RU2793366C1 - Intelligent system and method for measuring the flow rate of a two-phase flow of oil wells - Google Patents

Abstract

FIELD: oil industry.

SUBSTANCE: invention relates to the field of measuring the volume and mass flow rate of a liquid and can be used to measure the mass flow rate of a multi-phase flow in oil wells. The technical result is to increase the efficiency of control of multiphase flows by measuring the speed of movement of individual phases of the flow based on the registration of signals from two collimated flows of gamma quanta of a radioisotope density converter. In particular, an intelligent system for measuring the mass flow rate of a two-phase flow of oil wells is claimed, which consists of a gamma radiation source unit capable of producing two collimated 660 keV beams directed at a gamma radiation detector unit containing a NaJ(T1) crystal and a photomultiplier. At the same time, the block of gamma radiation detectors is configured to generate data frames and transfer them to the block of calculations and wavelet transforms. The output of the block of calculations and wavelet transforms is connected to the input of the block of neural networks, with the possibility of transmitting signals received in wavelet spaces. The neural network block is configured to extract data for calculating the fluid flow rate and transfer it to the input of the mass flow calculation block. Also disclosed is a method for measuring the flow rate implemented by the said system.

EFFECT: increase in the efficiency of control of multiphase flows.

5 cl, 7 dwg, 4 tbl

Description

An intelligent system and method for measuring the flow rate of a two-phase flow of oil wells relates to the field of measuring the volume or mass flow of a liquid, in particular for measuring the mass flow of a multi-phase flow in oil wells, where the produced product consists of two different phases (gaseous and liquid), where the device is controlled by a neural network for calibration and data analysis.

Known "Two-phase flowmeter "omega" to measure the flow rate of oil wells" ( RU 62 164 U1 IPC E21B 47/00, G01F 1/00, 27.03.2007) , The invention relates to the field of oil and gas geophysics and can be used to determine the mass flow of oil , gas and water in a multiphase flow passing through a pipeline from a well. The technical result of the invention is to increase the accuracy of determining the flow rate of the gas-liquid mixture (GZhM) and to ensure the protection of maintenance personnel from radioactive radiation. To do this, the volume of the gas component of the multi-phase flow is simultaneously measured using a gas meter and a water-oil phase by means of a multi-phase flow meter (MPF), in which the volume and mass of the water-oil phase are calculated. Moreover, the calculation is carried out by the method of variable pressure drop - differential pressure (DD) that occurs in the narrowed section of the MFR. In this case, the pressure difference in the constriction serves as a measure of the flow rate of the water-oil emulsion, which is related by the following relationship: the greater the pressure difference, the greater the flow rate of the fluid flowing through the MFR. Based on known mathematical expressions, knowing the value of DD, the flux density is determined as the quotient of dividing DD by the square of the flow rate. Then, using the obtained values of the flux density, the volume and, accordingly, the mass of the oil-water phase of the GLS passing through the MFR are calculated. The total consumption of oil, gas and water of the multiphase flow is determined based on the processing of the measurement results in the gas meter and the MFR and the subsequent correction of the results, taking into account correction factors that take into account the composition of the mixture and the ratio between water and oil fractions. This adjustment is carried out according to the results of measuring the capacitance and active conductivity in the MFR channel.

The disadvantage of this method is the need to introduce into a two-phase gas-oil flow, dividing it into gaseous and liquid components, which requires considerable time to control the well in real time.

Also known is the "Multi-phase ultrasonic flow meter for pipelines" (RU 2631495 IPC G01F 1/66, G01F 1/74, G01N 9/00, 10.01.2017).

The system is designed to determine the densities and proportions of the phases in the flow of multi-phase fluid (PMF), which may include the oil phase, the water phase and the gas phase from the well. The system comprises a first density meter that measures PMST at locations where phases of PMST are often separated, a second densitometer that measures PMST at the outlet of the phase mixer-homogenizer, and a third densitometer that measures real-time PMST where the gas phase begins to separate or has separated. from the liquid phase, but where the liquid phases have not separated. The system also includes one or more processors for executing one or more programs to determine oil phase density, water phase density, gas phase density, and phase proportions, including water cut and gas volume fraction, based on readings from the first, second, and third density meters. The technical result is an increase in accuracy and safety.

