Conductivity and flow pattern background prediction method, image reconstruction method, equipment and medium

Abstract

The present invention provides a conductivity and flow pattern background prediction method, image reconstruction method, device and medium, the method comprising: obtaining first capacitance spectrum data when the pipeline is full of air and second capacitance spectrum data of the current fluid in the pipeline; inputting capacitance difference spectrum data of the first capacitance spectrum data and the second capacitance spectrum data into a pre-built neural network, and outputting the flow pattern background and conductivity of the current fluid in the pipeline. The scheme can simplify the system operation steps and will not be affected by errors caused by environmental factors.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to a conductivity and flow pattern background prediction method, an image reconstruction method, a device and a medium. Background Art

[0002] As one of the current research hotspots of process tomography (PT), Electrical Resistance Tomography (ERT) technology has broad application prospects in many industrial fields such as metallurgy, petroleum, papermaking, and environmental engineering. Compared with other PT technologies such as nuclear magnetic resonance, it has the advantages of low cost, simple structure, good safety, and easy operation.

[0003] The ERT technology is used to obtain the conductivity distribution in the measurement area, thereby indirectly measuring the conductivity. According to the sensitive field back projection algorithm, it can be obtained:

[0004]

[0005] in, represents the conductivity distribution in the measurement area, b is the number of sub-domains into which the measurement area is divided; σ0 is the conductivity reference value when the initial pipe is filled with water; is the sensitive field matrix, where a is the number of independently measured electrical impedances; Respectively represent the fluid impedance values ​​at the time of measurement and initial reference; · / represents the division of the corresponding element values.

[0006] In the above method, in order to obtain the R of the sensor when the pipe is full of water r , the ERT system needs to be filled with water before operation and record R r However, when the system is running, if environmental factors such as temperature change, R needs to be updated. r The value means that the system needs to be stopped and refilled with water for measurement. However, in actual situations, the system is not allowed to stop frequently after it is running, which will cause measurement errors. Summary of the invention

[0007] The purpose of the embodiments of this specification is to provide a conductivity and flow pattern background prediction method, image reconstruction method, device and medium.

[0008] To solve the above technical problems, the embodiments of the present application are implemented in the following ways:

[0009] In a first aspect, the present application provides a conductivity and flow pattern background prediction method, the method comprising:

[0010] Acquire first capacitance spectrum data when the pipeline is filled with air and second capacitance spectrum data of the current fluid in the pipeline;

[0011] The capacitance difference spectrum data between the first capacitance spectrum data and the second capacitance spectrum data is input into a pre-built neural network, and the flow pattern background and conductivity of the current fluid in the pipeline are output.

[0012] In one embodiment, the first capacitance spectrum data and / or the second capacitance spectrum data are complex-valued capacitance spectrum data measured by a complex-valued multi-frequency capacitance tomography device.

[0013] In one embodiment, the complex-valued multi-frequency capacitance tomography device includes a plurality of electrode sensors, and the plurality of electrode sensors are disposed on an outer wall of a pipe.

[0014] In one embodiment, the pre-built neural network adopts an improved ResNet neural network.

[0015] In one embodiment, inputting capacitance difference spectrum data of the first capacitance spectrum data and the second capacitance spectrum data into a pre-built neural network comprises:

[0016] Normalizing the first capacitance spectrum data and the second capacitance spectrum data to obtain first normalized capacitance spectrum data and second normalized capacitance spectrum data;

[0017] The first normalized capacitance spectrum data and the normalized capacitance difference spectrum data of the second normalized capacitance spectrum data are input into the pre-built neural network.

[0018] In a second aspect, the present application provides an image reconstruction method, the method comprising:

[0019] According to the flow pattern background and conductivity of the current fluid predicted by the conductivity and flow pattern background prediction method of the first aspect, the capacitance spectrum data of the predicted flow pattern background are selected as linear points for image reconstruction.

[0020] In one of the embodiments, the predicted current fluid flow pattern background is air;

[0021] The capacitance spectrum data of the predicted flow pattern background is selected as the linear point for image reconstruction, including:

[0022] The first capacitance spectrum data is selected as a linear point for image reconstruction.

