FI88208C - FARING EQUIPMENT FOR ACOUSTIC MAINTENANCE AV ENTRY - Google Patents

Description

88208

Method and apparatus for acoustic measurement of gas flow Förfarande och anordning för acustisk mätning av en gasström 5

The invention relates to an acoustic flow measurement method for measuring the flow rate of gases and / or the quantities derived therefrom, in which a long-wave sound is transmitted to a measuring tube and two and quantities derived therefrom, and in which method a sound is selected to be sent to the measuring tube, the frequencies of which are selected between a certain minimum and maximum frequency.

The invention furthermore relates to a device for measuring the flow rate of gases and / or the quantities derived therefrom, such as volume flow and / or mass flow, which device comprises a measuring tube in which the gas flow to be measured flows, which device transmits audio signals and microphones as audio detectors. positioned: in connection with a measuring tube between said speakers at a certain sensing distance, the device comprising audio sequence devices for repeatedly supplying ____ in a certain frequency range; electrical signal sequences for said speakers.

The most important physical observation from the point of view of the measurement technique according to the invention is that in a rigid-walled pipe only a so-called a plane wave mode, the propagation speed of which does not depend on local variations in the medium, its temperature or flow rate, but only on the average values prevailing in the measuring range (Robertson 1977, -.1 1984), which enables accurate profile-independent flow measurement. For a circular cross - section pipe, said cut - off frequency can be calculated from the formula fc - c / (1.7 * D) (1) 2 882;.: Where c is the rate of sound propagation in the medium filling the pipe and D is the diameter of the pipe.

Finnish patent 76885 discloses a technical solution in which sound 5 is fed into a tube in the form of a frequency scan and the propagation time of sound is determined from the signals of microphones placed at the ends of the measuring interval by a polarity correlator. The flow times measured upstream and downstream can be used to determine with high accuracy both the average flow rate and the speed of sound in a resting medium. From the flow rate and the cross-sectional area of the pipe, the volume flow rate and from there the pressure and temperature measurement results can be calculated in a known manner by combining the mass flow rate. The same solution is also described in Ultrasonics International 1987 Conf. Proc. M.Kuuttila, P.Hiismäki: "An Acoustic Method for Hight Precision Gas Low Measure-15 ments.

Em. The method presented in FI patent 76885 directly correlates the signals detected by both microphones, but does not make any use of the frequency sweep that transmits information, which is fully known. The correlation function thus accumulates partly from the sound deliberately transmitted to the tube and partly from the noise present in the tube, the occurrence of which is difficult to prevent or control. In order to obtain a reliable measurement result, in practice it is necessary that the volume of the transmitted sound always overcomes the intensity of the background noise.

25 FI patent application 916102 (filing date 23.12.1991, applicant Instrument Instrument Factory Kytölä Oy) has sought to further develop the solution presented in the above-mentioned FI patent 76885 by reducing the effect of background noise by applying signals from both microphones to identical wipers with the same 30-30 delay. , for the filters that follow the frequency sweep to be transmitted, and only in this way are the filtered signals correlated. At any given time, only the frequencies that hit the narrow, sweep frequency window will pass through. Em. The solution presented in the FI application can work well in practice, as long as the frequency-35 scan is not too fast. As the sweep is shortened, the "instantaneous" 3 8 8 2 C o frequency becomes increasingly white to be determined and the operation of the filters is questioned.

When aiming for a meter whose display reacts quickly to changes in flow rate 5, it must be possible to measure the speeds of sound measured upstream and downstream without interfering with each other simultaneously and in the shortest possible time.

The main object of the present invention is to further develop the prior art discussed above and to eliminate the drawbacks therein.

The object of the invention is to create a new technical solution which is better suited to monitoring rapid changes in flow rate than the prior art.