This measurement method is based on the propagation of ultrasound in a multiphase fluid flow. The disadvantage of the method is the need to be introduced into the flow, the separation of the gas-water-oil flow into phases and the measurement in the selected phases of the flow. The multiphase separation ultrasonic flow meter proposed by the authors performs measurements in different media, which leads to additional measurement errors.

Also known is the method "Measurement of multiphase flow in a pipeline" (RU 2243510 C2 IPC G01F 1/74 (2000.01), G01F 1/56 (2000.01), G01F 1/34 (2000.01), G01N 27/08 (2000.01), 20.07.2003 ), The invention is intended for use in the oil and gas industry to measure the fractional composition and flow rate of phases in the flow of a three-phase mixture under any flow regime. By means of pairs of electrodes placed along the length of the throttling measuring section (case) installed in the pipeline, the electrical conductivity or dielectric constant of the mixture is measured. At the same time, the housing is used to create a differential pressure, which is measured in the annulus between the housing and the pipeline wall, as well as near the housing. By cross-correlation of the signals coming from the electrodes, the phase velocities are determined. Fractional composition is calculated by combining electrical measurements with pressure drop measurements. When combined with the velocity measurements, the volume flow rates of the phases are calculated, and when the latter are subsequently combined with the phase densities, the mass flow rates of the phases are calculated. The inventions take into account the slip between the gas and liquid phases and, thanks to a compact measuring instrument, make it possible to carry out all measurements practically in the same place.

The proposed measurement method is intrusive, violating the flow regime, since pairs of electrodes are installed in the flow, and a pressure drop is artificially created. A change in the electrical properties of the flow, for example, the salinity of the liquid circulating in the pipe, leads to an increase in the measurement errors of various fractions of the flow.

The closest technical solution to the claimed method is "Mass flow meter gas-liquid flow" (RU 2128328 C1 IPC G01F 1/74, G01F 1/86, 03/27/1999). The invention can be used to measure the mass flow rate of the liquid phase of the gas-oil-water flow in the oil industry. The flow meter contains a volumetric flow sensor, a density sensor, a control signal generation circuit, including a unit connected in series for determining the structural function of the flow density change process and a threshold device, a signal transmission unit and an indicator. EFFECT: invention improves the accuracy of calculating the mass flow rate of the liquid phase of the flow in the presence of gas inclusions randomly located in it in the form of small bubbles. 1 ill.

The advantage of this method of measurement is non-contact, non-separation, no change in the flow regime of a multiphase flow. The main disadvantage of the method is the use of the structural function, which depends on the velocities of both small and large bubbles, the velocity of which in a vertical pipe exceeds the velocity of the liquid. This fact increases the measurement error.

To improve the efficiency of oil production process control, it is necessary to perform non-separation, non-contact measurement of multi-phase flows of oil well production. It is proposed to use radioisotope measuring systems that provide a signal that depends on fluctuations in the two-phase flow density. The processing of this signal makes it possible to extract information about the structure of the two-phase flow, the velocity of the liquid and gas, and the relative content of the liquid. Extraction of information about the speed of fluid movement is based on digital processing of a non-stationary signal of a radioisotope converter, including a continuous or discrete non-decimated wavelet transform, an estimate of the cross-correlation of wavelet coefficients in decomposition subspaces.

The aim of the invention is a system and method for measuring the speed and mass flow of individual phases of a two-phase gas-oil flow of oil well production using a radioisotope flux density converter and cross-correlation of two collimated gamma-quantum fluxes in wavelet decomposition spaces.

The technical result is an intelligent system and method for measuring the mass flow rate of oil and gas of a two-phase flow of oil well production, using measurements of the speed of movement of individual phases of a two-phase gas-oil flow based on the registration of signals from two collimated flows of gamma quanta of a radioisotope density converter, using cross-correlation in wavelet spaces -decomposition and pre-trained neural network.