[0023] In one embodiment, the predicted current fluid flow pattern background is water;

[0024] The capacitance spectrum data of the predicted flow pattern background is selected as the linear point for image reconstruction, including:

[0025] According to the predicted conductivity, obtaining third capacitance spectrum data corresponding to the conductivity when the pipeline is filled with water;

[0026] The third capacitance spectrum data is selected as a linear point for image reconstruction.

[0027] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the conductivity and flow pattern background prediction method of the first aspect and / or the image reconstruction method of the second aspect are implemented.

[0028] In a fourth aspect, the present application provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the conductivity and flow pattern background prediction method of the first aspect and / or the image reconstruction method of the second aspect.

[0029] It can be seen from the technical solution provided in the above embodiments of this specification that the solution can predict the flow pattern background and conductivity of the current fluid in the pipeline by acquiring the first capacitance spectrum data when the pipeline is filled with air and the second capacitance spectrum data of the current fluid in the pipeline. There is no need to collect the capacitance spectrum data when the pipeline is filled with water, which can simplify the system operation steps and will not be affected by errors caused by environmental factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1 A schematic diagram of the flow chart of the conductivity and flow pattern background prediction method provided in this application;

[0032] FIG2 is a schematic diagram of the distribution of four typical gas-water two-phase flow patterns provided in this application;

[0033] FIG3 is a capacitance difference spectrum diagram obtained for the four flow patterns corresponding to FIG2 ;

[0034] Figure 4 A schematic diagram of the framework of the pre-built neural network provided for this application;

[0035] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0037] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0038] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present application description without departing from the scope or spirit of the present application. Other embodiments derived from the present application description will be apparent to those skilled in the art. The present application description and examples are exemplary only.

[0039] The words “include,” “including,” “have,” “contain,” etc. used in this document are open-ended terms, meaning including but not limited to.

[0040] Unless otherwise specified, "parts" in this application are all calculated by mass.

[0041] In the related art, the conductivity distribution in the measurement area is obtained by using ERT technology, so as to indirectly measure the conductivity. In order to obtain the R value of the sensor when the pipe is full of water, r , the ERT system needs to be filled with water before operation and record R r However, when the system is running, if environmental factors such as temperature change, R needs to be updated. r The value means that the system needs to be stopped and refilled with water for measurement. However, in actual situations, the system is not allowed to stop frequently after it is running, which will cause measurement errors.

[0042] Based on the above defects, the present application proposes a conductivity and flow pattern background prediction method, which does not need to collect the impedance value when the pipeline is full of water, thereby reducing the error caused by the drift of the impedance value when the pipeline is full of water.

[0043] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0044] Reference Figure 1 , which shows a flow chart of the conductivity and flow pattern background prediction method applicable to the embodiment of the present application.

[0045] like Figure 1 As shown, the conductivity and flow pattern background prediction method may include:

[0046] S110, acquiring first capacitance spectrum data when the pipeline is filled with air and second capacitance spectrum data of the current fluid in the pipeline;

[0047] S120, inputting capacitance difference spectrum data between the first capacitance spectrum data and the second capacitance spectrum data into a pre-built neural network, and outputting the flow pattern background and conductivity of the current fluid in the pipeline.

[0048] Optionally, the first capacitance spectrum data and / or the second capacitance spectrum data are complex-valued capacitance spectrum data measured by a complex-valued multi-frequency capacitance tomography device. The complex-valued multi-frequency capacitance tomography device includes a plurality of electrode sensors, and the plurality of electrode sensors are arranged on the outer wall of the pipeline.

[0049] Optionally, inputting capacitance difference spectrum data between the first capacitance spectrum data and the second capacitance spectrum data into a pre-built neural network includes:

[0050] Normalizing the first capacitance spectrum data and the second capacitance spectrum data to obtain first normalized capacitance spectrum data and second normalized capacitance spectrum data;

[0051] The first normalized capacitance spectrum data and the normalized capacitance difference spectrum data of the second normalized capacitance spectrum data are input into the pre-built neural network.

