阅读说明：本技术 一种油气两相流参数测定装置及方法 (Oil-gas two-phase flow parameter measuring device and method ) 是由 李轶 许卓群 伍国柱 于 2020-03-26 设计创作，主要内容包括：本申请提出了一种油气两相流参数测定装置,包括气相处理组件、气体流量计、油相处理组件、油相流量计和测量组件,所述气相处理组件与所述气体流量计连接,所述油相处理组件与所述油相流量计连接,所述气体流量计与所述测量组件连接,所述油相流量计与所述测量组件连接；所述测量组件包括依次连接的油气混合器、第一电容层析成像传感器、文丘里管和第二电容层析成像传感器,所述气体流量计与所述油气混合器连接,所述油相计量计与所述油气混合器连接；所述测量组件与分离罐连接,所述分离罐与所述油相处理组件连接。使用双电学层析成像(ECT)与文丘里管相结合的测量装置,通过文丘里管前后流型变化预测油气两相流参数。(The application provides an oil-gas two-phase flow parameter measuring device, which comprises a gas phase processing assembly, a gas flowmeter, an oil phase processing assembly, an oil phase flowmeter and a measuring assembly, wherein the gas phase processing assembly is connected with the gas flowmeter, the oil phase processing assembly is connected with the oil phase flowmeter, the gas flowmeter is connected with the measuring assembly, and the oil phase flowmeter is connected with the measuring assembly; the measuring component comprises an oil-gas mixer, a first capacitance tomography sensor, a Venturi tube and a second capacitance tomography sensor which are sequentially connected, the gas flowmeter is connected with the oil-gas mixer, and the oil phase meter is connected with the oil-gas mixer; the measuring component is connected with the separating tank, and the separating tank is connected with the oil phase processing component. And (3) predicting the oil-gas two-phase flow parameters through the flow pattern change before and after the Venturi tube by using a measuring device combining double electrical tomography (ECT) and the Venturi tube.)

1. An oil-gas two-phase flow parameter measuring device is characterized in that: the device comprises a gas phase processing assembly, a gas flowmeter, an oil phase processing assembly, an oil phase flowmeter and a measuring assembly, wherein the gas phase processing assembly is connected with the gas flowmeter, the oil phase processing assembly is connected with the oil phase flowmeter, the gas flowmeter is connected with the measuring assembly, and the oil phase flowmeter is connected with the measuring assembly;

the measuring component comprises an oil-gas mixer, a first capacitance tomography sensor, a Venturi tube and a second capacitance tomography sensor which are sequentially connected, the gas flowmeter is connected with the oil-gas mixer, and the oil phase meter is connected with the oil-gas mixer;

the measuring component is connected with the separating tank, and the separating tank is connected with the oil phase processing component.

2. The oil and gas two-phase flow parameter measurement device of claim 1, wherein: the gas phase treatment component comprises an air compressor, an air tank and a first valve which are connected in sequence, and the first valve is connected with the gas flowmeter.

3. The oil and gas two-phase flow parameter measurement device of claim 1, wherein: the oil phase processing assembly comprises an oil tank, a centrifugal oil pump and a second valve which are connected in sequence, and the second valve is connected with the oil phase flowmeter.

4. The oil and gas two-phase flow parameter measurement device of claim 1, wherein: the first capacitance tomography sensor sequentially comprises a stainless steel pipeline, a metal electrode and an insulation shielding layer from inside to outside, wherein the metal electrode is arranged on the stainless steel pipeline.

5. The oil and gas two-phase flow parameter measurement device of claim 1, wherein: the venturi tube comprises an inlet section, a contraction section, a throat and a diffusion section which are connected in sequence.

6. A method for measuring parameters of oil-gas two-phase flow is characterized by comprising the following steps: the determination method comprises the following steps:

1) designing an experimental scheme, setting different experimental working conditions, acquiring capacitance data of fixed duration under each working condition by using a capacitance tomography sensor, and measuring each phase flow of the oil-gas two-phase flow by using an oil phase flowmeter and a gas flowmeter;

2) collecting capacitance data of fixed duration under each working condition by using the oil-gas two-phase flow parameter measuring device of any one of claims 1 to 5,

3) normalizing the capacitance data in the step 1) and the step 2);

4) carrying out post-processing on the collected double-capacitance tomography capacitance data, and then carrying out image reconstruction;

5) establishing a convolution neural network model, respectively taking a Venturi tube front flow pattern diagram, a Venturi tube rear flow pattern diagram and a Venturi tube front and rear combined flow pattern diagram as the input of a convolution neural network, and predicting the oil content, the gas content and the section gas content under the corresponding states of the flow pattern diagrams;

6) and comparing the oil content, the gas content and the section gas content which are actually measured in the experiment with the oil content, the gas content and the section gas content which are obtained by predicting the convolutional neural network model, and evaluating the accuracy of model prediction.

