Method and device for determining the degree of degradation of a decarbonation absorbent, and ship management system

Abstract

The application provides a kind of decarburization absorbent degradation degree determination method, device and ship management system, the method includes the system parameter sample from the carbon dioxide capture and storage system and target result is input to the neural network model constructed, and the neural network model is trained, to obtain the target parameter of the neural network model;Based on the target parameter, the system parameter monitored by the carbon dioxide capture and storage system is input to the neural network model that has been trained, to obtain the predicted value of the decarburization absorbent degradation degree.The application can directly use the system parameter output by the carbon dioxide capture and storage system to provide real-time prediction of the decarburization absorbent degradation degree through the neural network model.

Description

Technical Field

[0001] This invention relates to the field of greenhouse gas control technology, and in particular to a method, apparatus and ship management system for measuring the degradation degree of decarbonization absorbent. Background Technology

[0002] The combustion of fossil fuels is the primary source of atmospheric CO2. Various equipment used in human production and daily life, especially transportation vehicles such as ships, burn large quantities of hydrocarbon fuels annually, thus emitting significant amounts of CO2 greenhouse gases into the atmosphere, contributing to the continuous rise in global average temperatures. From a technical perspective, compared to other technical solutions such as physical adsorption, chemical absorption methods based on organic amine absorbents have attracted considerable attention due to their high system maturity and economic advantages. However, tests on carbon dioxide capture and storage systems (CCS systems) in land-based power plants have demonstrated the degradation and deterioration of organic amine absorbents. Therefore, in shipboard CCS systems, there is an urgent need to develop a method for on-site, real-time determination of the degradation degree of decarbonization absorbents. Summary of the Invention

[0003] This invention provides a method, apparatus, and ship management system for determining the degradation degree of decarbonization absorbents, in order to solve the problem in the prior art that it is impossible to know in real time whether the decarbonization absorbent has degraded or deteriorated.

[0004] In a first aspect, the present invention provides a method for determining the degradation degree of a carbon removal absorbent, applied to a carbon dioxide capture and storage system, the method comprising:

[0005] The system parameter samples and target results from the carbon dioxide capture and storage system are input into the constructed neural network model, and the neural network model is trained to obtain the target parameters of the neural network model.

[0006] Based on the target parameters, the system parameters monitored by the carbon dioxide capture and storage system are input into the trained neural network model to obtain a predicted value of the degradation degree of the carbon removal absorbent.

[0007] In one embodiment of the present invention, the system parameters include state parameters reflecting the carbon dioxide capture and storage system and time series parameters reflecting the degradation process of the decarbonization absorbent.

[0008] The state parameters include one or more combinations of the following:

[0009] The height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the tower;

[0010] Operating parameters of pressure, temperature, and reboiler power within the absorption tower and regeneration tower;

[0011] The types, initial concentrations, flow rates, physical properties, and intrinsic parameters of organic amines;

[0012] The time series parameters include one or more combinations of the following:

[0013] The types and concentrations of the overhead gases in the absorption tower and regeneration tower;

[0014] The types and concentrations of degradation products at the bottom of the absorption tower and regeneration tower.

[0015] In one embodiment of the present invention, the step of inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model includes:

[0016] The system parameter sample x = [x l1 x l2 x l3 x l4 (4), x l5 The target result y, representing the degree of degradation, is input into the neural network model;

[0017] Where, x l1 This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers, x. l2 The x represents the operating parameters of pressure, temperature, and reboiler power within the absorber and regeneration towers. l3 The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines, x l4 (t) represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers, x l5 (t) represents the cumulative time series data of the types and concentration changes of degradation products at the bottom of the absorption tower and the regeneration tower.

[0018] In one embodiment of the present invention, the neural network model includes a feedforward network, a recurrent network, and a splicing module. The step of inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model and training the neural network model to obtain the target parameters of the neural network model further includes:

[0019] x represents the state parameter l1 x l2 and x l3 The input is fed into the feedforward network, and after undergoing nonlinear transformation by the feedforward network, the first result y1 is output.

