Hydrate method seabed carbon sequestration electrical resistance tomography inversion method and system

Abstract

The invention discloses a hydrate method seabed carbon sequestration electrical resistance tomography inversion method and system, and relates to the field of carbon sequestration and electrical resistance tomography. The hydrate method seabed carbon sequestration electrical resistance tomography inversion method and system mainly comprises the following steps: obtaining a simulation data set by using a finite element method; and performing image reconstruction by using a resistivity tomography method to obtain a training data set, training the neural network model to obtain a trained neural network model, and predicting the boundary voltage data to be measured to obtain a resistivity distribution image. By implementing the hydrate method seabed carbon sequestration electrical resistance tomography inversion method and system provided by the invention, real-time monitoring of hydrate generation, distribution and sequestration effect in the seabed carbon sequestration process can be realized.

Description

Technical Field

[0001] The present invention relates to the fields of carbon sequestration and electrical resistance tomography, and more particularly, to a method and system for inverse electrical resistance tomography of hydrate-based subsea carbon sequestration. Background Art

[0002] Carbon Capture, Utilization and Storage (CCUS) technology is one of the key technologies for achieving global climate change goals and China's "dual carbon" goals. The CCUS technology includes multiple links such as carbon capture, transportation, utilization, and storage. Among them, CO2 geological sequestration is an important means to achieve permanent CO2 emission reduction. Offshore carbon sequestration is considered to be the most promising geological sequestration site because it is far from aquifers and is blocked by seawater on the surface. However, real-time monitoring of the injected CO2 is the key to ensuring the safe implementation of the marine CCUS project. At present, reservoir evaluation based on electrical response characteristics plays an important role in CCUS technology, but there are still defects such as large amounts of data and complex data processing processes. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for inverse electrical resistance tomography of hydrate-based subsea carbon sequestration, which can realize real-time monitoring of the formation, distribution, and sequestration effect of hydrates during the subsea carbon sequestration process.

[0004] The present invention provides a method for inverse electrical resistance tomography of hydrate-based subsea carbon sequestration, including the following steps: S1: According to the two-dimensional imaging area, a simulation data set is obtained by using the finite element method; S2: According to the simulation data set, image reconstruction is performed by using the electrical resistance tomography method to obtain a training data set; S3: The neural network model is trained by using the training data set to obtain a trained neural network model; S4: The trained neural network model is used to predict the boundary voltage data to be measured, and a resistivity distribution image is obtained.

[0005] The present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for inverse electrical resistance tomography of hydrate-based subsea carbon sequestration are implemented.

[0006] The present invention also provides a resistive tomography inversion system for hydrate-based subsea carbon sequestration, which includes a resistive tomography module for reconstructing an image based on boundary voltage data to obtain a resistivity distribution image. The resistive tomography module includes the above-mentioned computer device. The resistive tomography inversion system for hydrate-based subsea carbon sequestration further includes a subsea environment simulation module and a boundary voltage measurement module. The subsea environment simulation module is used to simulate the subsea temperature and pressure environment and generate hydrates. The boundary voltage measurement module is used to collect boundary voltage data.

[0007] Implementing the resistive tomography inversion method and system for hydrate-based subsea carbon sequestration provided by the present invention has the following beneficial effects: The present invention obtains the voltage signal around the measured electrode and solves the inverse problem of electrical resistivity tomography (ERT) technology based on the hydrate method to obtain the spatial distribution of resistivity of the measured cross-section, so as to distinguish the occurrence form and spatial distribution of hydrates. The resistive tomography inversion system for hydrate-based subsea carbon sequestration of the present invention includes a subsea environment simulation module, a boundary voltage measurement module, and a resistive tomography (ERT) system module. The subsea environment simulation module simulates the subsea high-pressure and low-temperature environment through equipment such as high-pressure reactors and water tanks. The boundary voltage measurement module uses electrode plates, enameled wires, etc. to collect boundary voltage data. The ERT system module combines iterative algorithms, non-iterative algorithms, and neural network algorithms for image reconstruction to distinguish the occurrence form and spatial distribution of hydrates, and has the characteristics of non-invasive, non-radiative, fast response, simple structure, and low cost, providing strong support for improving the efficiency and safety guarantee technology of CO 2 sequestration in the sea area. The resistive tomography inversion system for hydrate-based subsea carbon sequestration can simulate the real subsea environment, establish a resistivity distribution image using the voltage signal around the measured object, and thus intuitively present the distribution of hydrates, which can be used to study the generation, distribution of hydrates and their impact on the sequestration effect during the process of hydrate-based subsea carbon sequestration, etc., and realize the real-time monitoring of the generation, distribution of hydrates and their sequestration effect during the subsea carbon sequestration process. Description of the Drawings