7. The oil and gas two-phase flow parameter determination method of claim 6, wherein: and 2) reconstructing an image by adopting a linear projection algorithm.

8. The oil and gas two-phase flow parameter determination method of claim 6, wherein: the convolutional neural network model in the step 3) is an initiation-v 3 model, the input of the model is a manifold graph, and the output is the oil content, the gas content and the section gas content obtained by predicting the model.

9. The oil and gas two-phase flow parameter determination method of claim 6, wherein: the convolutional neural network model in the step 3) adopts Elastic Net regression as a loss function.

10. The oil and gas two-phase flow parameter measurement method according to any one of claims 6 to 9, wherein: and predicting the oil content, the gas content and the section gas content in the corresponding state of the flow pattern by using a section gas content measurement algorithm based on the capacitance tomography image and a support vector machine algorithm, and comparing the result with the measurement effect of a convolutional neural network prediction model.

Technical Field

The application belongs to the technical field of oil-gas two-phase flow, and particularly relates to an oil-gas two-phase flow parameter measuring device and method.

Background

In modern society, the oil industry has become the foundation of the world's economy, and how to effectively recover oil is an urgent issue. The accurate measurement of oil and gas two-phase parameters is one of the bottlenecks of oil extraction technology. In a traditional oil and gas two-phase flow parameter measurement method, a multiphase mixture needs to be separated in an oil well to measure a single-phase flow rate. Although the method improves the measurement accuracy of the single-phase flow, the separation process is complex, the equipment is expensive, and the measurement efficiency is low. Therefore, an accurate and efficient multiphase flow online measurement technology needs to be found.

There are many techniques currently in use for flow measurement of two-phase oil and gas flow. The common methods include high-speed camera measurement, wire mesh sensor measurement, radiation attenuation measurement, and fiber optic probe measurement. These methods are feasible in a laboratory environment, but may be limited in their application in a practical working environment given the complexity of the oilfield environment. For example, in the high-speed camera measurement method, the transparent pipe section required for shooting by the camera cannot be realized in practical application, and the wire-mesh sensor required in the wire mesh sensor measurement method and the probe in contact with the fluid required in the optical line probe measurement method are difficult to repair when encountering severe weather in the oil field measurement environment. The ECT sensor has the characteristics of high measuring speed, low application cost, mature product development and the like, and is frequently used in oilfield flow detection. The Venturi tube has the characteristics of accurate measurement, low energy consumption, stable performance, convenient maintenance and the like, and has wide application in the flow detection of single-phase flow and multiphase flow.

In the detection of oil-gas two-phase flow, two-phase interfaces are distributed in different geometric shapes or structural forms, which are called as oil-gas two-phase flow patterns. The definition and classification of the flow pattern of the oil-gas two-phase flow are quite complex, and the common flow patterns at present comprise bubble flow, laminar flow, wavy flow, slug flow and annular flow. However, the flow accuracy of the oil content and the gas content of the oil-gas two-phase flow obtained by the existing measuring device is not high.

Disclosure of Invention

1. Technical problem to be solved

Based on the detection of oil-gas two-phase flow, the two-phase interface is distributed into different geometric shapes or structural forms, which is called as the oil-gas two-phase flow pattern. The definition and classification of the flow pattern of the oil-gas two-phase flow are quite complex, and the common flow patterns at present comprise bubble flow, laminar flow, wavy flow, slug flow and annular flow. But the problem that the flow precision of the oil content and the gas content of the oil-gas two-phase flow obtained by the existing measuring device is not high is solved.

