it is sufficient to have an estimate where the maximum argument is located. This can be done by estimating the absolute transit times Tabs, 12 and T abs, 21. Subtracting these transit times gives, by definition:

With low sampling rates, this is accurately enough to estimate n^*:

where n ̂^* denotes an estimate of n^*. In geometric terms, n ̂^ * is the delay in whole samples between the two filtered measurement signals y12[n] and y21[n] .

[0163] Finally, only a few points of cross-correlation (or any pattern matching) need be calculated around lag n ̂^*. The number of points depends on the interpolation method. The interpolation method increases the accuracy of the sampling period down to the picosecond range. As an example, cosine interpolation (needs three points) works great for these small band ultrasonic transducer signals with the cross correlation method. Cosine peak interpolation gives the distance d from lag n ̂^ * to the hidden maximum of the continuous cosine function. This interpolation is illustrated in Fig. 20.

Este d então corrige a estimativa de latência n ̂^ * para obter um ∆T altamente preciso:

[0164] From F. Viola, W. Walker, “A Spline-Based Algorithm for Continuous TimeDelay Estimation Using Sampled Data”, IEEE transactions on ultrasonics, ferroelectrics and frequency control, Vol. 52, No. 1, January 2005, it is known that the cosine peak interpolation of a function y[k] can be calculated as follows:

This d then corrects the latency estimate n ̂^* to obtain a highly accurate ∆T:

[0165] The multiplication with ½ can be omitted when considered in the final calculation of the flow velocity.

[0166] There is a slightly different method in relation to DTRAF (Digital TRA-Filtering), called DTRAC (Digital TRA-Correlation). The training and measurement process to obtain the trained FIR filters and measurement signals is identical to DTRAF. However, the signal processing chain changes. Instead of filtering the entire signal with the trained (time-inverted) FIR filters, only a few points are computed for a cross-correlation with the non-time-inverted FIR filters. This is done for the upstream and downstream measurement signal.

[0167] Through the interpolation of these correlation points, accurate absolute transit times are obtained and can be used to calculate the ∆T by subtraction. The DTRAC method is exemplarily illustrated in Fig. 21 as a block diagram

[0168] Similar to the DTRAF method, an adjustable pulse generator unit 76 can be used to perform DTRAC. In the following, the proposed DTRAC method is explained in more detail:

Isso produz os coeficientes do filtro FIR h_ (MF, 12) [n] e h_ (MF, 21) [n] com Ts como período de amostragem.[0169] Firstly, TRA training is done in the same way as for DTRAF with an input signal x [n] , while x (t) is the time continuous form of x [n] :

This produces the FIR filter coefficients h_ (MF, 12) [n] and h_ (MF, 21) [n] with Ts as the sampling period.

Os sinais de medição são quantizados e amostrados com Ts. Os sinais de medição são então correlacionados com os filtros digitais:

[0170] Measurement works in the same way as TRA training. The same input signal x[n] is used:

The measurement signals are quantized and sampled with Ts. The measurement signals are then correlated with the digital filters:

nxy corresponde à estimativa de n ^ *. Um grande período de amostragem está reduzindo o risco de uma estimativa errada de n ̂12 ^ * ou n ̂21 ^ *. O método de interpolação usa o conhecimento da função contínua no tempo por trás da correlação cruzada discreta no tempo (ou qualquer correspondência de padrão). No caso de transdutores ultrassônicos de banda estreita, a saída de correlação cruzada tende a ser muito semelhante a uma oscilação do cosseno em torno do pico máximo real. Este efeito pode ser ainda maior com um sinal de entrada, que excita apenas a frequência de ressonância do par de transdutores ultrassônicos como um seno longo, cosseno, oscilação triangular ou retangular. O método de interpolação estima o índice do máximo real da correlação cruzada discreta no tempo em subamostras (e tempos com a multiplicação do período de amostragem):

d12 e d21 correspondem à correção da subamostra do método de interpolação. Como os tempos de trânsito absolutos T(abs, 12) e T(abs, 21) são muito precisos, o ∆T pode ser obtido através da subtração destes:

A multiplicação com ½ pode ser omitida quando considerada no cálculo da velocidade do fluxo.[0171] The cross correlation is a time delay estimator. However, other time delay estimators (or pattern combiners) can be used such as sum of squared differences (SSD). The two cross-correlations need not be fully calculated; instead, only a few points around the actual maximum peak need to be calculated. This is enough for the calculation of the interpolation method:

nxy matches the estimate of n^*. A large sampling period is reducing the risk of a wrong estimate of n ̂12 ^ * or n ̂21 ^ *. The interpolation method uses knowledge of the continuous-time function behind the discrete-time cross-correlation (or any pattern matching). In the case of narrowband ultrasonic transducers, the cross-correlation output tends to be very similar to a cosine oscillation around the true maximum peak. This effect can be even greater with an input signal, which only excites the resonant frequency of the ultrasonic transducer pair as a long sine, cosine, triangular or rectangular oscillation. The interpolation method estimates the real maximum index of the discrete cross-correlation in time in subsamples (and times with the sampling period multiplication):

d12 and d21 correspond to the subsample correction of the interpolation method. As the absolute transit times T(abs, 12) and T(abs, 21) are very precise, ∆T can be obtained by subtracting them:

Multiplication with ½ can be omitted when considering flow velocity calculations.

[0172] The proposed DTRAF and DTRAC methods can model the channel (77) with the conduit (12) and the fluid in the computer. This can be explained as follows: if an LTI system has blocks or transfer functions A, B, C, D, then the resulting transfer function in the frequency domain is the product A * B * C * D of the transfer functions individual A, B, C, D. This product does not depend on the order of transfer functions A, B, C, D. Therefore, the calculation of the channel response can be "shifted" into the digital domain.

[0173] Fig. 14 shows a PWM signal 129 from a pulse width modulator, which has a proper modulated on/off cycle, such that the PWM signal (129) approximates a time-inverted TRA signal (130).

[0174] Fig. 15 shows a comparison of a response signal (131) for the PWM signal (129) and a response signal (132) for the time-inverted TRA signal (130).

[0174] According to the PWM-based TRA method, a time-inverted signal is approximated according to UA = A • (UE-UD) signal, where signal (x) = 1 for x> = 0 and 0 plus

[0175] By way of example, the pulse width modulation can be based on an example with a transducer, a triangular signal frequency of 2 MHz and a PWM time resolution of approximately 83 ns.

[0176] The approximation of the inverted signal in time, which is shown in Fig. 14, is quite good, although the PWM generation is kept simple and no additional filters are used.

[0177] The simulation with correlated TRA based on PWM shown in Fig. 16 demonstrates how ΔT performance changes. As for the simulations, the signal power is calculated and the noise is scaled, in these simulations the SNR for the PWM-based correlated TRA must be increased

[0178] The results of Figs. 16 and 17 are obtained by simulations, in which the system including signal generation, transducers, channel and signal reception is modeled as a time-invariant linear system. The simulations model measurements with zero flow.

[0179] Fig. 16 shows measured time delays for PWM-based correlated TRA. The measured time delays are indicated by the column diagram (133) and a resulting probability density is shown by the Gaussian curve (134).

[0180] Fig. 17 shows the measurement results for the signals generated with the digital TRA correlation method mentioned above. The measured time delays are indicated by the column diagram (135) and a resulting probability density is shown by the Gaussian curve (136).

[0181] Fig. 18 shows measurement results for signals generated with the digital TRA filtering method mentioned above. The measured time delays are indicated by the column diagram (135) and a resulting probability density is shown by the Gaussian curve (136).

[0182] A comparison between the methods yields relatively similar results (eg, with regard to mean, standard deviation). Note that the simulation parameters used to generate measurement results shown in Figures 16, 17 and 18 are differentiated.

[0183] For the measurement of Fig. 18, pulse trains with 25 pulses are used as an input signal. The first 50 TRA training measurements are calculated and stored in two FIR filters. ΔT_DTRAF is the result of the digital TRA filtering method (DTRAF). The test tube has closed ends to make zero flow measurements.

[0184] Figure 19 shows a visualization of the DTRAF method as a block diagram. Here, an adjustable pulse generating unit (76) is connected to an input signal source on an input side and a first ultrasonic transducer (11) on an output side. The first ultrasonic transducer (11) is coupled to a channel (77), like a conduit (12) with a fluid. A second ultrasonic transducer (13) is coupled to the channel (77) and an output of the second ultrasonic transducer (13) is coupled to an ADC (64). The same arrangement is also used during the measurement phase, which is shown in Fig. 19 below the training phase.