[0008] The present invention will be further described below in conjunction with the drawings and embodiments. In the drawings: Figure 1 is the flow chart of the resistive tomography inversion method for hydrate-based subsea carbon sequestration provided by the present invention; Figure 2 is the architecture diagram of the convolutional neural network CNN algorithm provided by the present invention; Figure 3 is the architecture diagram of the recurrent neural network RNN algorithm provided by the present invention; Figure 4It is a neural network ResNet algorithm architecture diagram provided by the present invention; Figure 5 It is a structural block diagram of the computer device provided by the present invention; Figure 6 It is a schematic diagram of the structure of the electrical resistance tomography inversion system for seafloor carbon storage using the hydrate method provided by the present invention; In the figure: 1. high pressure reactor, 2. water tank, 3. constant temperature controller, 4. temperature sensor, 5. pressure sensor, 6. flow pump, 7. solenoid valve, 8. controller, 9. gas cylinder, 10. electrode sheet, 11. enameled wire, 12. broadband combined alternating current source, 13. data acquisition instrument; Figure 7 This is a front view of the structure of the high-pressure reactor provided by the present invention; Figure 8 It is a left view of the structure of the high-pressure reactor provided by the present invention; Figure 9 It is a top view of the structure of the high-pressure reactor provided by the present invention; Figure 10 The invention provides a high-pressure reactor sealing device capable of passing a wire. In the figure: 14. Main body bolt, 15. Attachment bolt, 16. Connecting screw, 17. Metal gasket, 18. Resin gasket, 19. Zirconia ceramic gasket insulating gasket. DETAILED DESCRIPTION