[0052] Specifically, complex-valued multi-frequency electrical capacitance tomography (CVMF-ECT) is a non-invasive, non-contact and radiation-free emerging imaging technology that can be used to reconstruct the complex dielectric constant distribution of multiphase flow. Due to the use of complex-valued measurement technology, CVMF-ECT can effectively measure the complex-valued multi-frequency capacitance data of multiphase flow containing conductive phases, that is, the complex-valued capacitance spectrum. The complex-valued capacitance spectrum contains more information than traditional capacitance data, such as the trend of capacitance value changing with frequency and the relative size relationship between the real part and the imaginary part of capacitance, which broadens the application scope of capacitance tomography technology.

[0053] It is understandable that the complex-valued multi-frequency capacitance tomography device includes a plurality of electrode sensors, and the plurality of electrode sensors can be arranged at equal intervals on the outer wall of the pipeline. As shown in Figure 2, the complex-valued multi-frequency capacitance tomography device uses 8 electrode sensors, represented by 1, 2, ..., 8 respectively. Each electrode sensor is also connected to a measuring circuit for measuring the complex-valued capacitance between any two electrode sensors, and obtaining the complex-valued capacitance within a set frequency range to obtain capacitance spectrum data, which can be sent to the execution subject of the conductivity and flow pattern background prediction method. It is understandable that the first capacitance spectrum data when the pipeline is full of air can be pre-stored in an independent or component with a storage function such as a storage medium or a storage device. The execution subject of the conductivity and flow pattern background prediction method reads it from a storage medium or a storage device.

[0054] According to electromagnetic theory, the complex capacitance C measured by CVMF-ECT can be calculated by formula (1):

[0055]

[0056] Where V is the potential difference between the electrode pairs of the electrode sensor; ε(x, y) and σ(x, y) are the dielectric constant distribution and conductivity distribution of the measurement area, respectively; f is the circuit excitation frequency; Γ is the integrated area; φ(x, y) is the potential distribution; S is the area of ​​the electrode sensor; is the normal vector of the electrode surface of the electrode sensor.

[0057] As shown in Figure 2 (including Figure 2a , 2b , 2c, 2d) show four typical gas-water two-phase flow patterns. Figure 2a Shown are flow pattern distributions s1: single rod with water background; Figure 2b Shown are flow pattern distributions s2: water background double bars; Figure 2c Shown are flow pattern distributions s3: single rod with air background; Figure 2d The flow pattern distribution s4 is shown: air background double rod. The CVMF-ECT device in Figure 2 uses an 8-electrode sensor, which is represented by 1, 2, ..., 8 (where Figure 2b , 2c , 2d, the numbers of the electrode sensors are not marked). In Figure 2, the outer diameter of the pipe is 60 mm and the inner diameter is 56 mm; the diameter of the single rod under the water background and the air background is 24 mm, located in the center of the measurement area; the diameter of the double rod is 15 mm, close to electrode 1 and electrode 4 respectively.

[0058] Since the four two-phase flows in Figure 2 all contain conductive water, it can be seen from equation (1) that the complex capacitance measured by CVMF-ECT will change with frequency. Therefore, the capacitance difference spectrum obtained by these four flow types can be obtained, as shown in Figure 3 (including Figure 3a ,3b , 3c, 3d), wherein, Figure 3a , 3b , 3c, and 3d are Figure 2a , 2b , 2c, and 2d corresponding to the capacitance difference spectrum. air is the capacitance measurement value when the pipe is filled with air (i.e., the first capacitance spectrum data), C s1 , C s2 , C s3 and C s4 They are the capacitance measurement values ​​of the four flow types in Figure 3 (i.e., the second capacitance spectrum data); the horizontal axis is σ / f, with a range of 10 -5 -10 1 S / (m·MHz); the left vertical axis represents the real part of the complex capacitance difference, and the right vertical axis represents the imaginary part of the complex capacitance difference. Since an 8-electrode sensor is used for measurement, there are 28 independent complex capacitance measurement values, which are represented by ΔC r and ΔC i Here we select the capacitance difference between electrodes 1, 2 and electrodes 1, 5, that is, and Analyze for representatives.