[0185] The bottom row shows a signal evaluation of a digitized received response signal during a measurement phase. A first digital filter unit (120) and a second digital filter unit (121) are respectively connected to an output of the ADC (64). The respective outputs of the digital filter units (120), (121) are connected to an input of a cross-correlation unit (122) and an output of the cross-correlation unit (122) is connected to an input of an interpolation unit (123).

[0186] Fig. 20 shows an illustration of a general peak interpolation to determine a maximum of a correlation function. The general peak interpolation according to Fig. 20 was explained above and is not repeated in detail here. In short, a correlation function is locally approximated by a cosine function or a higher order polynomial. This cosine is found by means of interpolation with three interpolation points y [k-1] , y [k] and y [k + 1] . Then, the maximum of the cosine function is determined. If more interpolation points are available, a least squares method can be used to find the cosine. The distance of the interpolation points to the cosine can also be minimized according to a different type of norm.

[0187] Fig. 21 shows a visualization of the proposed DTRAC method as a block diagram. The arrangement of Fig. 21 is similar to the arrangement of Fig. 19. For the sake of brevity, only the components in the bottom row, which are different from those in Fig. 19, are explained below.

[0188] An input of a cross-correlation unit (122) is connected to an output of the ADC (64) and an output of the cross-correlation unit (122) is connected to an input of an interpolation unit (123).

[0189] Figs. 19 and 21 can also be seen as a possible realization of the DTRAC and DTRAF methods by hardware and/or software components. The various units shown in Figs. 19 and 21 can be realized by separate components or in a component and can be realized in hardware and/or software. For example, the cross-correlation unit (122) and the interpolation unit (123) can be realized on the same integrated circuit. The training and measurement procedures according to Figs. 19 to 21 were explained above and are not repeated here.

[0190] Fig. 22 shows, by way of example, another embodiment of a proposed flow measuring device (60) for measuring a flow velocity of a fluid in a fluid conduit (12) with an ultrasonic transit time flowmeter ( 60) using a digital acoustic filter time-reversal method.

[0191] The buffer (67) is connected to an adjustable pulse generator unit (76). The ADC (64) is furthermore connected to digital filter units (120), (121). The digital filter units (120), (121) are connected to a correlation unit (122), which is connected to an interpolation unit (123).

[0192] The interpolation unit (123) is connected to a speed calculation unit (71), which is connected to a result memory (72) and to the buffer (67).

Claims (13)

Method for determining a flow velocity of a fluid in a channel (77) with a fluid conduit (12) with an ultrasonic travel time flow meter (60) using a digital time reversal acoustic filter method, the method characterized by comprising the following steps:

• - performing a training process with an adjustable pulse generator unit (76), the training process comprising the steps:
• - applying an input signal (x [n] ) to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) at a first location;
• - receiving a first response signal to the input signal (x [n] ) at a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) at a second location, the second position upstream or downstream of the first location with respect to a flow direction (14) of the fluid;
• - converting the first response signal into a digitized first response signal using an analog-to-digital converter (64); and
• - deriving a first digital response filter (120) from the first digitized response signal;
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a second response signal to the input signal (x[n]) at the first ultrasonic transducer (11);
• - converting the second response signal into a second digitized response signal using the analog-to-digital converter (64); and
• - deriving a second digital response filter (121) from the second digitized response signal;
• - performing a measurement process with the adjustable pulse generator unit (76), the measurement process comprising the steps:
• - apply the input signal (x [n] ) to the first ultrasonic transducer (11);
• - receiving a third response signal from the input signal (x [n] ) in the second ultrasonic transducer (13);
• - converting the response signal into a third digitized response signal (y12' [n] ) using the analog-to-digital converter (64);
• - performing a reverse direction measurement process with the adjustable pulse generator unit (76), the reverse direction measurement process comprising the steps:
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a fourth response signal from the input signal (x [n] ) in the first ultrasonic transducer (11);
• - converting the fourth response signal into a fourth digitized reverse direction response signal (y21' [n] ) using the analog-to-digital converter (64);
• - deriving a first correlation input signal (y12 [n] ) from the third digitized response signal (y12´[n] ) and the first digital response filter (120), the derivation comprising inverting the third signal digitized response (y12´[n] ) or the first digital response filter (120) with respect to time;
• - deriving a second correlation input signal (y21 [n] ) from the fourth digitized response signal (y21´[n] ) and the second digital response filter (121), the derivation comprising inverting the fourth signal digitized response (y21´[n] ) or the second digital response filter (121) with respect to time; and
• - deriving a difference time of flight (ΔT) of the first correlation input signal (y12[n] ) and the second correlation input signal (y21[n] ) by calculating a discrete correlation of the first correlation input signal ( y12 [n] ) with the second correlation input signal (y21 [n] ), where a time index of a maximum value of the discrete correlation using an interpolation method is determined.