[0009] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0010] Figure 1 A schematic diagram of the hydrate method for seafloor carbon storage resistance tomography inversion method of this embodiment is shown. In this embodiment, the hydrate method for seafloor carbon storage resistance tomography inversion method includes the following steps: S1: According to the two-dimensional imaging area, the simulation data set is obtained using the finite element method; In an exemplary embodiment, step S1 specifically includes: S11: dividing the two-dimensional imaging area into a plurality of small areas, wherein the plurality of small areas include a sensitive field; and dividing the sensitive field into a plurality of triangular units using a triangulation method according to the plurality of small areas; S12: Based on multiple triangular elements, known resistivity distribution in the sensitive field and excitation current source, the finite element method is used to perform segmentation, interpolation and numerical calculation to obtain a numerical solution to the boundary problem; based on the numerical solution to the boundary problem, a simulation data set is obtained; As an exemplary embodiment, in step S12, according to multiple triangular elements, the resistivity distribution in the known sensitive field, and the excitation current source, the boundary problem is transformed into a corresponding variational problem. The finite element method is used for dissection and interpolation, and the variational problem is discretized into an extreme value problem of an ordinary multivariate function. The discretized variational problem is finally reduced to a set of multivariate algebraic equations and solved. The numerical calculation method is used to solve the multivariate algebraic equations to obtain the numerical solution of the boundary problem. It should be noted that the resistivity tomography inversion method for subsea carbon sequestration by the hydrate method mainly studies the problem-solving of electrical tomography. The main process is divided into a forward problem and an inverse problem. Among them, the solution of the forward problem is to solve the potential distribution at any point in the sensitive field on the premise of the known resistivity distribution and the excitation current source in the sensitive field. The inverse problem is an image reconstruction method that solves the resistivity distribution in the sensitive field from the measured boundary voltage and the excitation current, and then obtains the resistivity distribution image. The solution of the forward problem uses the finite element analysis method. First, the variational problem is determined. Secondly, the field domain is divided. The smaller the size of the discrete unit, the more discrete intervals are divided, and the more the combined body of the units approximates the solution domain. After dividing the solution domain, the linear interpolation function is used to solve the divided small discrete units, and the variational problem of the solution domain can be transformed into an extreme value problem of a multivariate function in the finite element subspace. Solving the extreme value of this linear function can complete the solution of the variational problem. The inverse problem solution methods can be divided into non-iterative methods, iterative methods, and ERT image reconstruction techniques based on neural networks. Among them, the non-iterative methods include the linear backprojection algorithm, etc. Its accuracy is low but the calculation speed is fast, which is suitable for real-time imaging. The iterative methods include the Gauss-Newton iterative algorithm, the TSVD algorithm, the conjugate gradient algorithm, etc. For the conjugate gradient algorithm, its accuracy is high but the calculation amount is large and the imaging time is long. This algorithm is an iterative algorithm improved on the basis of the steepest descent method. It corrects the iterative direction according to the conjugate relationship and is suitable for linear equations with a symmetric positive definite coefficient matrix. The calculation steps are: the normalization processing of the sensitivity matrix, iterative calculation until the residual after iteration is less than the set value to obtain the optimal solution. The ERT image reconstruction technology based on neural networks includes training models using the convolutional neural network CNN, the recurrent neural network RNN, and the neural network ResNet respectively to improve the accuracy and efficiency of image reconstruction. It should be noted that the forward problem solution method in this embodiment uses the finite element method. The steps mainly include: dividing the two-dimensional imaging area into multiple small areas, using the triangular division method to dissect the sensitive field into 1600 triangular elements; calculating the number E of independently measured boundary voltages according to the number N of electrodes, E = N(N - 3); transforming the boundary problem into a corresponding variational problem, and using dissection and interpolation to discretize the variational problem into an extreme value problem of an ordinary multivariate function; finally reducing the discretized variational problem to a set of multivariate algebraic equations and solving to obtain the numerical solution of the boundary problem. S2: According to the simulation data set, use the resistivity tomography method to perform image reconstruction to obtain a training data set; As an exemplary embodiment, the simulation data set contains 14,500 voltage-resistance sequence pairs; the neural network model realizes high-precision image reconstruction by learning the complex relationship between ERT measurement data and the resistivity of the medium; S3: Use the training data set to train the neural network model to obtain a trained neural network model; In an exemplary embodiment, the neural network model is a convolutional neural network or a recurrent neural network; Figure 2 It is the algorithm architecture diagram of the convolutional neural network CNN; since the CNN network is suitable for processing image data, and the data collected in this embodiment is measured voltage data, which does not match the shape of the picture data, before the data is input into the model, the shape of the data is transformed. The original sequence of 1*208 is transformed into a matrix of 13*16, and after zero padding, the output becomes a matrix of (18,18,1); then operations such as convolutional layer, BN layer and max pooling layer are performed to convert the output shape into 18*18*256; Figure 3 It is the algorithm architecture diagram of the recurrent neural network RNN; since the data obtained in this embodiment is the measured voltage data, that is, sequence data, no shape transformation is required, that is, the sequence input into the network is (208,1). It passes through two RNN layers with the number of neuron nodes being 512 and 750 respectively. At this time, the sequence shape is (208,750). The output sequence is flattened and passes through the first fully connected layer with 1000 neuron nodes. At this time, the output is (1000,1); 50% of the neurons are discarded to prevent possible overfitting; a fully connected neural network with 1600 neuron nodes is adopted, and the Sigmoid function is used for activation to make the output (1600,1); In an exemplary embodiment, Figure 4It is an architecture diagram of the ResNet algorithm for neural networks; the neural network model is configured as follows: perform shape transformation and zero-padding operations on the original sequence data to change the shape of the original sequence data to (18, 18, 1) to obtain the input data; perform upsampling and batch normalization (BN) layer processing on the input data to change the shape of the input data to (36, 36, 1), and pass through two convolutional layers with 256 and 512 filters respectively and a stride of 2 to change the data shape to (9, 9, 512), and a BN layer is connected after the convolutional layer; input the data into the Identity_block layer and the Conv_block layer to change the data shape to (9, 9, 512); input the data into the flattening layer, a fully connected layer with 1000 neural nodes, a Dropout layer, and a fully connected layer with 1600 neural nodes in sequence, and use the Sigmoid activation function to map the current data to the range [0, 1] to obtain the distribution image; It should be noted that in this embodiment, the neural network model is used to perform shape transformation and zero-padding operations on the original sequence, changing the shape of the data to (18, 18, 1). Then, upsampling (Upsampling) and batch normalization layer (batch_normalization, BN) are performed on the input data, and the shape of the data at this time is (36, 36, 1). After that, it passes through two convolutional layers with 256 and 512 filters respectively and a stride of 2, and a BN layer is connected after each convolutional layer. The shape of the data at this time is (9, 9, 512). Further, by inputting the data into the Identity_block layer and the Conv_block layer, the data shape is converted to (9, 9, 512). Finally, the data is input into the flattening layer, a fully connected layer with 1000 neural nodes, a Dropout layer, and a fully connected layer with 1600 neural nodes in sequence, and the Sigmoid activation function is used to map the current data to the range [0, 1]; It should be noted that in this embodiment, after designing the samples and making the samples, the data obtained from the samples can be preprocessed, model training can be carried out, sample testing can be carried out, and the results of the inversion imaging can be output; S4: Use the trained neural network model to predict the boundary voltage data to be measured to obtain the resistivity distribution image; It should be noted that for the above neural network-based ERT image reconstruction technology, CNN, ResNet, and RNN are built to train models respectively to solve the forward problem, so as to improve the accuracy and efficiency of image reconstruction; more specifically, the neural network-based ERT image reconstruction technology updates parameters, that is, important parameters of the inverse problem are set, including loss function, learning rate, and number of training epochs; it mainly includes the following steps: using a simulation dataset for ERT image reconstruction, and the dataset is obtained by solving the ERT forward problem through the finite element method; training a neural network model, and the model includes a convolutional neural network (CNN), a residual network (ResNet), or a recurrent neural network (RNN); dividing the training dataset into a training set and a validation set, and the ratio of the training set to the validation set is 28:1. In this embodiment, for the above imaging algorithm, performance testing and analysis are carried out, and the steps include: preparing data samples, establishing performance evaluation indicators, and result analysis; among them, preparing data samples includes the following steps: constructing a massive or nodular hydrate model sample, the model sample has massive and nodular growth characteristics, and a large high-resistivity hydrate structure accompanied by a small amount of sandstone particles; constructing a layered or vein-like hydrate model sample, the model sample grows in fractures and often has an irregular shape; by changing the position and quantity of the high-resistivity body (hydrate) generated in the model, designing simulation samples; among them, the performance evaluation indicators include: mean square error (MSE), mean absolute error (MAE), and correlation coefficient; among them, the performance evaluation indicators include the following steps: using a neural network model to process ERT measurement data, and the neural network model includes a convolutional neural network (CNN), a residual network (ResNet), or a recurrent neural network (RNN); evaluating the fitting performance of the model, and measuring the error between the predicted value and the true value by calculating the mean square error (MSE) and the mean absolute error (MAE); the correlation coefficient indicates the linear correlation between the predicted value and the true value; among them, the result analysis includes: evaluating the prediction performance of the neural network model with reference to the above evaluation indicators; it is determined that the neural network models all show relatively low mean square error (MSE) and mean absolute error (MAE), indicating that the error between their predicted values and the true values is small and they can effectively fit the data; it is determined that the recurrent neural network (RNN) has the lowest MSE, showing its better fitting performance; it is determined that the convolutional neural network (CNN) has the lowest mean square error, but its MAE is relatively high, probably because the fitting effect on a small number of extreme values is not ideal enough; it is determined that the RNN model achieves a better balance in terms of error and correlation, showing an overall stable performance; it is determined that the traditional algorithms have relatively high MSE and MAE, and the correlation coefficient is relatively low or even negative, indicating that these methods fail to capture the correct trend of the data.