15

In order to achieve the above and later objects, the method of the invention is mainly characterized in that the average flow rate and / or derived quantities 20, such as volumetric flow rate and mass flow rate, are determined in a known manner by calculating premature or delay is determined by transmitting periodically reproducible audio sequences of discrete frequency components to the measurement tube by speakers located outside the measurement interval, ha-35 silence and in the method of sequences transformed by the acoustic transmission path

4 882CS

the autocorrelation function to be generated is as sharp as possible, and that the signals detected by the sound detectors are applied to FIR filters, in which a sequence derived from the sequence measured downstream or upstream in the zero-flow situation is stored as coefficient vectors.

The device according to the invention, in turn, is mainly characterized in that the device comprises memory elements in which continuously reproducible measurement sound sequences and power amplifiers can be transmitted to the speakers, the outputs of said memory elements being connected via D / A converters and the power amplifiers being connected to said speakers connected to the measuring tube, that the device comprises FIR filters to which the necessary coefficient sequences can be loaded and to the input of the FIR filters said microphones connected to the measuring tube via A / D converters, that the device comprises memory circuits connected said FIR filters, that the device comprises a host processor adapted to control the 20 different operating sequences of the device and to calculate the average flow based on the sequences obtained from the latter memory circuits; opacity and / or volume flow and / or temperature and / or pressure measurement results by combining the mass flow rate.

25 Experience has shown that the correlation technique of wideband audio signals, but lower than the cut-off frequency fc described above, in itself provides a good starting point for accurately measuring the propagation time of sound. This is combined with the filter technology in such a way that the sounds advancing upstream and downstream in the pipe do not interfere with each other and that the measurement result is as independent as possible from the background noise.

In the present invention, the signal processing technique is not based on polarity correlators as in the prior art of the invention.

5,882G3

The invention applies FIR filters which are based on digital signal processing technology and which are so fast that the processing of audio-frequency signals can be carried out in real time.

The FIR filters mentioned in the invention do not correlate the two time series to determine their time difference, but compare the current time series with a coefficient vector fixedly fed to the FIR filter, whereby a real-time cross-correlation function of the coefficient vector and the input signal is obtained as the filter output signal. In the flow measurement according to the invention 10, the travel times of the sound upstream and downstream are then obtained directly as the time differences of the pulses appearing in the output signals of the above-mentioned FIR filters.

In the following, the invention will be described in detail with reference to the figures of the accompanying drawing, in connection with which the physical background of the invention and some application examples of the invention are described, to which the invention is not limited.

Figure 1 shows the power spectra of the sound sequences used in the sound propagation time measurement 20 applied in the invention, with the spectrum of even frequencies above and the corresponding spectrum of odd frequencies below.

Figure 2 shows the audio sequences used in the audio propagation time measurement 25 applied in the invention, with the sequence of even frequency components above and the sequence of odd frequency components below.

-. Figure 3 shows the correlation functions of the sequences used in the sound propagation time measurement used in the invention, with the autocorrelation function of the sequence of even frequency components at the top and the autocorrelation function of the sequence of odd frequency components at the bottom and the cross-correlation function of these in the middle.

6 88208

Figure 4 illustrates how the symmetric correlation peaks of Figure 3 become antisymmetric when the coefficient sequences are replaced by Hilbert transforms of the sequences used above.

Figure 5 shows a first embodiment of an acoustic flow meter according to the invention in block diagram form.

Figure 6 shows a block diagram of another embodiment of an acoustic flow meter according to the invention.

10

According to Figures 5 and 6, the gas flow, the flow rate v of which is acoustically measured, flows in a tube 10. Audio signals are transmitted to the measuring tube 10 upstream and downstream by speakers 12a and 12b, which are received by microphones 15a 13a and 13b interposed between speakers 12a and 12b. is essential for flow measurement. The diameter D of the measuring tube 10 and the measuring distance L are selected, for example, so that L «10D.