[0020] The cumulative parameter x representing the time series parameters l4 (t) and x l5The input (t) is fed into the cyclic network, and after undergoing a nonlinear transformation by the cyclic network, the second result y2(t) is output.

[0021] The first result y1 and the second result y2(t) are input into the splicing module to obtain the prediction result y^=f(x;θ), where θ represents all the training parameters of the feedforward network and the recurrent network, and f represents a function;

[0022] Based on the predicted result y^ and the target result y, the training parameters θ are iteratively optimized and solved using a regression loss function to obtain the target parameters θ~.

[0023] In one embodiment of the present invention, the step of iteratively optimizing the training parameters θ using a regression loss function to obtain the target parameters θ~ includes:

[0024] Calculate the loss value for the t-th iteration: loss = ψ(y^, y), where ψ represents the loss function;

[0025] Calculate the gradient of the loss function with respect to the training parameters θ.

[0026] The training parameters θ are updated using gradient descent with a preset learning rate α. t+1 =θ t -αg, to obtain θ t+1 Then, the process proceeds to the (t+1)th iteration until the loss value is lower than the set threshold, at which point the target parameter θ is obtained.

[0027] The trained neural network model is represented as y^=f(x;θ~).

[0028] In one embodiment of the present invention, the state parameters and the time series parameters are obtained by sensors on the carbon dioxide capture and storage system, and the target result is obtained by spectral analysis.

[0029] Secondly, the present invention provides a device for measuring the degradation degree of a decarbonization absorbent, the device comprising:

[0030] The training module is used to input system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model, and to train the neural network model to obtain the target parameters of the neural network model.

[0031] The prediction module is used to input the system parameters monitored by the carbon dioxide capture and storage system into the trained neural network model based on the target parameters, so as to obtain a predicted value of the degradation degree of the decarbonization absorbent.

[0032] In one embodiment of the present invention, the system parameter sample includes state parameters reflecting the carbon dioxide capture and storage system and time series parameters reflecting the degradation process of the carbon removal absorbent. The training module is further used for:

[0033] The system parameter sample x = [x l1 x l2 x l3 x l4 (t), x l5 The target result y, representing the degree of degradation, is input into the neural network model;

[0034] Where, x l1 This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers, x. l2 The x represents the operating parameters of pressure, temperature, and reboiler power within the absorber and regeneration towers. l3 The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines, x l4 (t) represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers, x l5 (t) represents the cumulative time series data of the types and concentration changes of degradation products at the bottom of the absorption tower and the regeneration tower.

[0035] In one embodiment of the present invention, the neural network model includes a feedforward network, a recurrent network, and a splicing module, and the training module is further used for:

[0036] x represents the state parameter l1 x l2 and x l3 The input is fed into the feedforward network, and after undergoing nonlinear transformation by the feedforward network, the first result y1 is output.

[0037] x represents the time series parameters l4 (t) and x l5 The input (t) is fed into the cyclic network, and after undergoing a nonlinear transformation by the cyclic network, the second result y2(t) is output.

[0038] The first result y1 and the second result y2(t) are input into the splicing module to obtain the prediction result y^=f(x;θ), where θ represents all the training parameters of the feedforward network and the recurrent network, and f represents a function;

[0039] Based on the predicted result y^ and the target result y, the training parameters θ are iteratively optimized and solved using a regression loss function to obtain the target parameters θ~.

[0040] Thirdly, the present invention provides a ship management system for performing the method for determining the degree of degradation of decarbonizing absorbent as described in any of the first aspects.

[0041] This invention provides a method, apparatus, and ship management system for measuring the degradation degree of a carbon capture and storage (CCS) absorbent. By constructing a neural network model for predicting the degradation degree of the CCS absorbent and training the model, a trained neural network model can be obtained. This neural network model can directly use the system parameters output by the CCS system to provide real-time prediction of the degradation degree of the CCS absorbent, solving the problem in the prior art that it is impossible to know in real time whether the CCS absorbent has degraded or deteriorated. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0043] Figure 1 This is a flowchart of the method for determining the degradation degree of decarbonization absorbent provided by the present invention;

[0044] Figure 2 This is a schematic diagram of the neural network model training process provided in an embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of the structure of the neural network model provided in an embodiment of the present invention;

[0046] Figure 4 This is a schematic diagram of the prediction process of the neural network model provided by the present invention;

[0047] Figure 5 This is a schematic diagram of the structure of the device for measuring the degradation degree of decarbonization absorbent provided by the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0049] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein.