The essence of the invention

The block diagram of the device is shown in Fig.1. The device has a direct radiation source (1), which controls the flux density (2) in the central vertical of the pipe (3). The detector unit (4) perceives gamma quanta and generates pulses, the amplitude of which is proportional to the energy of the detected radiation, and separates the channels of direct and scattered radiation using a threshold. Pulses with an amplitude less than the threshold are assigned to the scattered radiation channel, and more powerful pulses are assigned to the direct radiation channel.

To effectively control multiphase flows, it is necessary to measure the velocities of the liquid and gaseous phases of the flow and the content of each phase of the flow. Measuring the speed of movement of individual phases of a multiphase flow is a more difficult task. To control the production process, it is necessary to measure three separate well flow rates - oil, water and gas. To achieve the goal, an algorithm has been developed for processing sequences of pulses recorded by the detector, implemented in a microprocessor (5).

The measurement is based on the control of the density of the flowing gas-liquid mixture between the radiation source and receiver. Density fluctuations provide information about the flow rate. The presence of a bubble in a liquid causes small random spikes in the flux density signal, the amplitude and scale of which depend on the size and speed of the bubble. Therefore, the motivation for applying the wave approach to the study of the cross-correlation between density fluctuations present in two receiver signals, which is based on a continuous wavelet transform. The method for measuring the velocity of the liquid phase of the flow is used in the dispersed-bubble or annular flow regime based on the calculation of the mutual wavelet-correlation of the signals of radioisotope densitometers.

Cross Wavelet Correlation Based on Discrete Wavelet Transform . The movement of small bubbles generates a correlation that rapidly decreases with time. The movement of large bubbles is stronger and slowly decreasing in time. To isolate signals characterizing the movement of fluid and small bubbles, it is advisable to analyze signals in different ranges, which can be distinguished using wavelet filters.

Wavelet cross-correlation is a scale-localized version of the usual cross-correlation between two signals. Cross-correlation determines the similarity between two sequences at the same level of the wavelet decomposition by shifting one sequence relative to the other, multiplying the shifted sequences, and summing the result.

For deterministic sequences of discrete signals x(n) and y(n) of a radioisotope densitometer, the product is obtained:

– дискретные сигналы x(n) и y(n) на уровне j вейвлет-разложения сигнала, полученные от детекторов, расположенных на базовом расстоянии, n, m – дискретное время, m - переменная запаздывания, которая представляет сдвиг. Для действительных

и

комплексное сопряжение не требуется. Where

are discrete signals x(n) and y(n) at the level j of the wavelet decomposition of the signal obtained from detectors located at the base distance, n, m is discrete time, m is the delay variable that represents the shift. For valid

And

complex conjugation is not required.

Calculation of cross-correlation of signals is based on discrete wavelet transform using Feyer-Korovkin filters. The high pass filter and the low pass filter are shown in Fig. 2, where 6 is the scaling function; 7 - wavelet function; 8- high-frequency filter; 9 – low-frequency wavelet decomposition filter.

Ядро фильтра Фейера имеет вид:Among the wavelet filters used for multiresolution analysis, the Fejer-Korovkin filters have optimal frequency localization. The wavelet is more symmetrical but less smooth than the Daubechies wavelet. The frequency localization of Feyer-Korovkin filters guarantees a sufficiently good representation of the signal in individual frequency bands. Even non-negative kernels generate quadrature mirror filters with optimal frequency localization. The Fejér-Korovkin family forms compact, orthogonal wavelets allowing continuous and discrete wavelet transform, which can have order 4,6,8,14,18,22, and the order must not exceed an integer

The Fejér filter kernel has the form:

Signals x(n), y(n) represent at discrete times n=1…N. Each signal is represented as a linear combination of a scaling function ϕ(x) and a wavelet ψ(x) at various scales j and offsets k :

- количество уровней вейвлет разложения. Первая сумма представляет собой грубое приближение сигнала, а вторая сумма для

представляет детали в последовательных масштабах j. Where

- the number of levels of the wavelet decomposition. The first sum is a rough approximation of the signal, and the second sum for

presents details in successive scales j.