2. Technical scheme

In order to achieve the above object, the present application provides an oil-gas two-phase flow parameter measuring device, which includes a gas phase processing component, a gas flow meter, an oil phase processing component, an oil phase flow meter and a measuring component, wherein the gas phase processing component is connected to the gas flow meter, the oil phase processing component is connected to the oil phase flow meter, the gas flow meter is connected to the measuring component, and the oil phase flow meter is connected to the measuring component;

the measuring component comprises an oil-gas mixer, a first capacitance tomography sensor, a Venturi tube and a second capacitance tomography sensor which are sequentially connected, the gas flowmeter is connected with the oil-gas mixer, and the oil phase meter is connected with the oil-gas mixer;

the measuring component is connected with the separating tank, and the separating tank is connected with the oil phase processing component.

Another embodiment provided by the present application is: the gas phase treatment component comprises an air compressor, an air tank and a first valve which are connected in sequence, and the first valve is connected with the gas flowmeter.

Another embodiment provided by the present application is: the oil phase processing assembly comprises an oil tank, a centrifugal oil pump and a second valve which are connected in sequence, and the second valve is connected with the oil phase flowmeter.

Another embodiment provided by the present application is: the first capacitance tomography sensor sequentially comprises a stainless steel pipeline, a metal electrode and an insulation shielding layer from inside to outside, wherein the metal electrode is arranged on the stainless steel pipeline.

Another embodiment provided by the present application is: the venturi tube comprises an inlet section, a contraction section, a throat and a diffusion section which are connected in sequence.

The application also provides a method for measuring the parameters of the oil-gas two-phase flow, which comprises the following steps:

1) designing an experimental scheme, setting different experimental working conditions, acquiring capacitance data of fixed duration under each working condition by using a capacitance tomography sensor, and measuring each phase flow of the oil-gas two-phase flow by using an oil phase flowmeter and a gas flowmeter;

2) collecting capacitance data of fixed duration under each working condition by using the oil-gas two-phase flow parameter measuring device of any one of claims 1 to 5,

3) normalizing the capacitance data in the step 1) and the step 2);

4) carrying out post-processing on the collected double-capacitance tomography capacitance data, and then carrying out image reconstruction;

5) establishing a convolution neural network model, respectively taking a Venturi tube front flow pattern diagram, a Venturi tube rear flow pattern diagram and a Venturi tube front and rear combined flow pattern diagram as the input of a convolution neural network, and predicting the oil content, the gas content and the section gas content under the corresponding states of the flow pattern diagrams;

6) and comparing the oil content, the gas content and the section gas content which are actually measured in the experiment with the oil content, the gas content and the section gas content which are obtained by predicting the convolutional neural network model, and evaluating the accuracy of model prediction.

Another embodiment provided by the present application is: and 2) reconstructing an image by adopting a linear projection algorithm.

Another embodiment provided by the present application is: the convolutional neural network model in the step 3) is an initiation-v 3 model, the input of the model is a manifold graph, and the output is the oil content, the gas content and the section gas content obtained by predicting the model.

Another embodiment provided by the present application is: the convolutional neural network model in the step 3) adopts Elastic Net regression as a loss function.

Another embodiment provided by the present application is: and predicting the oil content, the gas content and the section gas content in the corresponding state of the flow pattern by using a section gas content measurement algorithm based on the capacitance tomography image and a support vector machine algorithm, and comparing the result with the measurement effect of a convolutional neural network prediction model.

3. Advantageous effects

Compared with the prior art, the oil-gas two-phase flow parameter measuring device and method provided by the application have the beneficial effects that:

the application provides an oil-gas two-phase flow parameter measuring device, which provides a Convolutional Neural Network (CNN) model for realizing nonlinear mapping between oil-gas two-phase flow parameters and a flow pattern diagram, and predicts the oil-gas two-phase flow parameters through flow pattern change before and after a Venturi tube by using a measuring device combining double electrical tomography (ECT) and the Venturi tube.

According to the oil-gas two-phase flow parameter measuring device, when oil-gas two-phase flow flows through the Venturi tube, the flow pattern of the two-phase flow can be changed, and the change of the flow pattern is related to the total flow and the oil-gas ratio. Through the phenomenon, the application provides an experimental scheme of combining the double ECT sensors and the Venturi tube, and flow patterns of oil-gas two-phase flow passing through the front and the back of the Venturi tube are respectively collected.

The oil-gas two-phase flow parameter measuring device provided by the application is used for establishing oil-gas two-phase flow measuring equipment combining a double capacitance tomography (ECT) sensor and a Venturi tube for the first time.

According to the oil-gas two-phase flow parameter measuring method, the relation between the flow pattern change of the oil-gas two-phase flow before and after passing through the Venturi tube and the total flow and the oil-gas ratio is explored, and the oil content and the gas content of the oil-gas two-phase flow are measured.