[0011] In an exemplary embodiment, the above resistivity tomography inversion method for subsea carbon sequestration by the hydrate method can also be implemented in the following manner. In this embodiment, the resistivity tomography inversion method for subsea carbon sequestration by the hydrate method is an ERT imaging based on the linear backprojection algorithm, including the following steps: finite element field discretization; generating a projection and equipotential line coverage matrix; determining the projection region to which each discretized element belongs; calculating the resistivity of each element under the i-th excitation according to the measured voltage and the projection region to which the element belongs; and reflected projection calculation.

[0012] It should be noted that the non-iterative method selects the backprojection method to inversely calculate the resistivity distribution inside the object from the measurement data of multiple electrodes (i.e., the current and voltage at different positions). Its steps include: obtaining projection data, recording the current (or measured voltage) passing through this path, and these current (voltage) values represent the resistivity information inside the object; backprojection integration; solving the inverse problem. Due to the uncertainty in the reconstruction process caused by the imperfection of the data (such as noise or incomplete measurement), a regularization method is used to stabilize the solution of the inverse problem.

[0013] In an exemplary embodiment, the above resistivity tomography inversion method for subsea carbon sequestration by the hydrate method can also be implemented in the following manner. In this embodiment, the resistivity tomography inversion method for subsea carbon sequestration by the hydrate method is an ERT imaging based on the conjugate gradient algorithm, including the following steps: Regularization processing step, where regularization processing is performed through the formula ; Iterative calculation step, including: calculating the step size: ; Updating the solution: ; Updating the residual: ; Calculating the coefficient: ; Updating the direction: ; Judgment step, if the residual is less than the set value, the iteration ends; otherwise, return to the iterative calculation step for multiple iterations.

[0014] This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the above resistivity tomography inversion method for subsea carbon sequestration by the hydrate method.

[0015] Such as Figure 5As shown, the computer device 120 may include: at least one processor 121, such as a Central Processing Unit (CPU), at least one communication interface 123, a memory 124, and at least one communication bus 122. Among them, the communication bus 122 is used to realize the connection and communication between these components. Among them, the communication interface 123 may include a display screen and a keyboard. Optionally, the communication interface 123 may also include a standard wired interface and a wireless interface. The memory 124 may be a high-speed random access memory (RAM), or a non-volatile memory, such as at least one disk memory. Optionally, the memory 124 may also be at least one storage device located far from the aforementioned processor 121. Among them, application programs are stored in the memory 124, and the processor 121 calls the program code stored in the memory 124 to execute any of the above method steps. Among them, the communication bus 122 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The communication bus 122 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5It is represented by only one line in the figure, but it does not mean that there is only one bus or one type of bus. Among them, the memory 124 may include volatile memory, such as random-access memory (RAM); the memory may also include non-volatile memory, such as flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 124 may also include a combination of the above types of memories. Among them, the processor 121 may be a central processing unit (CPU), a network processor (NP), or a combination of a CPU and an NP. Among them, the processor 121 may further include a hardware chip. The above hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The above PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof. Optionally, the memory 124 is also used to store program instructions. The processor 121 may call the program instructions to implement the hydrate method for subsea carbon sequestration electrical resistance tomography inversion method and system as in this embodiment.

[0016] This embodiment provides a hydrate method for subsea carbon sequestration electrical resistance tomography inversion system, including an electrical resistance tomography module for reconstructing an image based on boundary voltage data to obtain a resistivity distribution image. The electrical resistance tomography module includes the above computer device. It is characterized in that the hydrate method for subsea carbon sequestration electrical resistance tomography inversion system further includes a subsea environment simulation module and a boundary voltage measurement module; the subsea environment simulation module is used to simulate the subsea temperature and pressure environment and generate hydrates; the boundary voltage measurement module is used to collect boundary voltage data.

[0017] In an exemplary embodiment, the subsea environment simulation module includes a high-pressure reactor 1, a water tank 2, a constant temperature controller 3, a temperature sensor 4, a pressure sensor 5, a flow pump 6, a solenoid valve 7, a controller 8, and a gas cylinder 9; the boundary voltage measurement module includes a plurality of electrode plates 10, enameled wires 11, a broadband combined alternating current source 12, and a data acquisition instrument 13;Figure 6 It is a schematic structural diagram of a resistance tomography inversion system for undersea carbon sequestration by the hydrate method; Figure 6 In it, 1 is a high-pressure reactor, 2 is a water tank, 3 is a constant temperature controller, 4 is a temperature sensor, 5 is a pressure sensor, 6 is a flow pump, 7 is a solenoid valve, 8 is a controller, 9 is a gas cylinder, 10 is an electrode plate, 11 is an enameled wire, 12 is a broadband combined alternating current source, and 13 is a data acquisition instrument; In an exemplary embodiment, the upper part of the first side surface of the water tank 2 is provided with a first connection hole, the lower part of the first side surface of the water tank 2 is provided with a second connection hole, the upper part of the opposite surface of the first side surface of the water tank 2 is provided with a third connection hole, and the lower part of the opposite surface of the first side surface of the water tank 2 is provided with a fourth connection hole. The apertures of the first connection hole, the second connection hole, the third connection hole, and the fourth connection hole are the same; the first connection hole and the third connection hole are water outlet holes, and the second connection hole and the fourth connection hole are water inlet holes.