, полученных во фрейме длиной

определяют коэффициенты кросс-корреляции на каждом масштабе.The flow rate measurement must be performed in real time, so the analysis of radioisotope densitometer signals is performed within a frame sliding along the signals. For signals

received in frame length

determine the cross-correlation coefficients on each scale.

и

коэффициентов деталей

. Коэффициенты деталей получают на каждом уровне разложения, а коэффициенты масштаба – только на последнем. Для фрейма данных

получено

и общее число уровней

Total calculated N coefficients

And

detail coefficients

. Detail coefficients are obtained at each decomposition level, while scale factors are obtained only at the last one. For data frame

received

and the total number of levels

, где X - входные данные,

- коэффициенты деталей в пространстве масштаба j, а

- коэффициенты масштабирования на конечном уровня. Обозначим вейвлет-коэффициенты сигналов x и y как

и

.The signal energy is distributed over various scale and wavelet factors so that

, where X is the input data,

- coefficients of details in scale space j, and

- scaling factors at the final level. We denote the wavelet coefficients of signals x and y as

And

.

For locally stationary measuring signals, the Pearson correlation function is determined by the relation:

Using the non-decimated discrete wavelet decomposition at each level s , normalized sample values of the cross-correlation between the wavelet coefficients of two signals are calculated:

The amount of time delay m, at which the cross-correlation between discrete signals is maximum, is the delay characteristic that needs to be estimated. In FIG. 2 shows the Feyer-Korovkin functions and filters that are used to decompose and restore signals. The functions represent a scaling function (6), a wavelet function (7), and decomposition filters - a high pass filter (8) and a low pass filter.

The fluid velocity is determined by the formula:

where D is the distance between the gamma radiation detectors, p is the wavelet decomposition level at which the motion of small bubbles is decisive.

The gas velocity depends on the movement of large bubbles, and is determined in the space of a lower level by the formula:

where q is the level of the wavelet decomposition at which the motion of large bubbles is decisive.

), на вход которого подают значения

, где входной слой принимает номера пространств разложения

, внутренний слой имеет 3 нейрона, выходной слой формирует сигналы

и

, и скорости

.To determine the decomposition levels s and q, a neural network is used - a three-layer perceptron with the number of neurons (

), the input of which is supplied with the values

, where the input layer takes the numbers of the decomposition spaces

, the inner layer has 3 neurons, the output layer generates signals

And

, and speed

.

The density of a mixture of oil and gas is determined by the formula:

– количество гамма-квантов, полученное за время регистрации при градуировке системы на пустой трубе,

- среднее количество гамма-квантов, полученное за время регистрации в процессе измерения,

- количество гамма-квантов, полученное за время регистрации на интервалах, классифицированных как интервалы отсутствия свободного газа.Where

is the number of gamma quanta obtained during the registration time when calibrating the system on an empty pipe,

- the average number of gamma quanta obtained during the registration during the measurement process,

- the number of gamma quanta received during the registration at the intervals classified as the intervals of the absence of free gas.

и

с использованием нейронной сети для определения интервалов отсутствия свободного газа в потоке.Classification of frames for registration

And

using a neural network to determine the intervals of the absence of free gas in the stream.

The density of the liquid is determined at the intervals of the absence of free gas:

Volume fraction of gas in the mixture:

- объемная доля жидкости в смеси.Where

- volume fraction of liquid in the mixture.

Volumetric gas flow - Q _g :

where S is the cross-sectional area of the pipeline,

Fluid volume flow:

Mass flow rate of gas Q _g :

where S is the cross-sectional area of the pipeline,

Mass liquid flow:

A block diagram of an intelligent system for measuring the flow rate of a two-phase flow of oil wells is shown in Fig.3. The scheme includes a block of radiation sources (10), a block of radiation detectors (11), a wavelet transform block (12), a neural network block (13), and an oil and gas flow block (14).