[0050] To address the problem in existing technologies that cannot detect in real time whether a decarbonization absorbent has degraded or deteriorated, this invention provides a method, apparatus, and ship management system for measuring the degree of degradation of a decarbonization absorbent. By constructing a neural network model to predict the degree of degradation of the decarbonization absorbent and training the model, a trained neural network model can be obtained. This neural network model can directly utilize system parameters output by the carbon dioxide capture and storage system (CCS system) to provide real-time prediction of the degree of degradation of the decarbonization absorbent, thus solving the problem in existing technologies that cannot detect in real time whether a decarbonization absorbent has degraded or deteriorated.

[0051] The following is combined with Figures 1-5 The present invention describes a method, apparatus, and ship management system for determining the degradation degree of a decarbonization absorbent.

[0052] Please refer to Figure 1 , Figure 1 This is a flowchart of a method for determining the degradation degree of a carbon dioxide absorbent provided by the present invention. A method for determining the degradation degree of a carbon dioxide absorbent, applied to a carbon dioxide capture and storage system (i.e., a CCS system), the method comprising:

[0053] Step 110: Input the system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model, and train the neural network model to obtain the target parameters of the neural network model.

[0054] The system parameter samples are sample data from the ship's CCS system used to train the neural network model. The system parameters fall into two categories: the first category reflects the state parameters of the carbon dioxide capture and storage system, and the second category reflects the time-series parameters of the decarbonization absorbent degradation process.

[0055] For example, the state parameters include one or more combinations of the following:

[0056] Structural parameters of the absorption tower and regeneration tower, including tower height, diameter, and internal equipment;

[0057] Operating parameters such as pressure, temperature, and reboiler power within the absorption tower and regeneration tower;

[0058] The physical properties and intrinsic parameters of organic amines, such as type, initial concentration, and flow rate;

[0059] For example, the time series parameters include one or more combinations of the following:

[0060] The types and concentrations of the overhead gases in the absorption tower and regeneration tower;

[0061] The types and concentrations of degradation products at the bottom of the absorption tower and regeneration tower.

[0062] Therefore, it can be seen that the input of the neural network model constructed in this invention comes from the system parameters of the ship's CCS system.

[0063] Step 120: Based on the target parameters, the system parameters monitored by the carbon dioxide capture and storage system are input into the trained neural network model to obtain a predicted value of the degradation degree of the carbon removal absorbent.

[0064] Therefore, the output of the neural network model constructed in this invention reflects the actual degradation state of the decarbonization absorbent (e.g., organic amine absorbent) in the ship's CCS system. During the offline training phase, the actual concentration of the circulating organic amine absorbent is tested through periodic sampling. In the online monitoring phase, the predicted value of the organic amine absorbent's degradation level is output. The intermediate end of the neural network model connects the input and output ends. During the offline training phase, a certain amount of training data is used to mine the mapping relationship between the system parameters at the model input end and the degradation level of the decarbonization absorbent at the output end. In the online testing phase, the training parameters at the intermediate end are frozen as target parameters. Then, by inputting the system parameters of the actual ship, the predicted value of the decarbonization absorbent's degradation level can be dynamically output in real time.

[0065] The following describes the training of the model in step 110 and the model prediction in step 120.

[0066] Please refer to Figure 2 , Figure 3 As shown, Figure 2 This is a schematic diagram of the neural network model training process provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the neural network model provided in an embodiment of the present invention. Step 110 above, which involves inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model, includes:

[0067] Step 210, set the system parameter sample x = [x l1 x l2 x l3 x l4 (t), x l5 The target result y, representing the degree of degradation, is input into the neural network model.