Method according to claim 1, characterized in that the second digital response filter (121) is the same as the first digital response filter (120). Method according to claim 1, characterized in that the second digital response filter (121) is a digital reverse direction response filter, wherein a determination of the reverse direction response filter comprises the following steps:

• - performing a reverse direction training process with the adjustable pulse generator unit (76);
• - applying the reverse direction input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a reverse direction response signal for the reverse direction training signal at the first ultrasonic transducer (11);
• - converting the reverse direction response signal into a digitized reverse direction response signal using the analog-to-digital converter (64);
• - deriving digital reverse direction response filter taps (hMF,21 [n] ) from the digitized reverse direction response signal; and
• - deriving the second digital response filter (121) from digital reverse direction response filter derivations (hMF,21[n]).

Method for determining a flow velocity of a fluid in a channel (77) with a fluid conduit (12) with an ultrasonic travel time flow meter (60) using a digital time reversal acoustic filter method, the method characterized by comprising the steps:

• - performing a training process with an adjustable pulse generator unit (76), the training process comprising the steps:
• - applying an input signal (x [n] ) to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) at a first location;
• - receiving a first response signal to the input signal (x [n] ) at a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) at a second location, the second location upstream or downstream of the first location with respect to a fluid flow direction;
• - converting the first response signal into a digitized first response signal using an analog-to-digital converter (64);
• - determining a first digital response filter with filter taps (hMF, 12 [n] ) from the first digitized response signal;
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a second response signal at the first ultrasonic transducer (11);
• - converting the second response signal into a second digitized response signal using the analog-to-digital converter (64); and
• - determining a second digital response filter with filter derivations (hMF, 21 [n] ) from the second digitized response signal;
• - performing a measurement process with the adjustable pulse generator unit (76), the measurement process comprising the steps:
• - apply the input signal (x [n] ) to the first ultrasonic transducer (11);
• - receiving a third response signal from the input signal (x [n] ) in the second ultrasonic transducer (13);
• - converting the third response signal into a third digitized response signal (y12' [n] ) using the analog-to-digital converter (64);
• - performing a reverse direction measurement process with the adjustable pulse generator unit (76), the reverse direction measurement process comprising the steps:
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a fourth response signal from the input signal (x [n] ) in the first ultrasonic transducer (11);
• - converting the fourth response signal into a fourth digitized response signal (y21´ [n] ) using the analog-to-digital converter (64);
• - derive a first time of flight (Tabs,12) from the third digitized response signal (y12'[n] ) and the first digitized response signal by calculating a discrete correlation of the third digitized response signal (y12'[n] ) with the first digitized response signal, wherein a time index of a maximum value of the discrete correlation using an interpolation method is determined;
• - derive a second time of flight (Tabs,21) from the fourth digitized response signal (y21'[n] ) and the second digitized response signal by calculating a discrete correlation of the fourth digitized response signal (y21'[n] ) with the second digitized response signal, wherein a time index of a maximum value of the discrete correlation using an interpolation method is determined; and
• - deriving a time-of-flight difference (ΔT) by a subtraction of previously derived flight times (Tabs, 12, Tabs, 21).

Method according to claim 4, characterized in that the second digital response filter is the same as the first digital response filter. Method, according to claim 4, characterized in that the first digital response filter is a direct direction response filter; wherein the second digital response filter is a reverse direction response filter; wherein the filter taps (hMF,12) of the forward direction response filter are derived from the first digitized response signal; and wherein the filter taps (hMF, 21) of the reverse direction response filter are derived from the second digitized response signal Method for determining a flow velocity of a fluid in a fluid conduit (12) using an ultrasonic travel time flow meter, the method characterized by comprising the steps:

• - generating a training signal with an adjustable pulse generator unit (76);
• - applying the training signal to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) in a first position;
• - receiving a response signal to the training signal in a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) in a second position, the second position being upstream or downstream of the first position with respect to a fluid flow direction;
• - converting the response signal into a digitized response signal using an analog-to-digital converter (64);
• - inverting the digitized response signal with respect to time to obtain an inverted digitized response signal;
• - generating a pulse width modulated measurement signal from the inverted digitized response signal;
• - applying the pulse width modulated measurement signal to the first ultrasonic transducer (11);
• - receiving a response signal from the pulse width modulated measurement signal in the second ultrasonic transducer (13); and
• - derive a time of flight (ΔT) from the response signal.