According to the oil-gas two-phase flow parameter measuring method, a Convolutional Neural Network (CNN) model is established, the oil content and the gas content in the current state are predicted through an oil-gas two-phase flow pattern diagram, and the problem that the traditional algorithm cannot realize nonlinear mapping between high-dimensional image data and low-dimensional flow data is solved.

According to the oil-gas two-phase flow parameter measuring method, the Elastic Net regression is used as a loss function, and the problem of overfitting which may occur is effectively avoided.

Drawings

FIG. 1 is a schematic structural diagram of an oil-gas two-phase flow parameter measuring device of the present application;

FIG. 2 is a schematic diagram of a present capacitance tomography sensor of the present application;

FIG. 3 is a schematic structural view of the venturi of the present application;

FIG. 4 is a schematic diagram of the flow pattern of the oil and gas two-phase flow before and after the venturi tube in a typical GVF of the present application;

FIG. 5 is a graph of oil gas two-phase flow pattern before and after the venturi tube of the present application as a function of GVF;

FIG. 6 is a schematic diagram of the GVF measurement relative error scatter plot of a cross-sectional gas fraction measurement algorithm based on a capacitance tomography image of the present application;

FIG. 7 is a schematic diagram of the GVF measurement relative error scatter plot based on the SVM algorithm of the present application;

FIG. 8 is a schematic diagram of the GVF measurement relative error scatter plot based on the CNN algorithm of the present application;

in the figure: 101-air compressor, 102-air tank, 103-first valve, 104-gas flowmeter, 105-oil tank, 106-centrifugal oil pump, 107-second valve, 108-oil phase flowmeter, 109-oil-gas mixer, 110-first capacitance tomography sensor, 111-venturi tube, 112-second capacitance tomography sensor, 113-separation tank, 201-metal electrode, 202-stainless steel pipeline, 203-insulating shielding layer, 301-inlet section, 302-contraction section, 303-throat channel, 304-diffusion section.

Detailed Description

Hereinafter, specific embodiments of the present application will be described in detail with reference to the accompanying drawings, and it will be apparent to those skilled in the art from this detailed description that the present application can be practiced. Features from different embodiments may be combined to yield new embodiments, or certain features may be substituted for certain embodiments to yield yet further preferred embodiments, without departing from the principles of the present application.

The basic idea of SVM learning is to solve for the separating hyperplane that correctly partitions the training data set and has the largest geometric separation. Separating hyperplanes there are an infinite number of such hyperplanes (i.e., perceptrons) for linearly separable datasets, but the separating hyperplane with the largest geometric separation is unique.

ElasticNet is also called elastic network regression.

Referring to fig. 1 to 8, the present application provides an oil-gas two-phase flow parameter measurement device, including a gas phase processing component, a gas flow meter 104, an oil phase processing component, an oil phase flow meter 108 and a measurement component, where the gas phase processing component is connected to the gas flow meter 104, the oil phase processing component is connected to the oil phase flow meter 108, the gas flow meter 104 is connected to the measurement component, and the oil phase flow meter 108 is connected to the measurement component;

the measuring assembly comprises an oil-gas mixer 109, a first Electric Capacitance Tomography (ECT) sensor 110, a venturi tube 111 and a second electric capacitance tomography sensor 112 which are connected in sequence, the gas flowmeter 104 is connected with the oil-gas mixer 109, and the oil phase meter 108 is connected with the oil-gas mixer 109;

the measuring component is connected with a separation tank 113, and the separation tank 113 is connected with the oil phase processing component.

Further, the gas phase processing assembly comprises an air compressor 101, an air tank 102 and a first valve 103 which are connected in sequence, wherein the first valve 103 is connected with the gas flow meter 104.

Further, the oil phase processing assembly comprises an oil tank 105, a centrifugal oil pump 106 and a second valve 107 which are connected in sequence, wherein the second valve 107 is connected with the oil phase flow meter 108.

Further, the first capacitance tomography sensor 110 sequentially includes a stainless steel pipeline 202, a metal electrode 201 and an insulation shielding layer 203 from inside to outside, and the metal electrode 201 is disposed on the stainless steel pipeline 202.

Further, the venturi tube 111 includes an inlet section 301, a converging section 302, a throat 303, and a diverging section 304 connected in series.