[0059] During the measurement, the excitation frequency was set from 0.1 MHz to 1 MHz with a step size of 0.1 MHz, so the conductivity range reflected in Figure 3 is 10 -5 -1S / m. When the value of σ / f is too small (<10 -4 ) or too large (>0.1), ΔC r and ΔC i The conductivity of the sample is almost unchanged, so it is difficult to distinguish different conductivities based on the spectrum at this time. The following uses the spectrum corresponding to 0.001S / m and 0.01S / m as examples to illustrate that conductivity can be distinguished by spectrum data.

[0060] In Figure 3, the shaded area represents the spectrum at a conductivity of 0.001 S / m. Figure 3a As σ / f increases, It decreases slowly at the beginning, and then decreases rapidly; It kept increasing until it reached its highest point and then fell back slightly; is an accelerated increase; Decreases to a minimum value and then increases slightly. Figure 3b It can be seen that the ΔC of the flow pattern distribution s2 with the same water background r and ΔC i and Figure 3a In the air background, Figure 3c and Figure 3dThe overall variation trend of capacitance in the medium is different from that in the water background: and The trend of change is similar to that in the water background, but slower; and The change is very small. The above is the characteristic trend of the spectrum of each flow type distribution when the conductivity is 0.001S / m. Based on these characteristic trends that only exist when the conductivity is 0.001S / m, this conductivity can be identified from the measured spectrum. Similarly, analyzing the spectrum at 0.01S / m represented by the gray shaded area, it can be found that the change trend of capacitance at this time becomes very different: as σ / f increases, for all four distributions, It decreases rapidly at the beginning and then becomes stable; has been decreasing; Figure 3a , 3b The corresponding water background distribution, The trend is to increase rapidly; Figure 3c , 3d The corresponding air background distribution, The change trend of the four distributions is similar to that of the water background, but more gentle. Trends and their respective The trends are similar.

[0061] From the above analysis, it can be seen that no matter the air background or water background distribution, the spectrum diagram at 0.001S / m conductivity and that at 0.01S / m have different characteristic trends, and these characteristic trends can help identify different conductivities.

[0062] In one embodiment, the pre-built neural network may adopt an improved ResNet neural network. The pre-built neural network framework is as follows: Figure 4 As shown in the figure, the improved ResNet neural network can effectively extract the information of the capacitance spectrum data and accurately predict the background and conductivity.

[0063] The input of the pre-built neural network is the capacitance difference spectrum data (or normalized capacitance difference spectrum data), with a dimension of 10×56; the convolution kernel size of the first convolution is 7×7, and the number of output channels is 16; the convolution kernel size in the first to fourth layers of the convolution group is 3×3, and the number of output channels is 16, 32, 64, 128, respectively, and the step size is 1×1, 1×2, 1×2, 1×2, respectively. The fully connected layer output of the background prediction branch is 1×2, where the position of the maximum value is the final prediction result of the background, and the result is 1 for the prediction of air background, and 2 for water background. It can be understood that the fully connected layer output of the background prediction branch is determined according to the actual predicted number of flow pattern backgrounds, that is, when the predicted number of flow pattern backgrounds is 3, the fully connected layer output of the background prediction branch is 1×3. The output of the fully connected layer of the conductivity prediction branch is 1×4, where the maximum value is the final prediction result of the conductivity, and 1-4 respectively represent the predicted conductivity of 0.001S / m, 0.005S / m, 0.01S / m or 0.05S / m. It can be understood that the output of the fully connected layer of the conductivity prediction branch is determined according to the actual predicted number of conductivities, that is, when the predicted number of conductivities is 6, the output of the fully connected layer of the conductivity prediction branch is 1×6.

[0064] The loss function of each branch during training uses the cross entropy loss function:

[0065]

[0066] The loss function of the background prediction branch is L b , the loss function of the conductivity prediction branch is L c .