With respect to the theoretical background of the present invention, it should first be noted that a signal synthesized from discrete, multiple phase components of a finite frequency range of a finite frequency range of random frequency components represents pseudo-random, periodic noise. All the frequency components of such noise are constantly ringing at a constant amplitude, so that their filtering also takes place completely with stationary filters. Bearing in mind that the correlation function of periodic signals treated as discrete time vectors corresponds to the multiplication performed in frequency space by component of frequency vectors, it is seen that the product of multiplication gives a zero vector if the vectors to be multiplied do not have common non-zero frequency components. Thus, if two wideband, pseudo-random time series are generated with one having only even and the other having only odd frequency components, their correlation function is identically zero at any discrete time offset. Such time series are thus identically orthogonal. If one of these 35 is sent downstream of the measured gas flow in the measuring tube 10 and the other upstream, their 7 8820ο correlation functions can be calculated without the slightest disturbance to the other. To illustrate this, reference is made to the accompanying Figures 1, 2 and 3.

5 Figure 1 above shows the power spectrum parβ (ω) of even frequencies, the envelope of which is of type w2 * exp (-wz). This choice has been made to ensure that the correlation function does not have distant, slowly attenuating oscillations. Figure 1 below shows the corresponding power spectrum of the odd frequencies Ρ0 (ω). Said power spectra are in the frequency-10 range fo-fi · In the present invention, the upper frequency fx is generally selected to be the same as the cut-off frequency fc according to Equation (1). The low frequency f0 is preferably (0.1 ... 0.3) x fc and the center frequency fk is between them.

Figure 2 shows the time series Se (t) formed by the even frequency components above and the time series SQ (t) formed by the odd frequency components below. The total length of the time vectors is 128 channels. The even-frequency vector has two identical 64-channel periods, and the odd-frequency vector has the last 64 channels repeat the vector formed by the first ones as negative. The phases of both the 20 even and odd frequencies are selected to be identical and drawn from the same random numbers. For this reason, the time series are very similar at the beginning and end of the period and mirror images of each other in the middle of the period. However, from the point of view of the measurement principle, this fact is irrelevant and both phases could have been drawn from completely 25 independent random numbers. The choice of steps is worth · .; draws attention to the fact that the amplitude variations of the time series do not become too large, in which case the speakers 12a, 12b would have to be easily fed large ones leading to nonlinear behavior at the moment--. additional effects.

30

Figure 3 shows the even time series at the top of the autocorrelation function, the odd time series at the bottom, and the series cross-correlation function in the middle. The cross-correlation function is identically zero, as is the intention for the purposes of the invention. Every other 35 correlation peaks of the odd time series is negative and completely cancels the positive peak of the correlation function of the even time series at the corresponding position 8 8 8 2 C o.

The above division into even and odd time series requires in the invention that the amplitude information of the series is not lost. Thus, they cannot be replaced by corresponding polarity strings without losing the orthogonality property. Thus, the required signal processing technique cannot be based on polarity correlators, as in the above-mentioned FI patent 76885 and FI application 916102. The invention also applies fast, digital signal processing technique-based FIR (£ inite impulse Response) -suits 16, as described above in connection with Figures 5 and 6. 17a; 16b, 17b; 17i, 172, with which the processing of audio frequency signals can be implemented in real time. Em. Suitable for the FIR filters 16,17 are, for example, the Motorola filter-15 circuitry DSP 56200. In the invention, the FIR filters 16,17 do not correlate two time series to determine their time difference, but the input time series is compared to a coefficient vector fixed to the FIR filter 16,17, whereby the filter output signal is a real-time cross-correlation function of this coefficient vector and input signal. The travel times required for the Vir-20 background measurement are obtained directly as the time differences of the pulses present in the output signals of such FIR filters 16,17. When the time series detected separately by the microphones 13a, 13b at zero flow rate, without background noise, are selected as coefficient vectors separately, each FIR filter picks up at its output only 25 desired portions of the microphone signals 13a, 13b, effectively filtering out . The coefficient vectors can also be selected as Hilbert transforms of the vectors selected above, whereby the symmetric peaks of the correlation functions become antisymmetric. These are better suited for determining 30 exact time differences.