Implementation of the invention

Simulation of measuring the velocities of the flow components. To substantiate the method for measuring the speed of oil movement using wavelet decomposition of the gamma density meter signals, a computer experiment was performed using a simplified model of the movement of a gas-oil mixture in a pipe with a dispersed-bubble flow structure. The oil in the pipe absorbs the gamma quanta coming from the radiation source, the presence of bubbles reduces the absorption. In the process of modeling a gas-liquid mixture in a given section of the pipe, the number of bubbles and their random location in the flow are randomly selected. Bubbles have different diameters, which differ by 16 times. Larger bubbles are located closer to the center of the pipe and move faster than the oil. Smaller bubbles are located closer to the pipe walls and move at a speed equal to the fluid velocity. Therefore, it is small bubbles that are marks that allow you to measure the speed of fluid movement.

The intensity of radioisotope radiation is related to the coefficient of linear attenuation of radiation, the density and thickness of the absorbing layer according to Bouguer's law:

- скорость счета гамма-квантов прямого излучения при градуировке на пустой трубе,

– плотность поглотителя,

- линейный коэффициент поглощения, d - толщина слоя поглотителя. Результаты моделирования содержат шум, распределенной по закону Пуассона. Дисперсия шума равна по величине сигналу I(d). Отношение сигнал/шум в среднем составляет приблизительно

где

- среднеквадратическое отклонение сигнала и шума соответственно. Моделировалось движение мелких пузырей таким образом, что задержка между детекторами равна 0.025 c и движение крупных пузырей с задержкой – 0.075 c. Пунктирной линией показаны доверительный границы для взаимной корреляции. Размер фрейма для 1200 отсчетов.Where

- count rate of gamma rays of direct radiation when calibrating on an empty tube,

is the density of the absorber,

- linear absorption coefficient, d - thickness of the absorber layer. The simulation results contain noise distributed according to the Poisson law. The noise dispersion is equal in magnitude to the signal I(d). The signal-to-noise ratio averages approximately

Where

are the standard deviation of the signal and noise, respectively. The movement of small bubbles was simulated in such a way that the delay between the detectors is 0.025 s and the movement of large bubbles with a delay of 0.075 s. The dotted line shows the confidence limits for cross-correlation. Frame size for 1200 samples.

=0.025 с для мелких пузырей и

=0.075 с. для крупных пузырей. The cross-correlation function of two signals from radioisotope detectors is shown in Fig.4. The calculation shows that there is no any pronounced maximum of the mutual correlation of the two signals. Cross-correlation function in spaces 1, 2 and 4 of the wavelet decomposition are shown in Fig.5. A pronounced maximum is present in spaces 2 and 4. The maxima correspond to the simulated delay

=0.025 s for small bubbles and

=0.075 s. for large bubbles.

и

, показанную в таблице 1.Multiple simulations with different levels of Poisson noise and bubble sizes allowed us to estimate the delay measurement error

And

shown in Table 1.

и

Table 1 - Delay modeling results

And

SNR

0.53 0.021 0.080 0.47 0.020 0.080 1.13 0.025 0.080 1.11 0.025 0.079 0.49 0.020 0.085 -0.23 0.028 0.080 0.67 0.020 0.080

составило

; относительное с.к.о

. Среднеквадратическое отклонение для интервала времени

относительное с.к.о

. Standard deviation (RMS) for the time interval

amounted to

; relative r.m.s.

. Standard deviation for time interval

relative r.m.s.

.

Under real conditions, the flows under study are not stationary. Following the assumption that the system parameters change slowly enough, the cross-correlation is calculated within the frame, the size of which is chosen in this way, based on the number of levels of the wavelet decomposition of the signals.

Block neural network 13 is shown in Fig.3. The neural network is a multilayer perceptron (MLP). From the output of the wavelet transform blocks, the signals are fed to the block of extremal correlators. The delays corresponding to the maxima of the cross-correlation are fed to the block of the neural network.

Первый скрытый слой МСП имеет восемь полно-связных нейронов с активационной функциейThe number of nodes in the input layer is determined by the number of levels of the wavelet decomposition:

The first hidden layer of the MSP has eight fully connected neurons with an activation function

и газа

.where ν is the output signal of the neuron adder, characterizing its local field, e ≈ 2.71828 is a mathematical constant. The output layer contains two neurons that form fluid velocity signals.

and gas

.

