Digital twinborn body construction method for coal-fired flue gas CO2 trapping device

Abstract

The invention belongs to the technical field of oil and gas field surface engineering CCUS, and discloses a digital twinborn body construction method for a coal-fired flue gas CO2 trapping device. The digital twinborn building method comprises the following steps: collecting static parameters and dynamic parameters of each device in a system A, and building a twinborn space based on an entity system A; constructing a three-dimensional visual model based on the twin space; training and optimizing the three-dimensional visualization model; the three-dimensional visualization model is verified; and according to the verified three-dimensional visual model, pre-judging and regulating and controlling the concentration value of the final captured CO2 in the entity system A in real time. According to the method, the digital twinborn space of the coal-fired flue gas CO2 capturing system A is constructed, dynamic / static data are combined, data exchange and virtual pre-judgment between the digital twinborn space and the entity space are achieved, intelligent sensing and real-time monitoring of the CO2 capturing system A can be accurately achieved, and pre-judgment and analysis of the final captured CO2 concentration value can be accurately achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of CCUS for oil and gas field surface engineering, and specifically relates to a digital twin construction method for a coal-fired flue gas CO2 capture device. Background Art

[0002] During the operation of coal-fired units to generate electricity, a certain amount of CO2 will be emitted along with the flue gas. By capturing and recovering the CO2 in the exhaust flue gas and purifying and reinjecting it, CO2 emissions can be reduced, which is beneficial to environmental protection. Among them, the CO2 capture device is mainly composed of pre-washing towers, absorption towers, regeneration towers, carbon dioxide pressurization and carbon dioxide dehydration facilities. Due to the combined influence of static data (such as material, rated power, etc.) and dynamic data (such as flow, pressure, temperature, etc.) during the capture process, it is impossible to predict the concentration of injected CO2 in advance, and it is impossible to predict and control the CO2 capture rate in advance, which may lead to a decrease in the CO2 capture rate, resulting in insufficient carbon sources, and a significant reduction in the oil field recovery effect. Therefore, the process flow of traditional CO2 capture devices does not fully consider the analysis and application of dynamic / static data.

[0003] Digital twin technology is a digital model of a physical object. It can sense, diagnose and predict the state of a physical object in real time through actual measurement, simulation and data analysis, regulate the behavior of the physical object through optimization and instructions, and evolve itself through mutual learning between related digital models, thereby realizing interactive mapping between physical space and digital space. Digital twin technology is mainly used in traditional oil and gas field surface processes in the oil and gas field industry, such as oil transfer stations, crude oil processing stations, natural gas processing stations, etc., but has not been seen in CCUS oil field surface stations, especially in associated gas recycling and reinjection stations.

[0004] Chinese patent CN111210359A discloses the digital twin evolution mechanism and method for intelligent mine scenarios. The digital twin is mutually coupled and evolved by physical model, logical model, simulation model and data model, and includes the following method steps: (1) physical model definition; (2) logical model representation; (3) simulation model establishment; (4) simulation model optimization; (5) simulation model verification; (6) data model construction; (7) digital twin integration; the patent realizes data mirroring and information interaction between digital twin and physical entity by constructing digital twin, and realizes object twin, process twin and performance twin of physical entity in physical space and digital twin in virtual space; through the digital twin evolution mechanism and method, remote visual monitoring of physical space intelligent mine scene is realized in virtual space intelligent mine scene, as well as intelligent perception, real-time monitoring, precise positioning and health prediction of mine application scene equipment. However, the method of the patent is mainly applied to intelligent mine scene, and is not suitable for CCUS oil field ground station, especially for coal-fired flue gas CO2 capture device process scene. Summary of the invention

[0005] The present invention aims to solve at least one technical problem existing in the background technology, and provides a digital twin construction method for a coal-fired flue gas CO2 capture device. By constructing a digital twin space of a coal-fired flue gas CO2 capture system A and combining dynamic / static data, data exchange and virtual prediction between the digital twin space and the physical space are realized, and the CO2 capture system A can be accurately intelligently sensed, monitored in real time, and the final captured CO2 concentration value can be predicted and analyzed in advance.

[0006] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0007] A digital twin construction method for a coal-fired flue gas CO2 capture device is applied to a coal-fired flue gas CO2 capture system A, wherein the system A comprises a washing tower, a fan, an absorption tower, a heat exchanger, a regeneration tower, a cooling / separator, a pressurizing device, a dehydration and liquefaction device connected in sequence; the washing tower is connected to a coal-fired unit outside the device A; the washing tower is also connected to a cooling tower, which is connected back to the washing tower through a first return pipe to form a circulating washing system; the absorption tower is connected to a venting pipe; the cooling / separator is connected back to the regeneration tower through a second return pipe; the dehydration and liquefaction device is connected to a pipeline system outside the system A, and is used to externally transport the finally captured CO2 through the pipeline system;

[0008] The digital twin construction method specifically includes the following steps:

[0009] Step S1. Collect static parameters and dynamic parameters of each device in system A, and establish a twin space based on the physical system A;

[0010] The static parameters are factory data parameters of each device in system A, and the dynamic parameters are real-time changing process parameters that need to be monitored during the process flow in system A;

[0011] Step S2. Based on the twin space established in step S1, construct a three-dimensional visualization model;

[0012] Step S3. Training and optimizing the three-dimensional visualization model: collecting the original dynamic data of each device that affects the final captured CO2 concentration value in the physical system A in real time, training and optimizing the twin space, and feeding back the optimization results to the three-dimensional visualization model;

[0013] Step S4. Verify the three-dimensional visualization model: If the three-dimensional visualization model meets the prediction conditions for correctness verification, proceed to the next step S5; if not, return to step S1;

[0014] Step S5. Based on the verified three-dimensional visualization model, the final captured CO2 concentration value in the physical system A is predicted and regulated in real time.