Electronic processing unit (53) for determining a flow velocity of a fluid in a channel (77) with a fluid conduit (12) with an ultrasonic travel time flow meter (60) using an acoustic filter method of digital time reversal, characterized by comprising:

• - an adjustable pulse generating unit (76), the pulse generating unit (76) being operative to carry out a training process with an input signal (x[n]);
• - a transmission means, the transmission means being operative to apply the input signal (x [n] ) to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) at a first location;
• - a receiving means, the receiving means being operative to receive a first response signal to the input signal (x[n]) in a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the conduit of fluid (12) at a second location, the second location being upstream or downstream of the first location relative to a fluid flow direction;
• - an analogue to digital converter (64), the analogue to digital converter (64) being operative to convert the first response signal into a digitized first response signal;
• - a processing means, the processing means being operative to determine a first digital response filter (120) with FIR filter derivations (h_MF12[n]) from the digitized response signal;
• - the transmission means being further operative to apply the input signal (x [n] ) to a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) at the second location;
• - the receiving means further being operative to receive a second signal in response to the input signal (x[n]) in the first ultrasonic transducer (11);
• - the analogue to digital converter (64) being further operative to convert the second response signal into a second digitized response signal;
• - the processing means further being operative to determine a second digital response filter (121) with FIR filter taps (hMF,21 [n] ) from the second digitized response signal;
• - the adjustable pulse generating unit (76) being additionally operative to carry out a measurement process with the input signal (x[n]);
• - the transmission means being additionally operative to apply the input signal (x [n] ) of the measurement process to the first ultrasonic transducer (11);
• - the receiving means further being operative to receive a third response signal from the input signal (x[n]) of the measurement process in the second ultrasonic transducer (13);
• - the processing means being operative to convert the third response signal into a third digitized response signal (y12') using the analog-to-digital converter (64);
• - the adjustable pulse generator unit (76) being additionally operative to perform a reverse direction measurement process with the input signal (x[n]);
• - the transmission means being additionally operative to apply the input signal (x [n] ) of the reverse direction measurement process to the second ultrasonic transducer (13);
• - the receiving means further being operative to receive a fourth response signal from the reverse direction measurement process at the first ultrasonic transducer (11);
• - the analogue-digital converter (64) being further operative to convert the fourth response signal of the reverse direction measurement process into a fourth digitized response signal (y21') using the analogue-digital converter (64);
• - the processing means being operative for:
• - deriving a first correlation input signal (y12) from the third digitized response signal (y12') and the first digital response filter (120), the derivation comprising inverting the third digitized response signal (y12'); or the first digital response filter (120) with respect to time;
• - deriving a second correlation input signal (y21) from the fourth digitized response signal (y21') and the second digital response filter (121), the derivation comprising inverting the fourth digitized response signal or the second filter digital response (121) with respect to time;
• - deriving a difference time of flight (ΔT) of the first correlation input signal (y12 [n] ) and the second correlation input signal (y21 [n] ) by calculating a discrete correlation of the first correlation input signal ( y12 [n] ) and the second correlation input signal (y21 [n] ) and determine a time index of a maximum value of the discrete correlation using an interpolation method.

Electronic processing unit (53) according to claim 8, characterized in that the second digital response filter (121) is the same as the first digital response filter (120). Electronic processing unit (53) according to claim 8, characterized in that the second digital response filter (121) is a digital reverse direction response filter, and wherein the electronic processing unit (53) is still operative for

• - generating a reverse direction training signal with an adjustable pulse generator unit (76);
• - apply the reverse direction training signal to the second ultrasonic transducer (13);
• - receiving a reverse direction response signal for the reverse direction training signal at the first ultrasonic transducer (11);
• - converting the reverse direction response signal into a digitized reverse direction response signal using an analog-to-digital converter (64); and
• - deriving the second digital response filter (121) with FIR filter taps (hMF, 21 [n] ) from the digitized reverse direction response signal.