[0067] Among them, x is the prediction result of the pre-built neural network, k is the true value, and K is the number of categories. For example, L for the background prediction branch b K = 2 in the conductivity prediction branch. c K=4 in .

[0068] The total loss function used by the pre-built neural network during training is L cp :

[0069] L cp =λ1L b +λ2L c (3)

[0070] Among them, λ1 and λ2 are the weights of the loss function of the background prediction branch and the loss function of the conductivity prediction branch, respectively, which can be set according to actual needs. For example, λ1 is 0.1 and λ2 is 0.9.

[0071] The data used for training can be obtained from simulation, including single and double air rods with different diameters and positions under water background, single and double water rods with different diameters and positions under air background, and capacitance spectrum data of laminar flow with different heights and slopes (laminar flow is set as water background) at four conductivity levels of 0.001S / m, 0.005S / m, 0.01S / m and 0.05S / m. After data enhancement by adding noise, there are 42,000 sets of training data. The training was performed using the Pytorch framework on two NVIDIA 1070Ti GPUs, and the Adam optimizer was used for 15 rounds of optimization with a learning rate of 5e-5.

[0072] The actual effect of this application is verified by experiments. The experimental flow types are shown in the first column of Table 1, with a total of 7 types, among which gray is water and white is air. The experimental data was collected by Keysight Technologies' E4990A model impedance analyzer, and the final conductivity prediction results are listed in Table 1. Through the method invented in this article, the accuracy of background prediction reached 100%, and the accuracy of conductivity prediction under water background also reached 100%; the accuracy of conductivity prediction under air background reached 87.5%. However, even if the conductivity prediction under air background is wrong, it will not affect the subsequent imaging results.

[0073] Table 1 Experimental data conductivity prediction results

[0074]

[0075]

[0076] The embodiment of the present application can predict the flow pattern background and conductivity of the current fluid in the pipeline by acquiring the first capacitance spectrum data when the pipeline is filled with air and the second capacitance spectrum data of the current fluid in the pipeline. There is no need to collect the capacitance spectrum data when the pipeline is filled with water, which can simplify the system operation steps and will not be affected by errors caused by environmental factors.

[0077] The present application pre-builds a neural network and directly uses the complex-valued capacitance spectrum data measured by the CVMF-ECT device to predict conductivity without the need to install an additional conductivity sensor.

[0078] This application identifies the conductivity and flow pattern background based on the characteristic trend of complex capacitance spectrum data at different conductivities. The complex capacitance spectrum data is collected by the CVMF-ECT system, which does not require contact with the fluid, and achieves non-contact and non-invasive measurement of conductivity. And from the experimental results in Table 1, the background of flow pattern distribution can be accurately determined and the conductivity size can be predicted.

[0079] In order to simplify the calculation, CVMF-ECT uses a linearization method to solve the imaging problem and divides the complex value into two parts: the real part and the imaginary part. By introducing linear points, the forward problem of CVMF-ECT can be constructed as:

[0080]

[0081] In the formula, are the changes in the real and imaginary parts of the complex-valued capacitance due to the perturbation of the complex dielectric constant; are the changes in the real and imaginary parts of the complex dielectric constant from the linear point perturbation, where Δε r Represents the change in dielectric constant; is Δε r , Δε i Mapping to ΔC r The Jacobian matrix of is Δε r , Δε i Mapping to ΔC i The Jacobian matrix of the CVMF-ECT sensor. The superscript m is determined by the number of electrode sensors N of the CVMF-ECT sensor. For an imaging system composed of N electrode sensors, the number of independently measured capacitance values ​​is m = N (N-1) / 2. The superscript n is the number of pixels for imaging the dielectric constant of the measurement area. To simplify writing, equation (4) can be written as:

[0082] C=JG (5)

[0083] In the formula,

[0084] By solving the inverse problem of formula (5), the medium distribution image reconstructed by CVMF-ECT can be obtained. Usually, the inverse problem of formula (5) can be solved by expressing it as the following optimization problem:

[0085]

[0086] stC=JG (6)

[0087] Where l(·) and represent the regularization function and regularization parameter respectively.