[0015] Furthermore, the digital twin is formed by the mutual fusion and application of the twin space and the three-dimensional visualization model constructed based on the physical system A, and the data mutual mapping, information synchronous interaction and feedback between the twin space and the physical space are realized through the digital twin;

[0016] The digital twin includes target twin, state and process twin and function twin;

[0017] The data model of the digital twin includes an acquisition layer, a transmission layer, a data layer and an application layer; wherein,

[0018] The acquisition layer is used to collect static parameters and dynamic data of each device in system A;

[0019] The transmission layer is used for data transmission of dynamic / static parameters of each device in system A;

[0020] The data layer is used for the integration, calculation, analysis, data twinning and storage management of the data of each device in system A;

[0021] The application layer provides users with real-time monitoring and advance prediction results of system A.

[0022] Furthermore, the process flow of the system A for capturing CO2 from coal-fired flue gas is as follows:

[0023] After the exhaust gas from the coal-fired unit enters system A, it is circulated and washed through the washing tower and cooling tower, and after cooling and impurity removal, the pre-treated flue gas is obtained;

[0024] After pretreatment, the flue gas is pressurized by a fan and transported to the absorption tower;

[0025] The absorption tower is provided with a lean ammonia liquid, which is used to absorb the captured CO2 gas and then converted into a rich ammonia liquid, and the unabsorbed tail gas is discharged through the venting pipe; the discharged tail gas is simultaneously monitored to see whether it meets the set flue gas temperature and SO2 concentration. If not, the absorption liquid concentration and the number of cycle washing in the washing tower are predicted and adjusted to meet the process requirements;

[0026] The ammonia-rich liquid converted in the absorption tower enters the regeneration tower after being heated by the heat exchanger, and is converted into ammonia-lean liquid after releasing the captured CO2 gas in the regeneration tower, and then returns to the absorption tower after being cooled by the heat exchanger;

[0027] The CO2 gas released by the regeneration tower enters the cooling / separator for cooling and separation to obtain purified CO2 gas and separation liquid; the separation liquid flows back to the regeneration tower through the second return pipe, and synchronously monitors whether the concentration of the purified CO2 gas meets the set CO2 concentration: if not, the purified CO2 gas is returned to the regeneration tower through the second return pipe, and the dynamic parameters of the absorption tower, the heat exchanger and the regeneration tower are synchronously adjusted to meet the process requirements before being transported to the booster; if satisfied, the purified CO2 gas is sequentially transported to the booster for pressurization, the dehydration and liquefaction device for dehydration and liquefaction to obtain the final captured CO2 gas;

[0028] The final captured CO2 gas is transported externally through a pipeline system, and the water content in the final captured CO2 gas is simultaneously monitored to see if it meets the set water content. If not, the dynamic parameters of the dehydration and liquefaction device need to be predicted and adjusted to meet the process requirements.

[0029] Furthermore, the step S1 specifically includes the following steps:

[0030] Step S101. Collect the static parameters and dynamic parameters of each device in system A, build a three-dimensional model of each device in physical system A to obtain a twin physical system A, and define the dynamic / static parameter attributes of each device in the twin physical system A;

[0031] Step S102. Based on the twin physical system A, connect pipelines between the devices in the twin physical system A to obtain the twin space, and characterize the components and operation mechanism of the twin space through the process flow formed by the connect pipelines.

[0032] Furthermore, the step S2 specifically includes the following steps:

[0033] Step S201. 3D visualization model construction: construct a visualization model through a built-in lightweight 3D visualization engine to realize the display and roaming browsing of the twin space on the client or handheld terminal;

[0034] Step S202. Data association: Using data hyperlink technology, the data between the entity and the twin are interconnected. Each entity corresponds to its own 2D document and 3D visualization model. Clicking on any monomer in the 3D visualization model can quickly link to the corresponding 2D document and static / dynamic parameters, and visualize the entity space, the real-time status of the entity space and the process flow;

[0035] Among them, the input of the three-dimensional visualization model is: static and dynamic parameters of each device;

[0036] The outputs of the 3D visualization model are: control variables that affect the final captured CO2 concentration value;

[0037] Step S203. Interactive control: Realize real-time data collection, remote communication and real-time update of dynamic data through communication interfaces such as OPC, TCP / IP, MODUBUS, etc.

[0038] Furthermore, in step S202, the control variables affecting the final captured CO2 concentration value include: the flue gas temperature and SO2 concentration emitted by the absorption tower, the CO2 concentration in the purified CO2 gas, and the water content in the final captured CO2 gas.

[0039] Furthermore, the step S3 specifically includes the following steps:

[0040] Step S301. Real-time acquisition of original dynamic data of each device in the physical system A that affects the final captured CO2 concentration value, wherein the original dynamic data at least includes:

[0041] The exhaust gas temperature and SO2 concentration emitted by the absorption tower, as well as the absorption liquid concentration and the number of wash cycles in the corresponding washing tower;

[0042] The CO2 concentration in the purified CO2 gas and the corresponding dynamic parameters of the absorption tower, heat exchanger and regeneration tower;

[0043] The water content in the final captured CO2 gas and the corresponding dynamic parameters of the dehydration and liquefaction equipment;

[0044] Clean and filter the original dynamic data to obtain valid original data;

[0045] Step S302. In the twin space, a two-layer LSTM long short-term memory network model is constructed based on the valid original data;

[0046] Step S303. Use valid original data to train a double-layer LSTM long short-term memory network model to obtain an optimized double-layer LSTM long short-term memory network model, and feed the optimization result back to the three-dimensional visualization model.