The application also provides a method for measuring the parameters of the oil-gas two-phase flow, which comprises the following steps:

1) designing an experimental scheme, setting different experimental working conditions, acquiring capacitance data of fixed duration under each working condition by using a capacitance tomography sensor, and measuring each phase flow of the oil-gas two-phase flow by using an oil phase flowmeter and a gas flowmeter;

2) collecting capacitance data of fixed duration under each working condition by using the oil-gas two-phase flow parameter measuring device of any one of claims 1 to 5,

3) normalizing the capacitance data in the step 1) and the step 2);

4) carrying out post-processing on the collected double-capacitance tomography capacitance data, and then carrying out image reconstruction;

5) establishing a convolution neural network model, respectively taking a Venturi tube front flow pattern diagram, a Venturi tube rear flow pattern diagram and a Venturi tube front and rear combined flow pattern diagram as the input of a convolution neural network, and predicting the oil content, the gas content and the section gas content under the corresponding states of the flow pattern diagrams;

6) and comparing the oil content, the gas content and the cross-section gas content (GVF) which are actually measured in the experiment with the oil content, the gas content and the cross-section gas content which are obtained by predicting the convolutional neural network model, and evaluating the accuracy of model prediction.

Different conditions here refer to conditions at different total flows and gas-oil ratios. The working pressure of the experiment is 0.6MPa, and the working temperature is 33 ℃. The experimental data acquisition time under each working condition is 10 min.

In step 3), normalization is performed according to the following formula:

in the above formula, C_nIn order to be a normalized capacitance value,C_mfor the actual measured capacitance value, C_gIs the static capacitance value, C, when the pipeline is full of gas phase_oIs the static capacitance value when the pipeline is filled with oil phase.

Further, a linear projection algorithm (LBP) is used for image reconstruction in the step 2).

Further, the convolutional neural network model in the step 3) is an initiation-v 3 model, the input of the model is a manifold, and the output is the oil content, the gas content and the section gas content obtained by predicting the model.

Further, in the convolutional neural network model in the step 3), Elastic Net regression is adopted as a loss function. The Elastic Net regression is adopted as a loss function, compared with the least square regression, the loss function effectively avoids the possible over-fitting problem, and the target function expression is as follows:

in the above equation, e represents the error between the true value and the predicted value, h_ω(x⁽ⁱ⁾) Representing predicted oil and gas contents, y⁽ⁱ⁾Representing the actual oil and gas contents, n representing the number of data sets, lambda₁And λ₂For the regularization parameter, ω represents a vector containing the weights and biases between individual neurons.

And further predicting the oil content, the gas content and the cross-section gas content in the corresponding state of the flow pattern by using a cross-section gas content measurement algorithm based on the capacitance tomography image and a Support Vector Machine (SVM) algorithm, and comparing the measured results with the measurement effect of a convolutional neural network prediction model.

GVF is calculated by a section gas content measuring algorithm based on a capacitance tomography image, a flow pattern diagram before and after a Venturi tube is obtained by a linear projection algorithm (LBP), and a threshold value is set to process the flow pattern diagram. (oil phase) pixel points above the threshold are set to 1, and (gas phase) pixel points below the threshold are set to 0, and the cross-sectional gas fraction is calculated by the following equation:

in the above formula, M is the total number of the cross-sectional pixels, f_jIs the gray value of the jth pixel, A_jIs the area of the jth pixel, and a is the total area of the channel cross-section.

And (3) using an SVM algorithm, wherein the input independent variable x is a manifold graph obtained by using an LBP algorithm, and the output dependent variable y is the oil content, the gas content and the GVF corresponding to the manifold graph, and finally obtaining an optimal regression function between the image high-dimensional data and the feature vector.

In the experiment, besides the capacitance values measured by the two ECT sensors at different times, the flow rates of the two oil phases measured by the gas flowmeter 104 and the oil phase flowmeter 108 are recorded, and the specific detection method is as follows.

The method comprises the following steps: the experimental data under different working conditions are measured by designing an experimental scheme, wherein the different working conditions refer to different total flow and oil-gas ratios. In the experimental process, the oil phase flow is fixed, the gas flow is changed, and the oil content and the gas content in the oil-gas two-phase flow in a certain period of specific time under each working condition are respectively recorded. Wherein the gas fraction under each working condition is obtained according to respective flowmeters of oil and gas phases.