[0088] In the linearization method, the selection of linear points is crucial. If the linear points are not selected properly, the reconstructed image will be severely distorted. The linear points are set based on the flow pattern background and the conductivity of the water. Based on the above conductivity and flow pattern background prediction method, the flow pattern background and conductivity are predicted, and appropriate linear points can be selected.

[0089] The present application also proposes an image reconstruction method based on the conductivity and flow pattern background predicted by the above-mentioned conductivity and flow pattern background prediction method.

[0090] The image reconstruction method may include: selecting capacitance spectrum data of the predicted flow pattern background as linear points for image reconstruction according to the flow pattern background and conductivity of the current fluid predicted by the above-mentioned conductivity and flow pattern background prediction method.

[0091] Optionally, the predicted flow pattern background of the current fluid is air;

[0092] The capacitance spectrum data of the predicted flow pattern background is selected as the linear point for image reconstruction, including:

[0093] The first capacitance spectrum data is selected as a linear point for image reconstruction.

[0094] Optionally, the predicted flow pattern background of the current fluid is water;

[0095] The capacitance spectrum data of the predicted flow pattern background is selected as the linear point for image reconstruction, including:

[0096] According to the predicted conductivity, obtaining third capacitance spectrum data corresponding to the conductivity when the pipeline is filled with water;

[0097] The third capacitance spectrum data is selected as a linear point for image reconstruction.

[0098] After background prediction and conductivity prediction, the complex capacitance data at 0.5 MHz excitation frequency was selected for imaging using the Tikhonov algorithm. The results are listed in Table 2. The gray area represents water, and the pixel value is set to 1; the white area represents air, and the pixel value is set to 0. Δε represents the reconstructed dielectric constant distribution, and Δσ represents the reconstructed conductivity distribution.

[0099] Table 2 Experimental data imaging results

[0100]

[0101] It can be seen that when the linear point selection is inappropriate, that is, imaging with air linear points under water background distribution or imaging with water linear points under air background distribution, the imaging will have serious artifacts and the true distribution cannot be identified. However, after the air background and conductivity are predicted by the conductivity and flow pattern background prediction method of the present application, suitable linear points can be provided for each flow pattern distribution, which greatly improves the imaging quality and reduces artifacts.

[0102] The image reconstruction method of this embodiment is based on the conductivity and flow pattern background prediction method. It can accurately predict the selected conductivity and flow pattern background in real time without contacting the fluid, providing suitable linear points for image reconstruction, thereby improving the quality of the reconstructed image.

[0103] Figure 5FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 5 As shown, a structural schematic diagram of an electronic device 500 suitable for implementing an embodiment of the present application is shown.

[0104] like Figure 5 As shown, the electronic device 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 are also stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0105] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 506 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage section 508 as needed.

[0106] In particular, according to the embodiments of the present disclosure, the above reference Figure 1 The described process can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes program code for executing the above-mentioned conductivity and flow pattern background prediction method and / or image reconstruction method. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 509, and / or installed from the removable medium 511.

[0107] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of the code, and the aforementioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0108] The units or modules involved in the embodiments described in the present application may be implemented by software or hardware. The units or modules described may also be arranged in a processor. The names of these units or modules do not constitute limitations on the units or modules themselves in certain circumstances.

[0109] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop, a mobile phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0110] As another aspect, the present application further provides a storage medium, which may be a storage medium included in the aforementioned device in the above embodiment; or a storage medium that exists independently and is not assembled into the device. The storage medium stores one or more programs, and the aforementioned programs are used by one or more processors to execute the conductivity and flow pattern background prediction method and / or image reconstruction method described in the present application.