[0018] In an exemplary embodiment, the high-pressure reactor 1 includes a cylindrical hollow structure kettle body, an upper cover, and a lower cover. The upper cover and the lower cover are sealed with blind flanges having a plurality of circular holes with equal diameters. The upper cover and the lower cover are provided with ultrasonic probes, inlets and outlets, and pressure measurement ports; the inner layer of the kettle body is provided with a high-temperature sprayed insulation sealing material layer combined with its side wall gap, and the kettle body is also provided with 4 measurement surfaces, and 2 of the measurement surfaces are used for fixedly installing ultrasonic probes; the other 2 opposite measurement surfaces are respectively provided with 4 groups of interfaces arranged at equal intervals from top to bottom; the constant temperature circulator 3 includes a circulation pump and a compressor, and the constant temperature circulator is used to refrigerate or heat the circulating liquid in the water tank 2; the gas cylinder 9 is used to store liquid carbon dioxide, convert the liquid carbon dioxide into gaseous carbon dioxide, and inject the gaseous carbon dioxide into the high-pressure reactor.

[0019] As an exemplary embodiment, the water tank 2 includes: a square tank body; one connection hole is provided on one side surface of the upper part of the water tank 2; two connection holes are provided on the lower part of this side surface, and the apertures of these two connection holes are the same as the aperture of the upper connection hole; one and two connection holes are respectively provided on the upper and lower parts of the opposite surface of the water tank 2, and their apertures are the same as the apertures of the connection holes at the corresponding positions on the aforementioned side surface.

[0020] As an exemplary embodiment, the high-pressure reactor 1 includes: a designed pressure of 25 MPa, made of 316 material as a whole, with a height of approximately 540 mm, and a cylindrical hollow structure with a height of 300 mm and a diameter of 100 mm inside; the inside of the reactor body is processed with a high-temperature sprayed insulating and sealing material, which is more firmly combined with the side wall gap, preventing gas from infiltrating into the interlayer under high pressure, resulting in the bulging and then falling off of the lining material; the upper and lower covers of the high-pressure reactor are sealed with blind flanges having multiple circular holes of equal diameter, and 2 ultrasonic probes are fixedly installed on the upper and lower covers, and inlets, outlets and pressure measurement ports are reserved; the reactor body is provided with four measurement surfaces, two of which are used for fixedly installing 2 ultrasonic probes; on the other two opposite measurement surfaces, 4 groups of interfaces are arranged at equal intervals from top to bottom, with a total of 8 groups of interfaces on both sides; the interfaces are used to install electrodes or thermistors; two electrodes can be installed on each interface; further, the interface is a sealed hole of the high-pressure reactor through which wires can pass; the sealed hole of the high-pressure reactor includes a main bolt, an attached bolt, four connecting screws, a metal gasket, a resin gasket, and a zirconia ceramic washer insulating gasket. There are two holes with a diameter of 1.5 mm in the center of the main bolt for two wires to pass through; the liquid carbon dioxide in the gas cylinder is converted into gaseous carbon dioxide through a controller and connected to the high-pressure reactor to inject gas; a heat conduction probe is vertically placed at the lower side inside the reactor body; the tank body is an open-top square tank; there is one upper connection hole and two lower connection holes on each of the left and right side surfaces of the tank body, and all the hole diameters are approximately 120 mm; the arrangement of multiple acoustic, thermal, and electrical sensors is completed through the interfaces; the high-pressure reactor 1 is completely placed in the water tank 2, and the liquid inside the tank completely submerges the high-pressure reactor. The water bath method is adopted, with the liquid as the constant temperature medium. The two upper connection holes of the water tank 2 are water outlet holes, and the four lower connection holes are water inlet holes, a total of six holes, which are connected to a constant temperature controller 3 through hoses; the constant temperature circulator includes a circulation pump and a compressor, which are responsible for pumping the circulating liquid in the water tank 2 into the constant temperature circulator for refrigeration or heating during system operation, with a control range of -20°C to 80°C, and then flowing back into the water tank 2 through the bath liquid outlet; the four lower connection holes on the lower sides of the two side surfaces inside the tank body are water inlet holes, which are connected to the constant temperature controller through hoses, and the two upper connection holes are water outlet holes, which are connected with hoses, and the other ends of the hoses are directly placed in the liquid of the constant temperature circulator; the constant temperature controller includes a circulation pump and a compressor. When the system is operating, the constant temperature controller is responsible for pumping the circulating liquid in the water tank into the constant temperature controller for refrigeration or heating, with a control range of -20°C to 80°C, and then flowing back into the water tank through the bath liquid outlet.