[0068] Where, x l1This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers, x. l2 The x represents the operating parameters of pressure, temperature, and reboiler power within the absorber and regeneration towers. l3 The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines, x l4 (t) represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers, x l5 (t) represents the cumulative time series data of the types and concentration changes of degradation products at the bottom of the absorption tower and the regeneration tower.

[0069] For example, the above represents the state parameter x. l1 x l2 and x l3 and the cumulative parameter x representing the time series parameters l4 (t) and x l5 (t) can be measured by sensors in the ship's CCS system.

[0070] For example, the above-mentioned cumulative parameter x l4 (t) and x l5 (t) requires the cumulative value from the start to the current time as input, i.e. Where x4(t) represents the concentration of the gas at the top of the absorption tower and regeneration tower at each monitoring instant. Similarly... Where x5(t) is the concentration of degradation products at the bottom of the absorption tower and regeneration tower at each monitoring instant.

[0071] For example, the target result y can be obtained by spectral analysis, and is a spectral analysis measurement value that can be used to measure the training effect of the neural model.

[0072] For example, in step 110 above, the neural network model includes a feedforward network, a recurrent network, and a splicing module. The step of inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model and training the neural network model to obtain the target parameters of the neural network model further includes:

[0073] Step 220, x, representing the state parameter l1 x l2 and x l3 The input is fed into the feedforward network, and after undergoing nonlinear transformation by the feedforward network, the first result y1 is output.

[0074] For example, the feedforward network has two hidden layers, x1 = [x l1 x l2 x l3After two layers of nonlinear transformation, the first result y1 of the feedforward network output is obtained.

[0075] Step 230, the x representing the time series parameters l4 (t) and x l5 The input (t) is fed into the cyclic network, and after undergoing a nonlinear transformation by the cyclic network, the second result y2(t) is output.

[0076] For example, the recurrent network uses the LSTM (Long Short-Term Memory) model, x2 = [x l4 (t), x l5 After the nonlinear transformation of the LSTM model, the second result y2(t) of the recurrent network output is obtained.

[0077] Step 240: Input the first result y1 and the second result y2(t) into the splicing module to obtain the prediction result y^=f(x;θ), where θ represents all the training parameters of the feedforward network and the recurrent network, and f represents a function.

[0078] For example, the first result y1 output by the feedforward network and the second result y2(t) output by the recurrent network are concatenated. The concatenation method includes, but is not limited to, direct multiplication or combination into a vector. For example, if direct multiplication is used, y(t) = y1 × y2(t), that is, y^ = f(x; θ).

[0079] Step 250: Based on the predicted result y^ and the target result y, the training parameters θ are iteratively optimized using a regression loss function to obtain the target parameters θ~.

[0080] like Figure 2 , Figure 3 As shown, during offline training, based on the predicted result y^ and the target result y, the model parameters θ can be iteratively optimized and solved by selecting a suitable regression loss function, such as MAE (Mean absolute error, i.e., the average of the absolute errors between the predicted value and the true value) or MSE (Mean squared error, i.e., the average of the absolute squared errors between the predicted value and the true value), using the gradient descent method.

[0081] Specifically, the loss value loss = ψ(y^, y) is calculated for the t-th iteration, where ψ represents a loss function. Then, the gradient of the loss function with respect to the training parameters θ is calculated. Then, using a preset learning rate α, the training parameters θ are updated using gradient descent. t+1 =θ t-αg, to obtain θ t+1 Then, the process proceeds to the (t+1)th iteration until the loss value is lower than the set threshold, at which point the target parameter θ is obtained. θ is the optimal parameter for model training. Therefore, the trained neural network model is represented as y^=f(x;θ~).

[0082] It should be noted that, in order to construct the mapping relationship between the system parameters at the input end of the neural network model and the actual degradation state of the decarbonization absorbent at the output end of the model, the core intermediate end of the model can be based on a traditional regressor or a deep neural network, depending on the system prediction accuracy requirements and the relevant hardware (such as a graphics processing unit, GPU) conditions.