[0111] Storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0112] It should be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0113] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

Claims

1. A conductivity and flow pattern background prediction method, characterized in that: The method comprises: Acquire first capacitance spectrum data when the pipeline is filled with air and second capacitance spectrum data of the current fluid in the pipeline; Normalizing the first capacitance spectrum data and the second capacitance spectrum data to obtain first normalized capacitance spectrum data and second normalized capacitance spectrum data, wherein the first capacitance spectrum data and the second capacitance spectrum data are complex-valued capacitance spectrum data measured by a complex-valued multi-frequency capacitance tomography device, and the complex-valued multi-frequency capacitance tomography device includes a plurality of electrode sensors, and the plurality of electrode sensors are arranged on the outer wall of the pipeline; Inputting the normalized capacitance difference spectrum data of the first normalized capacitance spectrum data and the second normalized capacitance spectrum data into a pre-built neural network, and outputting the flow pattern background and conductivity of the current fluid in the pipeline, wherein the pre-built neural network adopts an improved ResNet neural network; The input of the pre-built neural network is the normalized capacitance difference spectrum data with a dimension of 10×56. The convolution kernel size of the first convolution is 7×7, and the number of output channels is 16. The convolution kernel size in the first to fourth layers of the convolution group is 3×3, and the number of output channels is 16, 32, 64, and 128, respectively, with a step size of 1×1, 1×2, 1×2, and 1×2, respectively. The fully connected layer output of the background prediction branch is 1×2, where the maximum value is the final prediction result of the background. The result 1 represents the predicted air background, and 2 represents the water background. The fully connected layer output of the background prediction branch is determined according to the actual predicted flow type background number. The output of the fully connected layer of the conductivity prediction branch is 1×4, where the position of the maximum value is the final prediction result of the conductivity, and 1-4 respectively represent the predicted conductivity of 0.001S / m, 0.005S / m, 0.01S / m or 0.05S / m. The output of the fully connected layer of the conductivity prediction branch is determined according to the actual predicted conductivity number, that is, when the predicted conductivity number is 6, the output of the fully connected layer of the conductivity prediction branch is 1×6; The loss function of each branch during training uses the cross entropy loss function: Where x is the prediction result of the pre-built neural network, k is the true value, and K is the number of categories; The total loss function used by the pre-built neural network during training is L cp : L cp =λ1L b +λ2L c ; Among them, λ1 and λ2 are the weights of the loss function of the background prediction branch and the loss function of the conductivity prediction branch, respectively. They are set according to actual needs. The loss function of the background prediction branch is L b , the loss function of the conductivity prediction branch is L c ; The data used for training were obtained through simulation, including single-rod and double-rod distribution of air with different diameters and positions under water background, single-rod and double-rod distribution of water with different diameters and positions under air background, and capacitance spectrum data of laminar flow with different heights and slopes at four conductivity levels of 0.001S / m, 0.005S / m, 0.01S / m and 0.05S / m. The Pytorch framework was used for training, and the Adam optimizer was used with a learning rate of 5e-5 for 15 rounds of optimization.

2. An image reconstruction method, characterized in that: The method comprises: According to the conductivity and flow pattern background prediction method of claim 1, the flow pattern background and conductivity of the current fluid are predicted, and the capacitance spectrum data of the predicted flow pattern background are selected as linear points for image reconstruction.

3. The image reconstruction method according to claim 2, characterized in that: The predicted flow pattern background of the current fluid is air; The selecting the predicted capacitance spectrum data of the flow pattern background as a linear point for image reconstruction comprises: The first capacitance spectrum data is selected as a linear point for image reconstruction.

4. The image reconstruction method according to claim 2, characterized in that: The predicted flow pattern background of the current fluid is water; The selecting the predicted capacitance spectrum data of the flow pattern background as a linear point for image reconstruction comprises: According to the predicted conductivity, obtaining third capacitance spectrum data corresponding to the conductivity when the pipeline is full of water; The third capacitance spectrum data is selected as a linear point for image reconstruction.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the conductivity and flow pattern background prediction method according to claim 1 and / or the image reconstruction method according to any one of claims 2-4 are implemented.

6. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the conductivity and flow pattern background prediction method as claimed in claim 1 and / or the image reconstruction method as claimed in any one of claims 2-4 are implemented.

Images