In the formula: GVF means gas fraction, G_gasIs a gas phase flow rate, G_oilThe oil phase flow rate.

The measurement range of the oil content in this application is 1-10m³H, the measurement range of the gas content is 20-150m³The GVF was measured in the range of 0.25 to 0.95.

Step two: an experimental device combining a double ECT sensor and a Venturi tube is adopted to carry out oil-gas two-phase flow phase measurement. And measuring capacitance values of the oil-gas two-phase flow before and after the oil-gas two-phase flow passes through the Venturi tube by using ECT sensors at the front end and the rear end of the Venturi tube respectively. The schematic structure of the ECT sensor is shown in fig. 2, and is composed of 8 electrode plates, and the number of capacitance values per frame is 28. For a sensor consisting of M electrodes, when only one electrode is energized and all other electrodes are held at zero potential, the number of independent capacitances is M (M-1)/2.

The working pressure in the experiment is 0.6MPa, and the working temperature is 33 ℃. The experimental data acquisition time under each working condition is 10 min. In order to ensure the flow stability under each working condition, the data collection of the ECT sensor is not started until the single-phase flow is stable and the oil-gas two-phase flow is mixed in the experimental process. Capacitance data of an Electric Capacitance Tomography (ECT) sensor is acquired through experiments, and before image reconstruction is carried out on the capacitance data, the instability of gas and oil-gas two-phase flow patterns is considered, and the ECT capacitance data of each 100 frames are subjected to average processing, so that stable capacitance data are obtained. Image reconstruction is performed using the averaged processed capacitance data.

Step three: and carrying out normalization processing on the capacitance value under each working condition.

In the above formula, C_nIs a normalized capacitance value, C_mFor the actual measured capacitance value, C_gIs the static capacitance value, C, when the pipeline is full of gas phase_oIs the static capacitance value when the pipeline is filled with oil phase. All the flow charts in fig. 3 and 4 are obtained by the LBP image reconstruction algorithm based on the normalized capacitance value.

Step four: the application uses a linear projection algorithm (LBP) to carry out image reconstruction on a manifold before and after a Venturi tube. ECT image reconstruction is the inverse problem of ECT, i.e. determining the distribution of permittivity within a pipe from the capacitance measurements between pairs of electrodes. The non-linear relationship between the measured capacitance and the dielectric constant can be simplified as follows:

λ＝Sg(3)

where λ is the normalized capacitance vector, S is the normalized sensitive field matrix, and g is the internal distribution matrix of the medium. The normalization method of the capacitance values is shown in step three.

For a two-dimensional field domain, the sensitive field matrix S can be solved by the following formula:

wherein S_i,j(k) Is the sensitivity of the electrode pair to the kth cell,indicating the application of a voltage excitation V to the electrode i_iThe distribution of the electric field strength under the condition that other electrodes are grounded,similarly, τ is the area of grid k. In imaging, it is often necessary to normalize the sensitivity field matrix as follows:

after obtaining the sensitive field matrix S, a formula for calculating the medium distribution matrix can be obtained:

g＝S^-1λ(6)

in most cases, the sensitive field matrix S is irreversible, i.e. S^-1Is absent, so there are many approximate solutions to S^-1The method is called as an image reconstruction algorithm. The linear projection algorithm (LBP) is commonly used in image reconstruction, which is a transpose matrix S using the sensitive field matrix S^TReplacing the inverse of the sensitive field matrix S^-1Calculating a medium distribution matrix g, namely:

g＝S^Tλ(7)

the algorithm principle is simple, the imaging speed is high, and the method is widely applied to ECT imaging. All the manifold maps in fig. 3 and 4 are obtained by the LBP image reconstruction algorithm.

Step five: the method applies an inclusion-V3 model in a CNN algorithm to solve the nonlinear mapping of oil-gas two-phase flow parameters (oil content, gas content and GVF) changing along with the flow pattern. The input of the CNN model is a flow pattern diagram before and after the Venturi tube, and the output is oil content, gas content and GVF under the corresponding state of the flow pattern. The traditional regression algorithm cannot solve the relationship between high-dimensional data (flow pattern image pixels) and low-dimensional data (oil-gas two-phase flow parameters). Compared with a fully connected neural network, the CNN network realizes local communication, weight sharing and down sampling. For the input image pixel information, the algorithm reserves important parameters as much as possible and removes a large number of unimportant parameters, thereby achieving better learning effect.