[0083] For example, when the system does not require high accuracy in predicting the degree of degradation or when the computing power of the ship's hardware is limited, various traditional regressors, including SVR (support vector regression) and Regression Forest, can be used. However, when the system requires high accuracy in predicting the degree of degradation and the computing power of the ship's hardware is sufficient, deep models, including but not limited to DNN (deep neural network), RNN (recurrent neural network), CNN (convolutional neural network), transformer (a model that uses attention mechanism to improve the training speed of the model), Bayesian neural network, and other neural network frameworks, can be used.

[0084] Please refer to Figure 4 , Figure 4 This is a schematic diagram of the neural network model prediction process provided by the present invention. During the online prediction process, the system parameter x = [x] is monitored in real time by sensors of the ship's CCS system. l1 x l2 x l3 x l4 (t), x l5 The input (t) is fed into the trained neural network model f(x; θ~) to obtain the predicted value y^=f(x) of the degree of degradation of the decarbonization absorbent.

[0085] In summary, during the offline training phase, this invention trains the constructed neural model based on the coefficient parameters reflected by sensor measurements and the parameters of the degree of degradation of the decarbonizing absorbent reflected by sampling detection. After the model training is completed, the target parameters of the trained model are frozen. Then, during online monitoring on a real ship, the predicted value of the degree of degradation of the decarbonizing absorbent can be output by inputting the corresponding system parameters.

[0086] Therefore, the method for determining the degradation degree of decarbonizing absorbent described in this invention can directly predict the degradation degree of decarbonizing absorbent using the system parameters of the ship's CCS system, solving the problem that existing technologies such as spectral analysis cannot be applied in ship space. Furthermore, the input parameters are all fixed system parameters or parameters that must be monitored, thus eliminating the need for additional equipment investment, resulting in high economic efficiency. It also provides real-time dynamic prediction information on the degradation status of the decarbonizing absorbent, facilitating operation for crew members.

[0087] The following describes the apparatus for measuring the degradation degree of decarbonizing absorbent provided by the present invention. The apparatus for measuring the degradation degree of decarbonizing absorbent described below can be referred to in correspondence with the method for measuring the degradation degree of decarbonizing absorbent described above.

[0088] Please refer to Figure 5 , Figure 5 This is a schematic diagram of the structure of the decarbonization degree measuring device for the decarbonization absorbent provided by the present invention. The decarbonization degree measuring device 500 includes a training module 510 and a prediction module 520.

[0089] For example, the training module 510 is used to input system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model, and train the neural network model to obtain the target parameters of the neural network model.

[0090] For example, the prediction module 520 is used to input the system parameters monitored by the carbon dioxide capture and storage system into the trained neural network model based on the target parameters to obtain a predicted value of the degradation degree of the carbon removal absorbent.

[0091] For example, the system parameters include state parameters reflecting the carbon dioxide capture and storage system and time-series parameters reflecting the degradation process of the carbon removal absorbent.

[0092] The state parameters include one or more combinations of the following:

[0093] The height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the tower;

[0094] Operating parameters of pressure, temperature, and reboiler power within the absorption tower and regeneration tower;

[0095] The types, initial concentrations, flow rates, physical properties, and intrinsic parameters of organic amines;

[0096] The time series parameters include one or more combinations of the following:

[0097] The types and concentrations of the overhead gases in the absorption tower and regeneration tower;

[0098] The types and concentrations of degradation products at the bottom of the absorption tower and regeneration tower.

[0099] For example, training module 510 is also used for:

[0100] The system parameter sample x = [x l1 x l2 x l3 x l4 (t), x l5 The target result y, representing the degree of degradation, is input into the neural network model;

[0101] Where, x l1 This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers, x. l2 The x represents the operating parameters of pressure, temperature, and reboiler power within the absorber and regeneration towers. l3 The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines, x l4 (t) represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers, x l5 (t) represents the cumulative time series data of the types and concentration changes of degradation products at the bottom of the absorption tower and the regeneration tower.

[0102] For example, training module 510 is also used for:

[0103] x represents the state parameter l1 x l2 and x l3 The input is fed into the feedforward network, and after undergoing nonlinear transformation by the feedforward network, the first result y1 is output.