Electronic processing unit (53) for determining a flow velocity of a fluid in a channel (77) with a fluid conduit (12) with an ultrasonic travel time flow meter (60) using an acoustic filter method of digital time reversal, in which the electronic processing unit (53) is characterized by being operative for:

• - carrying out a training process with an adjustable pulse generating unit (76);
• - applying an input signal (x [n] ) to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) at a first location;
• - receiving a first response signal to the input signal (x [n] ) at a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) at a second location, the second location upstream or downstream of the first position with respect to a fluid flow direction;
• - converting the first response signal into a digitized first response signal using an analog to digital converter (64);
• - deriving a digital response filter with FIR filter derivations (hMF,12 [n] ) from the first digitized response signal;
• - performing a reverse direction training process with the adjustable pulse generator unit (76);
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a second response signal at the first ultrasonic transducer (11);
• - converting the second response signal into a second digitized response signal using the analog-to-digital converter (64);
• - determining a digital response filter with filter derivations (hMF, 21 [n] ) from the second digitized reverse direction response signal;
• - carrying out a measurement process with the adjustable pulse generator unit (76);
• - apply the input signal (x [n] ) to the first ultrasonic transducer (11);
• - receiving a third response signal from the input signal (x [n] ) in the second ultrasonic transducer (13);
• - converting the third response signal into a third digitized response signal (y12' [n] ) using the analog-to-digital converter (64);
• - performing a reverse direction measurement process with the adjustable pulse generator unit (76);
• - apply the input signal (x [n] ) to the second ultrasonic transducer (13);
• - receiving a fourth response signal from the input signal (x [n] ) in the first ultrasonic transducer (11);
• - converting the fourth response signal into a fourth digitized response signal (y21´ [n] ) using the analog-to-digital converter (64);
• - deriving a first correlation signal (r12) from the third digitized response signal (y12´) and the digital response filter with filter derivations (hMF, 12 [n] ) calculating a convolution, the derivation comprising the inversion of the third digitized response signal (y12´) or the digital filter with the filter derivations (hMF, 12 [n] ) with respect to time;
• - deriving a second correlation signal (r21) from the fourth digitized response signal (y21') and a second digital response filter with filter taps (hMF,21 [n] ) calculating a convolution, the derivation comprising inverting the fourth response digitized signal or the digital filter with the derivations of the filter (hMF,21 [n] ) with respect to time;
• - deriving flight times (Tabs,12, Tabs,21) from each correlation signal (r12, r21) by an interpolation method; and
• - derive a time-of-flight difference (ΔT) by subtracting the previously derived flight times (Tabs, 12 - Tabs, 21).

Electronic processing unit (53), according to claim 11, characterized in that the digital response filter with the filter taps (hMF,21 [n] ) is equal to the digital response filter with the filter taps (hMF,12 [n] ). Electronic processing unit (53) according to claim 12, wherein the digital response filter with the filter taps (hMF,21 [n] ) is a digital reverse direction response filter and wherein a tap of the reverse direction response filter characterized by comprising the steps:

• - performing a reverse direction training process with the adjustable pulse generator unit (76);
• - applying the reverse direction training signal to the second ultrasonic transducer (13);
• - receiving a reverse direction response signal for the reverse direction training signal at the first ultrasonic transducer (11);
• - converting the reverse direction response signal into a digitized reverse direction response signal using an analog-to-digital converter (64);
• - derive the digital reverse direction response filter with FIR filter taps (hMF,21 [n] ) from the digitized reverse direction response signal. 14. Electronic processing unit (53) for determining a flow velocity of a fluid in a fluid conduit (12) using an ultrasonic travel time flowmeter, the electronic processing unit (53), characterized in that it is being operative for:
• - generating a training signal with an adjustable pulse generator unit (76);
• - applying an input signal (x [n] ) to a first ultrasonic transducer (11), the first ultrasonic transducer (11) being mounted in the fluid conduit (12) at a first location;
• - receiving a response signal to the training signal at a second ultrasonic transducer (13), the second ultrasonic transducer (13) being mounted in the fluid conduit (12) at a second location, the second location being upstream or downstream of the first location with respect to a fluid flow direction;
• - converting the response signal into a digitized response signal using an analog-to-digital converter (64);
• - inverting the digitized response signal with respect to time to obtain an inverted digitized response signal;
• - perform a pulse width modulated measurement process with the inverted digitized response signal,
• - applying the pulse width modulated measurement signal to the first ultrasonic transducer (11);
• - receiving a response signal from the pulse width modulated measurement signal in the second ultrasonic transducer (13); and
• - derive a time of flight (ΔT) from the response signal.

Images