[0047] Furthermore, the step S303 specifically includes the following steps:

[0048] Step S3031. Use the valid original data that affects the final captured CO2 concentration value as the input value of the double-layer LSTM long short-term memory network model, and automatically calculate and generate the predicted value of the control variable that affects the final captured CO2 concentration value through the model;

[0049] Step S3032. Compare the predicted value of the control variable that affects the final captured CO2 concentration value with the actual collected value, and use the mean relative error MRE as an evaluation standard to perform error comparison analysis;

[0050] Step S3033. Repeat steps S3031-S3032 to train and optimize the double-layer LSTM long short-term memory network model for a certain period of time; until the mean relative error MRE value is reduced to the set target value, the optimized double-layer LSTM long short-term memory network model can be obtained, and an expert experience library is automatically formed in the model;

[0051] Step S3034. The twin space synchronously feeds back the optimization results to the three-dimensional visualization model to achieve training optimization of the three-dimensional visualization model.

[0052] Furthermore, in step S3033, the mean relative error MRE value is reduced to ≤10%, and an optimized double-layer LSTM long short-term memory network model can be obtained.

[0053] Furthermore, in step S4, the three-dimensional visualization model satisfies the prediction condition for correctness verification as follows:

[0054] The predicted value of the control variable affecting the final captured CO2 concentration value by the three-dimensional visualization model is close to the actual collected value, and the error between the two is ≤10%.

[0055] Furthermore, the step S5 specifically includes the following steps:

[0056] Step S501. Collect the dynamic parameters of each device in the physical system A in real time and map them to the twin space;

[0057] Step S502: The twin space predicts the predicted value of the control variable that affects the final captured CO2 concentration value according to the mapped dynamic parameters;

[0058] Step S503. The twin space compares the predicted value of the control variable that affects the final captured CO2 concentration value with the judgment threshold:

[0059] If the tail gas emitted by the absorption tower meets the following conditions: the lower limit of the set flue gas temperature judgment threshold ≤ flue gas temperature < the upper limit of the set flue gas temperature judgment threshold, and the SO2 concentration < the set SO2 concentration judgment threshold, then the subsequent CO2 capture process is executed; if not, the absorption liquid concentration and the number of cycle washing times in the washing tower are predicted and adjusted to meet the process requirements;

[0060] If the purified CO2 gas separated by cooling / separation meets the following conditions: CO2 concentration ≥ the set CO2 concentration judgment threshold, the subsequent CO2 capture process is executed; if not, the dynamic parameters of the absorption tower, heat exchanger and regeneration tower are predicted and adjusted to meet the process requirements;

[0061] If the captured CO2 gas obtained after dehydration and liquefaction of CO2 in the dehydration and liquefaction device meets the following conditions: CO2 water content < set water content judgment threshold, the subsequent CO2 capture process will be executed; if not, the dynamic parameters of the dehydration and liquefaction device need to be predicted and adjusted to meet the process requirements.

[0062] Furthermore, the method for obtaining the judgment threshold of the control variable that affects the final captured CO2 concentration value is:

[0063] Based on the dynamic data collected in real time over a long period of time in the physical system A, the twin space is continuously trained and optimized, and finally the critical value of the control variable that can meet the captured CO2 concentration value is obtained, which is the judgment threshold.

[0064] Furthermore, the judgment threshold of the control variable affecting the final captured CO2 concentration value is:

[0065] The exhaust gas discharged from the absorption tower must meet the requirements of 40℃≤ flue gas temperature<45℃, and SO2 concentration in flue gas<15mg / Nm 3 ;

[0066] The purified CO2 gas separated by cooling / separation in the cooling / separation device must meet the following requirements: CO2 concentration ≥ 94%;

[0067] After dehydration and liquefaction of CO2 in the dehydration and liquefaction device, the final captured CO2 gas must meet the following requirements: CO2 water content <100ppm.

[0068] Furthermore, a positioning alarm module is also embedded in the three-dimensional visualization model of the digital twin;

[0069] The positioning alarm module obtains the normal operating value range of the dynamic parameters of each device in the physical system A based on the trained and optimized three-dimensional visualization model; at the same time, the positioning alarm module obtains the dynamic parameters of each device collected in the physical system A in real time through data transmission, and compares them with the normal operating value range of each dynamic parameter in real time;

[0070] When the positioning alarm module monitors that the dynamic parameters of each device in system A have operating data that exceeds the normal operating value range, a positioning alarm is issued on the display screen of the three-dimensional visualization model, and auxiliary suggestions and decisions are given;

[0071] Based on the auxiliary suggestions and decisions given, the system automatically adjusts parameters for maintenance or manually rectifies the fault.

[0072] Compared with the prior art, the beneficial effects of the present invention are:

[0073] (1) The present invention provides a digital twin construction method for a coal-fired flue gas CO2 capture device. By constructing a digital twin space of the coal-fired flue gas CO2 capture system A and combining dynamic / static data, data exchange and virtual prediction between the digital twin space and the physical space can be realized, and the intelligent perception and real-time monitoring of the CO2 capture system A can be accurately realized, as well as the advance prediction and analysis of the final captured CO2 concentration value.