MSP training is performed by the Levenberg-Marquard method of error backpropagation over time in real time. Two types of data are used to train a neural network. Data of the first type are obtained by modeling the passage of gamma quanta through matter.

The method for measuring the flow rate of a two-phase flow of oil wells regulates the following actions:

1. The detectors register the number of gamma quanta received during the registration time of 2 ms by two detectors.

2. Form data frames with a size of 1024 samples.

3. Wavelet transform of signal frames is performed according to formulas (3), (4) using filters (2).

4. Calculate the cross-correlation function (7) in the spaces of the wavelet decomposition.

5. Determine the extreme value of the correlation in each scale space of the wavelet decomposition and the velocity of the gas and liquid.

6. Classify data frames using a neural network and determine the density of oil and the average density of oil according to formulas (8) and (9).

7. Calculate the volume fraction of gas and oil in the mixture according to formula (11).

8. Calculate the volumetric and mass flow rate of gas and liquid according to formulas (12) - (15).

Brief description of the figures.

On the figures attached to the description it is given:

Figure 1 shows a block diagram of the device. The device has a direct radiation source (1), which controls the flux density (2) in the central vertical of the pipe (3). The detector unit (4) perceives gamma quanta and generates pulses, the amplitude of which is proportional to the energy of the detected radiation, and separates the channels of direct and scattered radiation using a threshold. Pulses with an amplitude less than the threshold are assigned to the scattered radiation channel, and more powerful pulses are assigned to the direct radiation channel. The processing of pulse sequences recorded by the detector is implemented in software in the microprocessor (5)

Figure 2 shows the Feyer-Korovkin functions and filters that are used to decompose and reconstruct signals. The functions represent a scaling function (6), a wavelet function (7), and decomposition filters - a high-pass filter (8) and a low-pass filter (9).

Figure 3 is a block diagram of an intelligent two-phase flow measurement system for oil wells. The scheme includes a block of radiation sources (10), a block of radiation detectors (11), a wavelet transform block (12), a neural network block (13), and an oil and gas flow block (14).

Figure 4 shows the cross-correlation function of two radioisotope detector signals.

Figure 5 represents the cross-correlation function in scale 1 (15), scale 2 (16) and scale 4 (17) wavelet decomposition of two signals of radioisotope detectors.

Figure 6 shows the mutual wavelet correlation between the signals of two radiation detectors at an oil flow rate of 0.56 m/s.

Figure 7. Mutual wavelet correlation between the signals of two radiation detectors at an oil flow rate of 0.28 m/s.

Example: A radiation source emits a collimated beam of radiation. Radiation source - radionuclide Cesium-137 (Cs'137). Standard units for generating γ-radiation (BGI-60A) make it possible to obtain collimated beams of γ-radiation with small scattering angles.

During the passage of γ-radiation with an energy of 400-800 kEV through a pipe filled with oil, the interaction of γ-radiation with the atoms of matter occurs. During this γ-quantum interaction, either Compton scattering (loss of part of the energy) or photoelectric absorption (loss of all energy) occurs. Accordingly, there was a weakening of the intensity of the γ-radiation flux due to photoelectric absorption and the release of γ-quanta from the controlled region due to Compton scattering. To register γ-radiation, we used a NaI(Tl) crystal, a photomultiplier, an electronic unit, and a filter tuned to receive γ-quanta with an energy of 660 keV.

In the present example, more than 500,000 measurements were obtained on an empty pipe, on a pipe completely filled with oil, and on a stream of oil. The registration time was 4ms. Tables 2-4 show the first 10 results of registration of gamma quanta registered by the radiation detector.