In order for the correlation functions to be as local as possible, without distant, slowly attenuating features, the frequency spectrum of the time series must be well controlled without sudden jumps. It can be assumed that the frequency response of both the flow measurement tube 10 and the microphones 13a, 13b is reasonably uniform. Instead, the speakers 12a, 12b 9 8 8 2 C 8 generally have to be placed in a side branch associated with the tube, which easily acts as a resonator and results in a very uneven frequency response. For this reason, the time series detected by the microphones 13a, 13b may differ significantly in shape from the time series transmitted by the power amplifiers 15a, 13b to the speaker 5 12a, 12b, but be similar to each other regardless of the time difference. Thus, the frequency spectrum of the signals transmitted by the power amplifiers 15a, 15b by the speakers 12a, 12b can be corrected so that the frequency spectra of the microphone signals can be obtained in exactly the desired shape.

10

The following are the calculation formulas used in the flow measurement according to the invention.

Flow rate [m / s] v - 0.5 * L * (t ^ 1 - t2’1) (2) 15

Volume flow [m3 / s] Q - v * A (3)

Mass flow [kg / s] M - Q * p (4) 20 v average flow rate L - distance between microphones 13a and 13b - flow time of sound downstream t2 - flow time of sound upstream "Q - volume flow; ··; 25 A - cross-sectional area of tube 10 area M - mass flow p - gas density

According to Figure 5, the operation of the system is controlled by the host processor 18.

The memory elements 14a, 14b store the measurement sequences Se (t) and S „(t) described above, which are continuously transmitted to the loudspeakers 12a, 12b via the power amplifiers 15a, 15b. Blocks 14a and 14b also include D / A converters for generating an analog signal to amplifiers 15a and 15b. According to Figure 5, the system 35 includes four FIR filters 16a, 17a; 16b, 17b, to which the audio signals converted into electrical signals detected by the microphones 13a; 13b are sent. Blocks 16a, 17a, 16b, 17b of Figure 5, as well as blocks 17χ and 172 of Figure 6, include A / D converters for converting analog signals from microphones 13a, 13b to digital input for input to FIR filters. A first FIR filter 16a connected to the microphone 13a is connected to the memory circuit 19a, and a second FIR filter 17a is respectively connected to the second memory circuit 20a. The first FIR filter 16b connected to the microphone 13b is connected to the memory circuit 19b and the second FIR filter 17b is connected to the memory circuit 20b, respectively. The outputs of the FIR filters 16a, 16b; 17a, 17b are stored in said memory circuits 19a, 20a; 19b, 20b under the control of a microprocessor 18. The microprocessor 18 calculates the times t1, t2a, tlb, t2b of the center of the peaks Pia »P2a» Pib »P2b stored in the memory circuits 19a, 20a; 19b, 20b. The flow time of the monitored gas downstream is thus in the direction of the gas flow rate v tL - tlb-tla and the flow time of the sound upstream t2 - t2a-t2b.

15

In the measurement system shown in Fig. 6, instead of four FIR filters, two FIR filters 17x and 172 are used, one of which stores the downstream sound sequence and the other the upstream sound detector sequence. The host processor 18 controls the memory circuits 14a and 14b in which the cyclically reproducible transmission sequences Sa (t) and S0 (t) are stored. Blocks 14a and 14b also represent D / A converters, the outputs of which supply power amplifiers 15a and 15b, which in turn supply measurement sound sequences to speakers 12a, 12b transmitting the measurement tube 10. The signals of both microphones 13a, 13b are fed to the inputs of the separation and summing amplifier 25 22. The output of amplifier 22 is connected to FIR filters 17x and 172. Blocks 17i and 172 also include the aforementioned A / D converters. The FIR filters met 171; 172 are connected to memory circuits 18χ and 182, one of which collects 18x the cumulative downstream correlation peaks p1 and p1b and their times t1a and p1bb accumulated at different times of each microphone 13a and 13b, and the second memory circuit 182 and their times t2b and t2a. The travel time of sound downstream tj ”tlb-tla and the travel time of sound downstream t2 - t2a-t2b, respectively.