[0021] Figure 7 is the front view of the high-pressure reactor structure; Figure 8 is the left view of the high-pressure reactor structure; Figure 9 is the top view of the high-pressure reactor structure; Figure 10It is a high-pressure reactor sealing device that can conduct wires; among them, 14 is the main body bolt, 15 is the attached body bolt, 16 is the connecting screw, 17 is the metal gasket, 18 is the resin gasket, and 19 is the zirconia ceramic washer insulating gasket.

[0022] The following introduces the temperature control method of the high-pressure reactor 1: The dynamic water bath is equipped with a circulation pump device, so that the injected liquid can flow in the water tank 2, the temperature distribution is more uniform, and the stability is better. The PID algorithm is used to achieve temperature control, and the control accuracy is ±0.05°C; The pressure control method of the high-pressure reactor 1 includes the following steps: Continuously inject water with a constant flow rate into the high-pressure reactor 1 using the flow pump 6; Measure the pressure in the high-pressure reactor 1 through the pressure sensor 5 and transmit the pressure signal to the controller 8; The controller 8 calculates the difference between the actual pressure and the set pressure; The controller 8 calculates and outputs an electrical signal of an appropriate magnitude through the PID algorithm to control the opening of the solenoid valve 7, thereby controlling the pressure in the high-pressure reactor 1; When the pressure is insufficient, reduce the opening of the solenoid valve 7 to reduce the flow rate of the discharged liquid, so that the liquid injection flow rate of the flow pump 6 is greater than the liquid discharge flow rate of the solenoid valve 7, so that the liquid in the high-pressure reactor 1 is compressed and the pressure increases; When the pressure is too high, increase the flow rate of the discharged liquid of the solenoid valve 7 to reduce the pressure in the high-pressure reactor 1; Multiple PID adjustments make the flow rate of the injected water and the flow rate of the discharged water basically remain unchanged, and the pressure in the high-pressure reactor 1 tends to be stable.

[0023] The following introduces the operation of the undersea environment simulation module: When the reaction device is under the simulated real temperature and pressure conditions of the seabed, hydrates can start to be generated, and the current temperature and pressure conditions can be monitored by a computer; Its steps mainly include: First, install several electrode plates 10 and enameled wires 11 in the high-pressure reactor 1, and the enameled wire 11 passes through the interface. Use a high-pressure reactor sealing device that can conduct wires to pass through as Figure 5As shown, it is sealed; the electrode plates are placed in the kettle body at equal intervals; the temperature sensor 4 and the pressure sensor 5 are inserted into the kettle body from the top and bottom of the high-pressure reactor 1; appropriate granular objects are selected to fill the inside of the kettle body. The high-pressure reactor 1 is connected to the gas cylinder 9 for airtightness inspection; after the high-pressure reactor 1 is sealed, it is placed in the water tank 2. There are a total of four connection holes at the lower parts of the two side surfaces inside the tank body of the water tank 2, which are water inlet holes and are connected to the thermostat through hoses. The two upper connection holes are water outlet holes and are connected by hoses. The other ends of the hoses are directly placed in the liquid of the thermostat 3; the thermostat 3 is responsible for pumping the circulating liquid in the water tank into the thermostat for refrigeration or heating, and the control range is -20°C to 80°C. Then it flows back into the water inlet in the water tank through the bath liquid outlet. The circulating pump device in the thermostat 3 enables the injected liquid to flow in the water tank, with uniform temperature distribution and ensuring stability. At the same time, combined with the PID algorithm, the control accuracy can reach ±0.05°C; adjust the pressure environment of the high-pressure reactor 1. The pressure sensor 5 is connected to the controller 8, and the controller 8 is connected to the solenoid valve 7. The water inlet and outlet holes of the water tank 2 can be directly controlled through the solenoid valve 7 and the stop valve connected to the solenoid valve 7; the flow pump 6 is used to continuously inject water with a constant flow rate into the high-pressure reactor 1. The pressure in the high-pressure reactor 1 is measured by the pressure sensor 5, and the pressure signal is transmitted to the controller 8. The controller 8 calculates the difference between the actual pressure and the set pressure; the controller 8 calculates and outputs an electrical signal of an appropriate magnitude through the PID algorithm to control the opening of the solenoid valve 7, thereby controlling the pressure in the high-pressure reactor 1. When the pressure is insufficient, the opening of the solenoid valve 7 is reduced to reduce the flow rate of the discharged liquid, so that the liquid injection flow rate of the flow pump 6 is greater than the liquid discharge flow rate of the solenoid valve 7, so that the liquid in the high-pressure reactor 1 is compressed and the pressure increases. After multiple PID adjustments, the flow rate of the injected water and the flow rate of the discharged water are basically kept unchanged, and the pressure in the high-pressure reactor 1 tends to be stable.