[0104] x represents the time series parameters l4 (t) and x l5 The input (t) is fed into the cyclic network, and after undergoing a nonlinear transformation by the cyclic network, the second result y2(t) is output.

[0105] The first result y1 and the second result y2(t) are input into the splicing module to obtain the prediction result y^=f(x;θ), where θ represents all the training parameters of the feedforward network and the recurrent network, and f represents a function;

[0106] Based on the predicted result y^ and the target result y, the training parameters θ are iteratively optimized and solved using a regression loss function to obtain the target parameters θ~.

[0107] For example, training module 510 is also used for:

[0108] Calculate the loss value for the t-th iteration: loss = ψ(y^, y), where ψ represents the loss function;

[0109] Calculate the gradient of the loss function with respect to the training parameters θ.

[0110] The training parameters θ are updated using gradient descent with a preset learning rate α. t+1 =θ t -αg, to obtain θ t+1 Then, the process proceeds to the (t+1)th iteration until the loss value is lower than the set threshold, at which point the target parameter θ is obtained.

[0111] The trained neural network model is represented as y^=f(x;θ~).

[0112] For example, the state parameters and the time series parameters are measured by sensors on the carbon dioxide capture and storage system, and the target result y is measured using a spectral analysis method.

[0113] In some embodiments of the present invention, the present invention also provides a ship management system for performing the method for determining the degree of degradation of decarbonizing absorbent as described above.

[0114] For example, the neural network model constructed by the method for determining the degree of degradation of the decarbonizing absorbent described in this invention can be embedded in the central control system of the ship management system and can realize automatic forecasting and early warning with a high degree of automation.

[0115] It should be noted that the above-mentioned decarbonization absorbent degradation degree measuring device provided in the embodiments of the present invention can realize all the method steps implemented in the above method embodiments and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiments and the beneficial effects will not be described in detail.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the degree of degradation of a carbon removal absorbent, applied to a carbon dioxide capture and storage system, characterized in that, The method includes: The system parameter samples and target results from the carbon dioxide capture and storage system are input into the constructed neural network model, and the neural network model is trained to obtain the target parameters of the neural network model. Based on the target parameters, the system parameters monitored by the carbon dioxide capture and storage system are input into the trained neural network model to obtain a predicted value of the degradation degree of the carbon removal absorbent. The system parameter sample includes: The parameters reflecting the state of the carbon dioxide capture and storage system include the height and diameter of the absorption tower and regeneration tower, the structural parameters of the equipment inside the towers, the operating parameters of the pressure and temperature inside the absorption tower and regeneration tower, and the physical properties and intrinsic parameters of the organic amines, including their type, initial concentration, and flow rate; and, The time series parameters reflecting the degradation process of the decarbonization absorbent include one or more of the following: the type and concentration of the overhead gas in the absorption tower and the regeneration tower, and the type and concentration of the degradation products at the bottom of the absorption tower and the regeneration tower. The neural network model is a hybrid network structure, including a feedforward network for processing the state parameters, a recurrent network for processing the time series parameters, and a splicing module for fusing the outputs of the feedforward network and the recurrent network.

2. The method for determining the degree of degradation of the decarbonization absorbent according to claim 1, characterized in that, The step of inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model includes: System parameter samples The target result y, representing the degree of degradation, is input into the neural network model; in, This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers. These represent the operating parameters, including pressure, temperature, and reboiler power, within the absorption tower and regeneration tower. The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines. This represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers. This represents the cumulative time series data showing the types and concentration changes of degradation products at the bottom of the absorption and regeneration towers.

3. The method for determining the degree of degradation of the decarbonization absorbent according to claim 2, characterized in that, The step of inputting system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model and training the neural network model to obtain the target parameters of the neural network model further includes: The parameters representing the state. , as well as The input is fed into the feedforward network, and after undergoing a nonlinear transformation by the feedforward network, a first result is output. y 1; The cumulative parameters representing time series parameters and The input is fed into the recurrent network, and after undergoing a nonlinear transformation by the recurrent network, a second result is output. y 2( t ); The first result y 1 and the second result y 2( t The input is given to the stitching module to obtain the prediction result. θ represents all the training parameters of the feedforward network and the recurrent network. Represents a function; Based on the prediction results The training parameters θ are iteratively optimized using a regression loss function, along with the target result y, to obtain the target parameters. .