35 Based on the above formulas (2), (3) and (4), the microprocessor 18 calculates the temperature and pressure measurement outputs of the average flow rate v, the volume flow Q and il 882C8 by combining the mass flow M. The measurement results are shown in Fig. 5 and 6 with the measurement results display and and / or printing device 21.

The simplest implementation of the method and apparatus of the invention is achieved by four FIR filters 16a, 17a fitted according to Figure 5; 16b, 17b, which are two for each microphone 13a, 13b, one for picking up forward and the other for upstream sound. When the frequency spectrum is carefully selected, the signals of the microphones 13a, 13b can be summed or separated as shown in Fig. 6, whereby only two FIR filters 17a and 17b are met. The output of each then has two spikes, the time difference between which is directly the time difference of the passage of sound in the distance between the microphones 13a, 13b.

15 If the measurement sequence is long, the FIR filters also become 16.17 long. In this case, long FIR filters can be implemented by cascading short filters. Cascading is described, for example, in the DSP 56200 data sheet for the Motorola filter circuit. The signal-to-noise ratio can also be improved by settling for short filters, but calculating the outputs of several consecutive measurement sequences as a moving average. Since there are two identical half-periods in the output sequence of even frequencies, it is also advantageous to add them together directly before determining the center of the travel time peak, since the noise portions of the first and second half-periods are independent of each other. Correspondingly, the half-cycles of the output sequences of the odd frequencies are of opposite sign, so that the latter half-cycle is preferably subtracted from the first half-cycle before the center of the transit time peak is calculated.

The following claims set forth within the scope of the inventive idea defined by the various details of the invention may vary and differ from those set forth above by way of example only.

Claims (8)