[0024] The boundary voltage measurement module includes electrode plates and enameled wire 11. During use, an electric field is established. The current source is an alternating current source with a frequency of 50 Hz and an excitation current of 10 mA. The excitation mode used is the adjacent excitation mode. A data acquisition instrument 13 is used to record the boundary voltage data. The electrode plate is a rectangular electrode with a length of 2 cm and a width of 1 cm. The electric field generated by the rectangular electrode is approximately a parallel field, and the current density distribution within the field domain is relatively uniform, which can provide a more uniform electric field distribution and has a good improvement effect on the uneven current density distribution within the electric field. Considering that the marine environment has strong corrosiveness, the electrode plate should not only have good electrical conductivity but also corrosion resistance, and the polarization effect on the electrode surface should be considered. Considering economic issues, the electrode plate is finally selected to be made of copper. The output of the broadband combined alternating current source is a sine wave. The adjacent excitation mode includes: connecting the output wire ends of the broadband combined alternating current source 12 to any two adjacent electrodes to establish an electric field within the object. According to the clockwise or counterclockwise order, the voltage is measured between other adjacent electrode pairs that have not been applied with an excitation signal, and 13 boundary voltage values are obtained to complete the measurement of the first round of boundary voltages. According to the clockwise or counterclockwise order, the broadband combined alternating current source 12 is connected to the next group of adjacent electrodes, and the measurement steps of the first round are repeated until excitation current has been applied to all adjacent electrode pairs. At this time, a total of 208 boundary voltage values are measured. According to the reciprocity theorem, the measurement values obtained when the current source and the voltmeter are interchanged have the same effect. Therefore, when the N-electrode electrical resistance tomography system uses the adjacent excitation mode for measurement, (N - 3)N / 2 independent measurement values can be obtained, and among the 208 values measured using 16 electrodes, the relatively independent values are 104.

[0025] The following introduces the operation of the boundary voltage measurement module, which is mainly used to obtain the boundary voltage values around the object to be measured and provide data for the electrical resistance tomography (ERT) system module. It mainly includes: connecting the enameled wire 11 to the terminal block of the data acquisition instrument 13. After determining the channel access, connect the output wire ends of the broadband combined alternating current source 12 to any two adjacent electrodes to establish an electric field within the object, output a sine wave, select an excitation current of 10 mA, and use the adjacent excitation mode, that is, according to the clockwise or counterclockwise order, measure the voltage between other adjacent electrode pairs that have not been applied with an excitation signal, and 13 boundary voltage values are obtained to complete the measurement of the first round of boundary voltages. According to the clockwise or counterclockwise order, connect the broadband combined alternating current source to the next group of adjacent electrodes, and repeat the measurement steps of the first round until excitation current has been applied to all adjacent electrode pairs. At this time, a total of 208 boundary voltage values are measured. According to the reciprocity theorem, the measurement values obtained when the current source and the voltmeter are interchanged have the same effect. Therefore, when the N-electrode electrical resistance tomography system uses the adjacent excitation mode for measurement, (N - 3)N / 2 independent measurement values can be obtained, and among the 208 values measured using 16 electrodes, the relatively independent values are 104.

[0026] It should be noted that in the boundary voltage measurement module, the electrode sheet 10 is made of copper plated with gold and inert electrodes such as silver (Ag) and silver chloride (AgCl), which can greatly weaken the electrode polarization effect generated by the excitation current. After the electrode sheet 10 is welded to the enameled wire 11, it is placed in an equally spaced square groove circular ring made of acrylic material and longitudinally placed in the high-pressure reactor 1. 16 electrode sheets 10 are placed at equal intervals in each layer of the ring, with a total of 64 in four layers; after the enameled wire 11 passes through the sealed hole of the high-pressure reactor 1 that can conduct wires from the side of the reactor body of the high-pressure reactor 1, it is connected to the connection terminal of the data acquisition instrument 13; the temperature and pressure are adjusted by using the above temperature control method and pressure control method to make them close to the real seabed environment; during use, an electric field is established, and the current source is a broadband programmable alternating current source. This excitation current source combines a signal generator and a high-voltage amplifier to output a controllable waveform, different frequency excitation current signals, the frequency is adjustable in the range of 1 Hz - 40 KHz, the excitation current is adjustable in the range of 0 - 80 mA, its value is affected by the load of the medium to be measured, and its magnitude is adjustable by loading voltage through the high-voltage amplifier. Finally, a stable high-frequency and low-noise excitation signal output is achieved, and the excitation mode used is the adjacent excitation mode; the data acquisition instrument 13 is used to record accurate boundary voltage data.

[0027] It should be noted that the above adjacent excitation mode includes the following steps: Connect the output wire ends of the broadband combined alternating current source 12 to any two adjacent electrodes to establish a stable and sensitive electric field inside the object; Measure the voltage between other adjacent electrode pairs that have not been applied with excitation signals in a clockwise or counterclockwise order to obtain 13 boundary voltage values and complete the measurement of the first round of boundary voltages; Connect the broadband combined alternating current source 12 to the next group of adjacent electrodes in a clockwise or counterclockwise order and repeat the measurement steps of the first round until excitation current has been applied to all adjacent electrode pairs. At this time, a total of 208 boundary voltage values are measured; According to the reciprocity theorem, the measured values obtained when the current source and the voltmeter are interchanged have the same effect. Therefore, when the N-electrode electrical resistance tomography system uses the adjacent excitation mode for measurement, (N - 3)N / 2 independent measurement values can be obtained, and among the 208 values measured with 16 electrodes, the relatively independent values are 104.

[0028] It should be noted that during the actual operation process, according to the medium to be measured, the frequency and current magnitude of the excitation signal need to be adjusted, and whether the waveform is a sine wave or a specific square wave also needs to be adjusted; Selecting the most optimized excitation current signal to cooperate with neural network inversion has the best effect, especially for complex formations (inhomogeneous resistance layers).