[0074] (2) The digital twin construction method for the coal-fired flue gas CO2 capture device provided by the present invention can realize the collection, uploading and storage of data based on the twin space constructed by the three-dimensional model, and can also realize automatic analysis and prediction of data, give reasonable control suggestions, and accurately execute control instructions; through the synchronous mapping of "static data + real-time dynamic data + prediction model", it can realize the accurate prediction and control of the final captured CO2 concentration value;

[0075] (3) The present invention provides a method for constructing a digital twin for a coal-fired flue gas CO2 capture device. Based on the digital twin, big data research is conducted to establish an intelligent operation and maintenance main line of "real-time data - positioning alarm - rapid prediction - auxiliary decision-making". Through the advance prediction of the digital twin, the abnormal operating parameters of the device can be discovered and adjusted in time, which can effectively improve the reliability of the entire system operation and achieve a final captured CO2 concentration value of ≥99% (including saturated water vapor), which meets the requirements of CO2 flooding for oil production in oil reservoirs in actual engineering, and effectively improves the CCUS-EOR (carbon dioxide capture storage and enhanced oil recovery) recovery rate and oilfield efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 This is a schematic structural diagram of a coal-fired flue gas CO2 capture system A according to an embodiment of the present invention;

[0077] Figure 2 An overall block diagram of the digital twin construction analysis of an embodiment of the present invention;

[0078] Figure 3 A flowchart of training and optimizing a three-dimensional visualization model according to an embodiment of the present invention;

[0079] Description of the markings in the figure:

[0080] A1-washing tower; A2-cooling tower; A3-fan; A4-absorption tower; A5-heat exchanger; A6-regeneration tower; A7-cooling / separator; A8-boosting device; A9-dehydration and liquefaction device;

[0081] 1-exhaust flue gas; 2-pretreated flue gas; 3-second return pipeline; 4-purified CO2 gas; 5-final captured CO2 gas. DETAILED DESCRIPTION

[0082] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0083] An embodiment of the present invention provides a digital twin construction method for a coal-fired flue gas CO2 capture device, which is applied to a coal-fired flue gas CO2 capture system A to achieve accurate advance prediction and analysis and regulation of the captured CO2 concentration value (referring to volume concentration, the same throughout the text).

[0084] like Figure 1 As shown, the system A includes a washing tower A1, a fan A3, an absorption tower A4, a heat exchanger A5, a regeneration tower A6, a cooling / separator A7, a booster device A8, and a dehydration and liquefaction device A9 which are connected in sequence; the washing tower A1 is connected to the coal-fired unit outside the device A, and is used to receive the exhaust flue gas 1 of the coal-fired unit; the washing tower A1 is also connected to a cooling tower A2, and the cooling tower A2 is connected back to the washing tower A1 through a first return pipe to form a circulating washing system for circulating washing of the exhaust flue gas 1; the absorption tower A4 is connected to a venting pipe for exhausting tail gas; the cooling / separator A7 is connected back to the regeneration tower A6 through a second return pipe 3 to return the separation liquid or the gas that does not meet the process requirements to the regeneration tower; the dehydration and liquefaction device A9 is connected to the pipeline system outside the system A, and is used to transport the finally captured CO2 through the pipeline system.

[0085] The process flow of the system A for capturing CO2 from coal-fired flue gas is as follows:

[0086] After the exhaust flue gas 1 from the coal-fired unit enters the system A, it is circulated and washed through the washing tower A1 and the cooling tower A2 to reduce the flue gas temperature and remove SO2 and other impurities in the flue gas, thereby obtaining the pre-treated flue gas 2;

[0087] The pre-treated flue gas 2 is pressurized by the fan A3 and then transported to the absorption tower A4;

[0088] The absorption tower A4 is provided with ammonia-poor liquid, which is used to absorb the captured CO2 gas and then converted into ammonia-rich liquid. The unabsorbed tail gas is discharged through the venting pipe; the discharged tail gas is monitored simultaneously to see whether it meets the following conditions: 40℃≤flue gas temperature<45℃, SO2 concentration in flue gas<35mg / Nm 3 If not, adjust the absorption liquid concentration and the number of washing cycles in the washing tower A1 to meet the process requirements;

[0089] The ammonia-rich liquid converted in the absorption tower A4 enters the regeneration tower A6 after being heated by heat exchanger A5, and is converted into ammonia-lean liquid after releasing the captured CO2 gas in the regeneration tower A6, and then returns to the absorption tower A4 after being cooled by heat exchanger A5;

[0090] The CO2 gas released from the regeneration tower A6 enters the cooling / separator A7 for cooling and separation to obtain purified CO2 gas and separation liquid; the separation liquid flows back to the regeneration tower through the second return pipe 3, and the concentration of the purified CO2 gas 4 is simultaneously monitored to see whether it meets the following conditions: CO2 concentration ≥ 94% (including saturated water vapor):

[0091] If not, the purified CO2 gas 4 is returned to the regeneration tower A6 through the second return pipe 3, and the dynamic parameters such as temperature / pressure / flow rate of the absorption tower A4, heat exchanger A5 and regeneration tower A6 are adjusted synchronously to meet the process requirements before being transported to the booster device;

[0092] If the conditions are met, the purified CO2 gas 4 is sequentially transported to the pressurizing device A8 for pressurization, and the dehydration and liquefaction device A9 for dehydration and liquefaction to obtain the final captured CO2 gas 5;

[0093] The final captured CO2 gas 5 is transported externally through a pipeline system, and the water content in the final captured CO2 gas 5 is simultaneously monitored to see whether it meets the following requirements: CO2 water content <100ppm. If not, the dynamic parameters of the dehydration and liquefaction device need to be predicted and adjusted to meet the process requirements.