The forward propagation algorithm in a convolutional neural network can be expressed as:

in the above formula, whereinIs the h th^thMth in layer neural network^thOutput value of individual neuron, k_n ^h-1Is the first (h-1)^thN-th in layer neural network^thOutput of individual neuron, w_mn ^hIs the first (h-1)^thN th of layer^thFrom neuron to h^thM th of layer^thWeight of individual neuron, q_m ^hIs the h th^thMth in layer neural network^thDeviation term for individual neurons. The Relu function is used herein as the activation function.

f(x)＝max(x,0)(9)

Step six: the loss function of the convolutional neural network in this application uses ElasticNet regression, which is a hybrid of Ridge and Lasso regression techniques. Elastic Net regression effectively avoids the over-fitting problem that can occur, as compared to least squares regression. The objective function is shown in equation (10).

In the above formula, e is the true value and predictionError between values, h_ω(x⁽ⁱ⁾) To predict the oil and gas contents, y⁽ⁱ⁾For the oil content and gas content obtained by experimental measurement, n is the number of data sets, lambda₁And λ₂Is a regularization parameter. ω is a vector containing the weights and biases between individual neurons.

Step seven: this application has contrasted the measurement result of venturi tube front flow pattern, venturi tube rear flow pattern, venturi tube front end and the horizontal combination of rear end flow pattern under same operating mode respectively. The measurement result shows that the flow pattern change of the oil-gas two-phase flow passing through the front and the back of the Venturi tube is related to the total flow and the oil-gas ratio in the current state, and the prediction effect of the oil-gas two-phase flow parameter by using the flow pattern diagram transversely combined with the front end and the back end of the Venturi tube is better.

Step eight: the oil content, the gas content and the GVF predicted by the CNN algorithm are compared with the oil content, the gas content and the GVF obtained by real measurement of experiments, the prediction relative error under each working condition is solved, and the prediction accuracy under different flow pattern input and different prediction methods is compared. FIG. 8 shows a GVF measurement relative error scatter diagram based on the CNN algorithm, which represents the prediction accuracy of the CNN algorithm on oil-gas two-phase flow parameters.

Step nine: the method predicts the oil-gas two-phase flow parameters by using a GVF measurement algorithm and an SVM algorithm based on an ECT image, and compares the predicted results with the predicted results of a CNN algorithm. For the section gas content measuring algorithm based on the capacitance tomography image, firstly, a linear projection algorithm (LBP) is used for obtaining a flow pattern diagram before and after a Venturi tube, and a threshold value is set for processing the flow pattern diagram. (oil phase) pixel points above the threshold are set to 1, and (gas phase) pixel points below the threshold are set to 0, and the cross-sectional gas fraction is calculated by the following equation:

in the above formula, M is the total number of the cross-sectional pixels, f_jIs the gray value of the jth pixel, A_jIs the area of the jth pixel, and a is the total area of the channel cross-section.

For the SVM algorithm, the independent variable x input by the model is a flow chart obtained by using an LBP algorithm, the dependent variable y output by the model is the oil content, the gas content and the GVF corresponding to the flow chart, and finally the optimal regression function between the image high-dimensional data and the feature vector is obtained. Fig. 6 and 7 show GVF measurement relative error scatter plots using the cross-sectional air void fraction measurement algorithm based on the electrical capacitance tomography image and the SVM algorithm, representing the measurement accuracy of the GVF measurement algorithm and the SVM algorithm based on the ECT image.

The application provides a convolution neural network model for predicting oil content and gas content in the current state through a manifold. Compared with the traditional algorithm, the convolutional neural network model realizes the nonlinear mapping between the high-dimensional image data and the low-dimensional flow data, and the Elastic Net regression is adopted as a loss function, so that the problem of overfitting which may occur is effectively avoided.

The method establishes the relation between the flow pattern change of the front and back of the Venturi tube and the total flow and the oil-gas ratio, predicts the oil content, the gas content and the gas content of the oil-gas two-phase flow, and greatly improves the measurement precision of the convolutional neural network algorithm by comparing with the traditional cross-section gas content measurement algorithm and SVM algorithm based on capacitance tomography images.

Although the present application has been described above with reference to specific embodiments, those skilled in the art will recognize that many changes may be made in the configuration and details of the present application within the principles and scope of the present application. The scope of protection of the present application is determined by the appended claims, and all changes that come within the meaning and range of equivalency of the claims are intended to be embraced therein.