Table 2 - Results of registration of γ-quanta on an empty pipe

No. 1 2 3 4 5 6 7 8 9 10 Nemp 297 273 280 279 298 281 300 301 275 277 No. eleven 12 13 14 15 16 17 18 19 20 Nemp 287 325 316 286 270 268 302 321 283 271

Table 3 - Results of registration of γ-quanta on a filled pipe

No. 1 2 3 4 5 6 7 8 9 10 Nfull 305 294 293 287 276 321 286 287 286 280 No. eleven 12 13 14 15 16 17 18 19 20 Nfull 257 284 298 286 272 280 321 294 324 280

Table 4 - Results of registration of γ-quanta on the oil flow in the pipe

No. 1 2 3 4 5 6 7 8 9 10 Nflow 171 181 190 173 187 176 160 193 172 204 No. eleven 12 13 14 15 16 17 18 19 20 Nflow 183 172 171 200 189 179 194 175 172 168

, коэффициент линейного поглощения нефти

, диаметр трубы 0.26 м, период регистрации

. Базовое расстояние между приемниками излучения составило 0.5 м. Для вычисления корреляции результаты измерения были разделены на фреймы длительностью 4.1 с, содержащие 1024 отчета. Oil Density:

,oil absorption coefficient

,pipe diameter 0.26 m, registration period

. The basic distance between the radiation receivers was 0.5 m. To calculate the correlation, the measurement results were divided into 4.1 s frames containing 1024 reports.

The first experiment was carried out with the oil flow moving at a speed of 0.56 m/s. Figures 6 and 7 show the correlation values between the wavelet coefficients of the signals of two radiation receivers. On the scales of the wavelet decomposition, which are responsible for the fine details of the signal (corresponding to the movement of small bubbles), the delay value was determined from the correlation maximum. In figure 6, at the maximum correlation value of 0.98, the delay was 0.282 s (the actual delay value is 0.28 s). In figure 7, at the maximum correlation value of 0.98, the delay was 0.142 s (the actual delay value is 0.14 s).

The volume flow is as follows:

Mass flow of oil :

.

.

Claims (5)

1. An intelligent system for measuring the mass flow rate of a two-phase flow of oil wells, characterized in that it consists of a gamma ray source unit configured to produce two 660 keV collimated beams directed to a gamma ray detector unit containing a NaJ(T1) crystal and a photomultiplier , wherein the block of gamma radiation detectors is configured to generate data frames and transmit them to the block of calculations and wavelet transforms, the output of the block of calculations and wavelet transforms is connected to the input of the block of neural networks, with the possibility of transmitting signals received in wavelet spaces, the neural network block is configured to extract data for calculating the fluid flow rate and transmit them to the input of the mass flow calculation block. 2. An intelligent system according to claim 1, characterized in that the block for calculating and wavelet-transforming frames of signals from two gamma radiation detectors and the block of neural networks are configured to determine the space of wavelet decomposition and classify frames for registering the average number of gamma quanta, obtained during the registration time in the measurement process and the number of gamma quanta obtained during the registration time at intervals classified as intervals of the absence of free gas. 3. The method of intelligent flow measurement, implemented by the system according to claim 1, characterized in that it includes the steps, which: emit through the gamma radiation unit two collimated beams of 660 keV; register the number of pulses of gamma quanta using a block of detectors for a given registration time; forming data frames, performing a wavelet transform of the data frames to form wavelet decomposition spaces, and calculating a cross-correlation function in the wavelet decomposition spaces using Feyer-Korovkin filters by means of a calculation block and wavelet transforms; determine the extreme value of the correlation in each scale space of the wavelet decomposition, the time of movement of oil and the speed of movement of gas and liquid; classifying the data frames using the neural network and determining the density of the mixture of oil and gas and the density of the oil, taking into account the movement of small gas bubbles in the stream; determine the volume fraction of gas in the mixture; and calculating the volumetric and mass flow rate of the liquid in the flow calculation unit. 4. The method according to claim 3, characterized in that small gas bubbles are used as marks of the movement of the liquid phase of well production, and the determination of the time of oil movement is carried out by the maximum of cross-correlation in the wavelet decomposition space corresponding to the movement of small gas bubbles. 5. The method according to claim 3, characterized in that the neural network block is used to determine the number of the wavelet decomposition space, which is used in determining the time of movement of the liquid phase of the flow and classifying frames and searching for frames to measure the average density of the liquid.

Images