[0094] Through the above process flow, a digital twin of the physical CO2 capture device is constructed in the virtual space, data exchange and virtual prejudgment between the digital twin capture device and the physical capture device are realized, and the early prediction of the CO2 capture rate and the intelligent regulation of the capture system A are realized.

[0095] like Figure 2 As shown, the digital twin is formed by the fusion and application of the twin space and the three-dimensional visualization model constructed based on the physical system A, and the digital twin is used to realize the mutual mapping of data, synchronous interaction of information and feedback between the twin space and the physical space.

[0096] The digital twin includes target twin, state and process twin and function twin;

[0097] The data model of the digital twin includes an acquisition layer, a transmission layer, a data layer and an application layer; wherein,

[0098] The acquisition layer is used to collect static parameters and dynamic data of each device in system A; wherein the static parameters are factory data parameters of each device in system A, including equipment material, equipment size, medium processed by the equipment, etc.; the dynamic parameters are real-time changing process parameters that need to be monitored during the process flow, including temperature, pressure, flow, etc. of each device;

[0099] The transmission layer is used for data transmission of dynamic / static parameters of each device in system A, and the transmission mode of the transmission layer adopts wired transmission;

[0100] The data layer is used for the integration, calculation, analysis, data twinning and storage management of the data of each device in system A;

[0101] The application layer provides users with real-time monitoring and advance prediction results of system A.

[0102] The digital twin construction method specifically includes the following steps:

[0103] Step S1. Establishing a twin space, specifically including:

[0104] Step S101. Collect the static parameters and dynamic parameters of each device in system A, build a three-dimensional model of each device in physical system A to obtain a twin physical system A, and define the dynamic / static parameter attributes of each device in the twin physical system A;

[0105] Step S102. Based on the twin physical system A, connect pipelines between the devices in the twin physical system A to obtain the twin space. The process flow formed by the connect pipelines is used to characterize the components and operation mechanism of the twin space, so as to feed back the dynamic / static parameter attributes and behaviors of each device to the twin space.

[0106] Step S2. Based on the twin space established in step S1, a three-dimensional visualization model is constructed, specifically including:

[0107] Step S201. 3D visualization model construction: construct a visualization model through a built-in lightweight 3D visualization engine to realize the display and roaming browsing of the twin space on the client or handheld terminal;

[0108] Step S202. Data association: Using data hyperlink technology, the data between the entity and the twin are interconnected. Each entity corresponds to its own 2D document and 3D visualization model. Clicking on any monomer in the 3D visualization model can quickly link to the corresponding 2D document and static / dynamic parameters, and visualize the entity space, the real-time status of the entity space and the process flow;

[0109] Among them, the input of the three-dimensional visualization model is: static and dynamic parameters of each device;

[0110] The output of the three-dimensional visualization model is: control variables that affect the final captured CO2 concentration value, including: flue gas temperature and SO2 concentration emitted by absorption tower A4, CO2 concentration in the purified CO2 gas 4, water content in the final captured CO2 gas 5, etc.

[0111] Step S203. Interactive control: Real-time data collection, remote communication and real-time update of dynamic data are realized through communication interfaces such as OPC, TCP / IP, MODUBUS, etc. That is, during the production operation, as the dynamic parameters continue to change, the two-dimensional documents and three-dimensional visualization models corresponding to each entity will also change accordingly, realizing real-time positioning and updating.

[0112] Step S3. Training and optimizing the three-dimensional visualization model: real-time acquisition of the original dynamic data of each device in the physical system A that affects the final captured CO2 concentration value, training and optimizing the twin space, and feeding back the optimization results to the three-dimensional visualization model; specifically including the following steps:

[0113] Step S301. At 30-minute intervals, collect and obtain the original dynamic data of each device in the physical system A that affects the final captured CO2 concentration value in real time. The original dynamic data particularly includes:

[0114] The exhaust gas temperature and SO2 concentration emitted by absorption tower A4, and the corresponding absorption liquid concentration and number of wash cycles in washing tower A1;

[0115] The concentration of CO2 in the purified CO2 gas 4, and the corresponding dynamic parameters such as temperature, pressure and flow rate of the absorption tower A4, the heat exchanger A5 and the regeneration tower A6;

[0116] The water content in the final captured CO2 gas 5, and the corresponding dynamic parameters such as temperature, pressure, flow rate, etc. of the dehydration and liquefaction device A9;

[0117] Pay special attention to and collect the above parameters. For each parameter, data from 48 time points is collected every day, which will generate a large amount of data, with tens of thousands of data collected every day. By checking data consistency, processing invalid data values, identifying abnormal data, and using multiple repeated review, verification and annotation methods, the large amount of original dynamic data collected in real time is cleaned and screened to obtain valid original data.