12 8 8 2 C 3 An acoustic flow measurement method for measuring the gas flow rate (v) and / or the quantities derived therefrom, in which a long-wave sound is transmitted to the measuring tube (10) and two sound detectors (13a, 13b) connected to the measuring tube are detected in the gas flow. - and countercurrent audio signals, each of which uses a correlation to determine the flow rate (v) of the gas flowing in the measuring tube (10) and / or the quantities derived therefrom, and in which the measuring tube (10) is sent a sequence of frequencies selected at a new minimum and maximum frequency ( f0 · ^), characterized in that the average flow rate and / or the quantities derived therefrom, such as the volume flow rate and the mass flow rate, are determined in a known manner by calculating zero of the advance or delay of the sound detected by the sound detectors (13a, 13b). that said advance or delay is determined by transmitting periodically reproducible sound sequences (S, (t) consisting of discrete frequency components to the measuring tube (10) by means of speakers (12a, 12b) located outside the measuring interval (L)). (t)), 25 that said audio sequence to be transmitted in the second direction (Se (t)) contains only even and said odd frequency sequence to be transmitted in the opposite direction (SQ (t)) only odd frequency components, 30 that said audio sequences are selected so that the audio detectors (13a) , 13b) the autocorrelation function generated in the method of the sequences detected and converted by the acoustic transmission path is as sharp as possible, and 35 that the signals detected by the sound detectors (13a, 13b) are applied to 13 8 82C3 FIR filters (16a, 17a, 16b, 17b; 17lf 172); stored as coefficient vectors in the zero flow situation a a sequence derived from a sequence measured downstream or upstream. Method according to Claim 1, characterized in that the method uses two sound detectors (13a, 13b), the signals of each of which are each applied to two FIR filters (16a, 17a; 16b, 17b), each of which is stored as a coefficient vector. a sequence appropriately derived from a sequence (S. (t)) measured downstream in the zero-flow situation, e.g. during the flow-15 downstream / countercurrent, the time difference measured from the filter of the centers of the flow times / countercurrent filters of each sound detector (13a, 13b) to the other is used (Fig. 5). Gas flow measurement method according to Claim 2, characterized in that the coefficient vectors of the FIR filters (16a, 17a, 16b, 17b) are, as such, zero-detector sequences measured in a zero-flow situation, the autocorrelation functions of which are spaced at the frequency of the transmission sequences. 25 man sharp. A gas flow measurement method according to claim 3, characterized in that Hilbert transforms of said sequences are selected as coefficient vectors for FIR filters. .- • 3 30 A gas flow measurement method according to claim 1, characterized in that there are two FIR filters (171172), one ... ... (17 :) for downstream sound and the other (172) for countercurrent sound, and that the signals of the two sound detectors before for the above FIR filters. 35 (171,17z) exports are added together or subtracted from each other (22), and that during the sound flow downstream / countercurrent, the autocorrelation functions of the autocorrelation functions occurring at different times at the output of the filter / countercurrent filter are used for each rubber with the sound detector (13a, 13b). the time difference between the centers (ίχ / ί2) (Figure 6). An apparatus for measuring the flow rate of gases and / or quantities derived therefrom, such as volume flow and / or mass flow, comprising a measuring tube (10) in which the gas flow to be measured flows, said speakers (12a, 12b) transmitting audio signals and microphones (13a) , 13b) arranged in connection with the measuring tube (10) between said speakers (12a, 12b) at a certain known mutual distance (L), which device comprises audio sequence devices (14a, 14b, 15a, 15b, 18) repeatedly supplying electrical signal sequences in a certain frequency range (f0-fi) to said speakers (12a, 12b), characterized in that the device 15 comprises memory elements (14a, 14b) in which continuously reproducible measurement sound sequences and power amplifiers transmitted to the speakers (12a, 12b) can be stored (15a, 15b), the inputs of which are the outputs of said memory elements (14a, 14b) connected between D / A converters and the outputs of the power amplifiers (15a, 15b) are connected to said speakers (12a, 12b) connected to said measuring tube (10), that the device comprises FIR filters (16a, 17a; 16b, 17b; 17 ^ 172) to which rechargeable the necessary multiplier sequences and having the microphones (13a, 13b) connected to said measuring tube (10) connected to the input of the FIR filters via A / D converters, the device comprising memory circuits 25 (19a, 20a, 19b, 20b; 18 !, 182), connected to the outputs of said FIR filters, that the device comprises a host processor (18) adapted to control the different operating sequences of the device and to calculate the average flow rate and / or based on the sequences obtained from the latter memory circuits (19a, 20a; 19b, 20b; 181,18z); or the results of volume flow and / or temperature and / or pressure measurements by combining the mass flow rate. Device according to claim 6, characterized in that the device comprises two pairs of FIR filters (16a, 17a, 16b, 17b), of which 35 one pair (16a, 17a) is connected to the second microphone) 13a) and the other pair (16b). 17b) to the second microphone (13b), respectively, that the output of each 882C8 FIR filter is connected to its memory circuit (19a, 20a; 19b, 20b) in which the output of the FIR filters can be stored and that the system host microprocessor (18) is adapted to calculate the number of stored data in said memory circuits. the centers of the peaks in the sequences (Pia> P2aiPibiP2b) 5 and the differences between them and the mutual (L) speeds of sound upstream and downstream of the microphones (13a, 13b) and further the average flow rate and / or volume flow and / or temperature and pressure measurement results by combining mass flow Figure 5). 10 Device according to claim 6, characterized in that the outputs of said microphones (13a, 13b) are connected to a summing or difference amplifier (22), the output of which amplifier (22) is connected to a pair of FIR filters (171,172), one of which a downstream and a countercurrent detector sequence can be stored that each said FIR filter (171,17z) is connected to its memory circuit (18 ^ 182), the other (18 ^) of which can be collected at different times of time by each microphone (13a, 13b). correlation function peaks (Pia.Pib) and to the second (182) countercurrent correlation function peaks 20 (P2biP2a), respectively, of the time differences (tlb-tla; t2a-t2b) of said correlation function peaks, said host processor (18) is arranged to calculate flow rates (18). 16 882C8