[0029] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit of the present invention and the scope protected by the claims. All of these fall within the protection scope of the present invention.

Claims

1. A method for electrical resistance tomography inversion of seafloor carbon storage by hydrate method, characterized in that: The following steps are involved: S1: According to the two-dimensional imaging area, the simulation data set is obtained using the finite element method; S2: Reconstructing images using a resistivity tomography method according to the simulation data set to obtain a training data set; S3: Using the training data set to train the neural network model to obtain a trained neural network model; S4: Using the trained neural network model to predict the boundary voltage data to be measured, to obtain a resistivity distribution image.

2. The method for electrical resistance tomography inversion of hydrate-based seafloor carbon storage according to claim 1, characterized in that: Step S1 specifically includes: S11: dividing the two-dimensional imaging area into a plurality of small areas, wherein the plurality of small areas include a sensitive field; and dividing the sensitive field into a plurality of triangular units using a triangulation method according to the plurality of small areas; S12: According to the multiple triangular units, the resistivity distribution in the known sensitive field and the excitation current source, the finite element method is used to perform dissection, interpolation and numerical calculation to obtain a numerical solution to the boundary problem; according to the numerical solution to the boundary problem, a simulation data set is obtained.

3. The method for electrical resistance tomography inversion of hydrate-based seafloor carbon storage according to claim 1, characterized in that: The neural network model is a convolutional neural network or a recurrent neural network.

4. The method for electrical resistance tomography inversion of hydrate-based seafloor carbon storage according to claim 1, characterized in that: The neural network model is configured as follows: shape conversion and zero padding operations are performed on the original sequence data to change the shape of the original sequence data to (18, 18, 1) to obtain input data; upsampling and batch normalization layer processing are performed on the input data to change the shape of the input data to (36, 36, 1), and after passing through two convolutional layers with 256 and 512 filters respectively and a step size of 2, the data shape is changed to (9, 9, 512), and a BN layer is connected after the convolutional layer; the data is input into the Identity_block layer and the Conv_block layer to change the data shape to (9, 9, 512); the data is sequentially input into the flattening layer, the fully connected layer with 1000 neural nodes, the Dropout layer, and the fully connected layer with 1600 neural nodes, and the Sigmoid activation function is used to map the current data to between [0, 1] to obtain a distribution image.

5. A computer 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 computer program, the steps of the hydrate-based seafloor carbon storage electrical resistance tomography inversion method as described in any one of claims 1 to 5 are implemented.

6. A hydrate-based seafloor carbon storage resistance tomography inversion system, comprising a resistance tomography module for performing image reconstruction based on the boundary voltage data to obtain a resistivity distribution image, the resistance tomography module comprising the computer device according to claim 5, characterized in that: The hydrate method seabed carbon storage resistance tomography inversion system also includes a seabed environment simulation module and a boundary voltage measurement module; the seabed environment simulation module is used to simulate the seabed temperature and pressure environment to generate hydrates; the boundary voltage measurement module is used to collect boundary voltage data.

7. The hydrate method seabed carbon storage electrical resistance tomography inversion system according to claim 6, characterized in that: The seabed environment simulation module comprises a high-pressure reactor (1), a water tank (2), a constant temperature controller (3), a temperature sensor (4), a pressure sensor (5), a flow pump (6), a solenoid valve (7), a controller (8), and a gas cylinder (9); the boundary voltage measurement module comprises a plurality of electrode sheets (10), enameled wires (11), a broadband combined alternating current source (12), and a data acquisition instrument (13).

8. The hydrate method seabed carbon storage electrical resistance tomography inversion system according to claim 7, characterized in that: A first connection hole is provided at the upper portion of the first side surface of the water tank (2), a second connection hole is provided at the lower portion of the first side surface of the water tank (2), a third connection hole is provided at the upper portion of the opposite surface of the first side surface of the water tank (2), and a fourth connection hole is provided at the lower portion of the opposite surface of the first side surface of the water tank (2), and the first connection hole, the second connection hole, the third connection hole and the fourth connection hole have the same aperture; the first connection hole and the third connection hole are water outlet holes, and the second connection hole and the fourth connection hole are water inlet holes.

9. The hydrate method seabed carbon storage electrical resistance tomography inversion system according to claim 7, characterized in that: The high-pressure reactor (1) comprises a reactor body, an upper cover and a lower cover which are arranged in a cylindrical hollow structure. The upper cover and the lower cover are sealed by a blind flange having a plurality of circular holes of equal diameter, and the upper cover and the lower cover are provided with an ultrasonic probe, an inlet and an outlet, and a pressure measuring port; The inner layer of the kettle body is provided with a high-temperature sprayed insulating sealing material layer combined with the gap of the side wall, and the kettle body is also provided with 4 measuring surfaces, 2 of which are used for fixing and installing ultrasonic probes; the other 2 measuring surfaces opposite to each other are respectively provided with 4 groups of interfaces arranged at equal intervals from top to bottom; The constant temperature circulator (3) comprises a circulation pump and a compressor, and is used to cool or heat the circulating liquid in the water tank (2); The gas cylinder (9) is used to store liquid carbon dioxide, convert the liquid carbon dioxide into gaseous carbon dioxide, and inject the gaseous carbon dioxide into the high-pressure reactor.

Images