[0118] Step S302. In the twin space, a two-layer LSTM long short-term memory network model is constructed based on valid original data;

[0119] Step S303: Use the valid original data to train the double-layer LSTM long short-term memory network model to obtain an optimized double-layer LSTM long short-term memory network model, and feed the optimization result back to the three-dimensional visualization model, specifically including:

[0120] Step S3031. Use the valid original data that affects the final captured CO2 concentration value as the input value of the double-layer LSTM long short-term memory network model, and automatically calculate and generate the predicted value of the control variable that affects the final captured CO2 concentration value through the model;

[0121] Step S3032. Compare the predicted value of the control variable that affects the final captured CO2 concentration value with the actual collected value, and use the mean relative error MRE as an evaluation standard to perform error comparison analysis;

[0122] Step S3033. Repeat steps S3031-S3032 to train and optimize the double-layer LSTM long short-term memory network model for a long time (e.g., 1-3 years); at the beginning, the mean relative error MRE value is generally large, and with continuous training and optimization, the mean relative error MRE value will continue to decrease until it drops below 10%, and the optimized double-layer LSTM long short-term memory network model can be obtained, and an expert experience library is automatically formed in the model;

[0123] Step S3034. The twin space synchronously feeds back the optimization results to the three-dimensional visualization model to achieve training optimization of the three-dimensional visualization model.

[0124] Step S4. Verify the three-dimensional visualization model: Verify the correctness of the three-dimensional visualization model. If the prediction conditions of the three-dimensional visualization model are met, that is, the predicted value of the control variable affecting the final captured CO2 concentration value by the three-dimensional visualization model is close to the actual collected value, and the error between the two is ≤10%, proceed to the next step S5; if not, return to step S1.

[0125] Step S5. Predict and regulate the final captured CO2 concentration value in the physical system A in real time based on the verified three-dimensional visualization model; specifically, the following steps are included:

[0126] Step S501. Collect the dynamic parameters of each device in the physical system A in real time and map them to the twin space;

[0127] Step S502: The twin space predicts the predicted value of the control variable that affects the final captured CO2 concentration value according to the mapped dynamic parameters;

[0128] Step S503. The twin space compares the predicted value of the control variable that affects the final captured CO2 concentration value with the judgment threshold:

[0129] If the tail gas discharged from the absorption tower A4 meets the following conditions: 40℃≤ flue gas temperature<45℃, SO2 concentration<35mg / Nm 3 , then execute the subsequent CO2 capture process; if not, predict and adjust the absorption liquid concentration and the number of cycle washing in the washing tower A1 to meet the process requirements;

[0130] If the purified CO2 gas 4 cooled and separated by the cooler / separator A7 meets the following conditions: CO2 concentration ≥ 94% (including saturated water vapor), the subsequent CO2 capture process is executed; if not, the dynamic parameters such as temperature, pressure, flow rate, etc. of the absorption tower A4, heat exchanger A5 and regeneration tower A6 are predicted and adjusted to meet the process requirements;

[0131] If the captured CO2 gas 5 obtained after dehydration and liquefaction of CO2 in the dehydration and liquefaction device A9 meets the following conditions: CO2 water content <100ppm, the subsequent CO2 capture process is executed; if not, the temperature, pressure, flow rate and other dynamic parameters of the dehydration and liquefaction device A9 need to be predicted and adjusted to meet the process requirements;

[0132] Among them, the method for obtaining the judgment threshold of the control variables that affect the final captured CO2 concentration value is: based on the dynamic data collected in real time over a long period of time in the physical system A (the collection time can be as long as 1-3 years), the twin space is continuously trained and optimized, and finally the critical value of the above-mentioned control variables that can meet the captured CO2 concentration value is obtained, which is the judgment threshold.

[0133] Through the above method, it can be ensured that the final captured CO2 concentration value can stably meet the target value, and the final CO2 capture concentration value can be achieved ≥ 99% (including saturated water vapor), thereby significantly improving the working efficiency of the entire system and reducing operating costs.

[0134] In addition, in some other embodiments, a positioning alarm module is also embedded in the three-dimensional visualization model of the digital twin; the positioning alarm module obtains the normal operating value range of the dynamic parameters of each device in the physical system A based on the trained and optimized three-dimensional visualization model; at the same time, the positioning alarm module obtains the dynamic parameters of each device collected in the physical system A in real time through data transmission, and compares them with the normal operating value range of each dynamic parameter in real time;

[0135] When the positioning alarm module monitors that the dynamic parameters of each device in system A have operating data that exceeds the normal operating value range, a positioning alarm will be issued on the display screen of the three-dimensional visualization model, and auxiliary suggestions and decisions will be given; based on the auxiliary suggestions and decisions given, the system will automatically perform parameter adjustment and maintenance, or the fault will be manually corrected.

[0136] The embodiment of the present invention is equipped with a positioning alarm module in the three-dimensional visualization model, conducts big data research based on the digital twin, and establishes an intelligent operation and maintenance main line of "real-time data - positioning alarm - rapid prediction - auxiliary decision making". Through the advance prediction of the digital twin, the abnormal operating parameters of the device can be discovered and adjusted in time, which can effectively improve the reliability of the entire system operation.

[0137] The above description is only an embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement and improvement made within the application scope of the present invention shall be included in the protection scope of the present invention.