4. The method for determining the degree of degradation of the decarbonization absorbent according to claim 3, characterized in that, The training parameters θ are iteratively optimized using a regression loss function to obtain the target parameters. The steps include: Calculate the loss value of the t-th iteration. , Represents the loss function; Calculate the gradient of the loss function with respect to the training parameters θ. ; The training parameters θ are updated using gradient descent with a preset learning rate α. In order to obtain Then, the process proceeds to the (t+1)th iteration until the loss value is lower than the set threshold, at which point the target parameter is obtained. ; The trained neural network model is represented as follows: .

5. The method for determining the degree of degradation of the decarbonization absorbent according to claim 1, characterized in that, The state parameters and the time series parameters are obtained by sensors on the carbon dioxide capture and storage system, and the target result is obtained by spectral analysis.

6. A device for determining the degree of degradation of a decarbonization absorbent, characterized in that, The device includes: The training module is used to input system parameter samples and target results from the carbon dioxide capture and storage system into the constructed neural network model and train the neural network model to obtain the target parameters of the neural network model. The prediction module is used to input the system parameters monitored by the carbon dioxide capture and storage system into the trained neural network model based on the target parameters, so as to obtain the predicted value of the degradation degree of the carbon removal absorbent; The system parameter sample includes: The parameters reflecting the state of the carbon dioxide capture and storage system include the height and diameter of the absorption tower and regeneration tower, the structural parameters of the equipment inside the towers, the operating parameters of the pressure and temperature inside the absorption tower and regeneration tower, and the physical properties and intrinsic parameters of the organic amines, including their type, initial concentration, and flow rate; and, The time series parameters reflecting the degradation process of the decarbonization absorbent include one or more of the following: the type and concentration of the overhead gas in the absorption tower and the regeneration tower, and the type and concentration of the degradation products at the bottom of the absorption tower and the regeneration tower. The neural network model is a hybrid network structure, including a feedforward network for processing the state parameters, a recurrent network for processing the time series parameters, and a splicing module for fusing the outputs of the feedforward network and the recurrent network.

7. The apparatus for determining the degree of degradation of decarbonization absorbent according to claim 6, characterized in that, The system parameter samples include state parameters reflecting the carbon dioxide capture and storage system and time series parameters reflecting the degradation process of the carbon removal absorbent. The training module is also used for: System parameter samples The target result y, representing the degree of degradation, is input into the neural network model; in, This indicates the height and diameter of the absorption tower and regeneration tower, as well as the structural parameters of the equipment inside the towers. These represent the operating parameters, including pressure, temperature, and reboiler power, within the absorption tower and regeneration tower. The physical properties and intrinsic parameters representing the type, initial concentration, and flow rate of organic amines. This represents the cumulative time series data of the types and concentration changes of the overhead gases in the absorption and regeneration towers. This represents the cumulative time series data showing the types and concentration changes of degradation products at the bottom of the absorption and regeneration towers.

8. The apparatus for determining the degree of degradation of the decarbonization absorbent according to claim 7, characterized in that, The training module is also used for: The parameters representing the state. , as well as The input is fed into the feedforward network, and after undergoing a nonlinear transformation by the feedforward network, a first result is output. y 1; The parameters representing time series and The input is fed into the recurrent network, and after undergoing a nonlinear transformation by the recurrent network, a second result is output. y 2( t ); The first result y 1 and the second result y 2( t The input is given to the stitching module to obtain the prediction result. θ represents all the training parameters of the feedforward network and the recurrent network. Represents a function; Based on the prediction results The training parameters θ are iteratively optimized using a regression loss function, along with the target result y, to obtain the target parameters. .

9. A ship management system, characterized in that, The system is used to perform the method for determining the degree of degradation of the decarbonization absorbent as described in any one of claims 1 to 5.

Images