Claims

1. A digital twin construction method for a coal-fired flue gas CO2 capture device, applied to a coal-fired flue gas CO2 capture system A, characterized in that: The system A comprises a washing tower, a fan, an absorption tower, a heat exchanger, a regeneration tower, a cooling / separator, a pressurizing device, a dehydration and liquefaction device connected in sequence; the washing tower is connected to a coal-fired unit outside the device A; the washing tower is also connected to a cooling tower, which is connected back to the washing tower through a first return pipe to form a circulating washing system; the absorption tower is connected to a venting pipe; the cooling / separator is connected back to the regeneration tower through a second return pipe; the dehydration and liquefaction device is connected to a pipeline system outside the system A, and is used to transport the finally captured CO2 through the pipeline system; The digital twin construction method specifically includes the following steps: Step S1. Collect static parameters and dynamic parameters of each device in system A, and establish a twin space based on the physical system A; The static parameters are factory data parameters of each device in system A, and the dynamic parameters are real-time changing process parameters that need to be monitored during the process flow in system A; Step S2. Based on the twin space established in step S1, construct a three-dimensional visualization model; Step S3. Training and optimizing the three-dimensional visualization model: collecting the original dynamic data of each device that affects the final captured CO2 concentration value in the physical system A in real time, training and optimizing the twin space, and feeding back the optimization results to the three-dimensional visualization model; Step S4. Verify the three-dimensional visualization model: If the three-dimensional visualization model meets the prediction conditions for correctness verification, proceed to the next step S5; if not, return to step S1; Step S5. Based on the verified three-dimensional visualization model, the final captured CO2 concentration value in the physical system A is predicted and regulated in real time.

2. The method according to claim 1, characterized in that: The digital twin is formed by the mutual fusion and application of the twin space and the three-dimensional visualization model constructed based on the physical system A, and realizes the mutual mapping of data, synchronous information interaction and feedback between the twin space and the physical space through the digital twin; The digital twin includes target twin, state and process twin and function twin; The data model of the digital twin includes an acquisition layer, a transmission layer, a data layer and an application layer; wherein, The acquisition layer is used to collect static parameters and dynamic data of each device in system A; The transmission layer is used for data transmission of dynamic / static parameters of each device in system A; The data layer is used for the integration, calculation, analysis, data twinning and storage management of the data of each device in system A; The application layer provides users with real-time monitoring and advance prediction results of system A.

3. The method according to claim 1, characterized in that: The process flow of the system A for capturing CO2 from coal-fired flue gas is as follows: After the exhaust gas from the coal-fired unit enters system A, it is circulated and washed through the washing tower and cooling tower, and after cooling and impurity removal, the pre-treated flue gas is obtained; After pretreatment, the flue gas is pressurized by a fan and transported to the absorption tower; The absorption tower is provided with a lean ammonia liquid, which is used to absorb the captured CO2 gas and then converted into a rich ammonia liquid, and the unabsorbed tail gas is discharged through the venting pipe; the discharged tail gas is simultaneously monitored to see whether it meets the set flue gas temperature and SO2 concentration. If not, the absorption liquid concentration and the number of cycle washing in the washing tower are predicted and adjusted to meet the process requirements; The ammonia-rich liquid converted in the absorption tower enters the regeneration tower after being heated by the heat exchanger, and is converted into ammonia-lean liquid after releasing the captured CO2 gas in the regeneration tower, and then returns to the absorption tower after being cooled by the heat exchanger; The CO2 gas released by the regeneration tower enters the cooling / separator for cooling and separation to obtain purified CO2 gas and separation liquid; the separation liquid flows back to the regeneration tower through the second return pipe, and synchronously monitors whether the concentration of the purified CO2 gas meets the set CO2 concentration: if not, the purified CO2 gas is returned to the regeneration tower through the second return pipe, and the dynamic parameters of the absorption tower, the heat exchanger and the regeneration tower are synchronously adjusted to meet the process requirements before being transported to the booster; if satisfied, the purified CO2 gas is sequentially transported to the booster for pressurization, the dehydration and liquefaction device for dehydration and liquefaction to obtain the final captured CO2 gas; The final captured CO2 gas is transported externally through a pipeline system, and the water content in the final captured CO2 gas is simultaneously monitored to see if it meets the set water content. If not, the dynamic parameters of the dehydration and liquefaction device need to be predicted and adjusted to meet the process requirements.

4. The method according to claim 1, characterized in that: The step S1 specifically includes the following steps: Step S101. Collect the static parameters and dynamic parameters of each device in system A, build a three-dimensional model of each device in physical system A to obtain a twin physical system A, and define the dynamic / static parameter attributes of each device in the twin physical system A; Step S102. Based on the twin physical system A, connect pipelines between the devices in the twin physical system A to obtain the twin space, and characterize the components and operation mechanism of the twin space through the process flow formed by the connect pipelines.

5. The method according to claim 1, characterized in that The step S2 specifically includes the following steps: Step S201. 3D visualization model construction: construct a visualization model through a built-in lightweight 3D visualization engine to realize the display and roaming browsing of the twin space on the client or handheld terminal; Step S202. Data association: Using data hyperlink technology, the data between the entity and the twin are interconnected. Each entity corresponds to its own 2D document and 3D visualization model. Clicking on any monomer in the 3D visualization model can quickly link to the corresponding 2D document and static / dynamic parameters, and visualize the entity space, the real-time status of the entity space and the process flow; Among them, the input of the three-dimensional visualization model is: static and dynamic parameters of each device; The outputs of the 3D visualization model are: control variables that affect the final captured CO2 concentration value; Step S203. Interactive control: Realize real-time data collection, remote communication and real-time update of dynamic data through communication interfaces such as OPC, TCP / IP, MODUBUS, etc.

6. The method according to claim 5, characterized in that In step S202, the control variables affecting the final captured CO2 concentration value include: the flue gas temperature and SO2 concentration emitted by the absorption tower, the CO2 concentration in the purified CO2 gas, and the water content in the final captured CO2 gas.

7. The method according to claim 1, characterized in that The step S3 specifically includes the following steps: Step S301. Real-time acquisition of original dynamic data of each device in the physical system A that affects the final captured CO2 concentration value, wherein the original dynamic data at least includes: The exhaust gas temperature and SO2 concentration emitted by the absorption tower, as well as the absorption liquid concentration and the number of wash cycles in the corresponding washing tower; The CO2 concentration in the purified CO2 gas and the corresponding dynamic parameters of the absorption tower, heat exchanger and regeneration tower; The water content in the final captured CO2 gas and the corresponding dynamic parameters of the dehydration and liquefaction equipment; Clean and filter the original dynamic data to obtain valid original data; Step S302. In the twin space, a two-layer LSTM long short-term memory network model is constructed based on the valid original data; Step S303. Use valid original data to train a double-layer LSTM long short-term memory network model to obtain an optimized double-layer LSTM long short-term memory network model, and feed the optimization result back to the three-dimensional visualization model.

8. The method according to claim 7, characterized in that The step S303 specifically includes the following steps: Step S3031. Use the valid original data that affects the final captured CO2 concentration value as the input value of the double-layer LSTM long short-term memory network model, and automatically calculate and generate the predicted value of the control variable that affects the final captured CO2 concentration value through the model; Step S3032. Compare the predicted value of the control variable that affects the final captured CO2 concentration value with the actual collected value, and use the mean relative error MRE as an evaluation standard to perform error comparison analysis; Step S3033. Repeat steps S3031-S3032 to train and optimize the double-layer LSTM long short-term memory network model for a certain period of time; until the mean relative error MRE value is reduced to the set target value, the optimized double-layer LSTM long short-term memory network model can be obtained, and an expert experience library is automatically formed in the model; Step S3034. The twin space synchronously feeds back the optimization results to the three-dimensional visualization model to achieve training optimization of the three-dimensional visualization model.

9. The method according to claim 8, characterized in that In step S3033, the mean relative error MRE value is reduced to ≤10%, and an optimized double-layer LSTM long short-term memory network model can be obtained.

10. The method according to claim 1, characterized in that In step S4, the prediction condition that the three-dimensional visualization model satisfies the correctness verification is: The predicted value of the control variable affecting the final captured CO2 concentration value by the three-dimensional visualization model is close to the actual collected value, and the error between the two is ≤10%.

11. The method according to any one of claims 1 to 10, characterized in that: The step S5 specifically includes the following steps: Step S501. Collect the dynamic parameters of each device in the physical system A in real time and map them to the twin space; Step S502: The twin space predicts the predicted value of the control variable that affects the final captured CO2 concentration value according to the mapped dynamic parameters; Step S503. The twin space compares the predicted value of the control variable that affects the final captured CO2 concentration value with the judgment threshold: If the tail gas emitted by the absorption tower meets the following conditions: the lower limit of the set flue gas temperature judgment threshold ≤ flue gas temperature < the upper limit of the set flue gas temperature judgment threshold, and the SO2 concentration < the set SO2 concentration judgment threshold, then the subsequent CO2 capture process is executed; if not, the absorption liquid concentration and the number of cycle washing times in the washing tower are predicted and adjusted to meet the process requirements; If the purified CO2 gas separated by cooling / separation meets the following conditions: CO2 concentration ≥ the set CO2 concentration judgment threshold, the subsequent CO2 capture process is executed; If not, predict and adjust the dynamic parameters of the absorption tower, heat exchanger and regeneration tower to meet the process requirements; If the captured CO2 gas obtained after dehydration and liquefaction of CO2 in the dehydration and liquefaction device meets the following conditions: CO2 water content < set water content judgment threshold, the subsequent CO2 capture process will be executed; if not, the dynamic parameters of the dehydration and liquefaction device need to be predicted and adjusted to meet the process requirements.

12. The method according to claim 11, characterized in that The method for obtaining the judgment threshold of the control variable that affects the final captured CO2 concentration value is: Based on the dynamic data collected in real time over a long period of time in the physical system A, the twin space is continuously trained and optimized, and finally the critical value of the control variable that can meet the captured CO2 concentration value is obtained, which is the judgment threshold.

13. The method according to claim 12, characterized in that The judgment threshold of the control variable that affects the final captured CO2 concentration value is: The tail gas discharged from the absorption tower must meet the requirements of 40℃≤ flue gas temperature<45℃, and SO2 concentration in flue gas<15mg / Nm 3 ; The purified CO2 gas separated by cooling / separation in the cooling / separation device must meet the following requirements: CO2 concentration ≥ 94%; After dehydration and liquefaction of CO2 in the dehydration and liquefaction device, the final captured CO2 gas must meet the following requirements: CO2 water content <100ppm.

14. The method according to claim 1, characterized in that A positioning alarm module is also embedded in the three-dimensional visualization model of the digital twin; The positioning alarm module obtains the normal operating value range of the dynamic parameters of each device in the physical system A based on the trained and optimized three-dimensional visualization model; at the same time, the positioning alarm module obtains the dynamic parameters of each device collected in the physical system A in real time through data transmission, and compares them with the normal operating value range of each dynamic parameter in real time; When the positioning alarm module monitors that the dynamic parameters of each device in system A have operating data that exceeds the normal operating value range, a positioning alarm is issued on the display screen of the three-dimensional visualization model, and auxiliary suggestions and decisions are given; Based on the auxiliary suggestions and decisions given, the system automatically adjusts parameters for maintenance or manually rectifies the fault.

Images