Dispersion predictive control method for post-combustion carbon capture system with absorbent storage

Abstract

The invention discloses a post-combustion carbon capture system decentralized predictive control method with absorbent storage, which adopts a conventional MPC method to track a carbon capture rate set value at an absorption side to adapt to flue gas fluctuation and meet the emission reduction target, and adopts an EMPC method at a desorption side to improve the economy in a dynamic operation process, so that the carbon capture rate set value can meet the emission reduction target. The flexible supporting function of the carbon capture system with absorbent storage is exerted, rapid adjustment of the carbon capture rate is achieved to meet the target of the carbon capture rate, the CO2 yield in the dynamic operation process of the system is maximized, the steam extraction consumption of a reboiler is reduced, the disturbance influence of upstream flue gas on the carbon capture system is effectively inhibited, and the carbon capture efficiency is improved. Therefore, flexible, economical and efficient operation of the post-combustion carbon capture system with absorbent storage is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of low-carbon economic energy systems, and in particular to a decentralized predictive control method for a post-combustion carbon capture system with absorbent storage. Background Art

[0002] Deep greenhouse gas emission reductions rely on a low-carbon transformation of the energy system. On the one hand, it is necessary to reduce the use of fossil fuels and vigorously develop renewable energy; on the other hand, it is necessary to promote the clean and efficient use of fossil energy and adopt carbon capture and storage technology in the remaining fossil fuel system. Carbon capture and storage (CCS) technology should be configured for coal-fired power units so that they can continue to serve while meeting climate goals. The post-combustion carbon capture (PCC) method based on solvent chemical absorption has the advantages of large gas processing capacity, fast reaction rate, suitability for processing low partial pressure and low concentration CO2 gas sources, and relatively economical and mature technology, making it suitable for carbon capture in coal-fired power units.

[0003] To support the high-proportion integration of intermittent renewable energy into the power grid, coal-fired power plants have gradually transformed from traditional primary power sources to flexible, adjustable power sources that participate extensively in peak and frequency regulation. If carbon capture systems cannot adapt to flue gas and available extraction steam flow rates caused by fluctuating loads in coal-fired power plants, and meet fluctuating CO2 production demands, they will not be able to fully realize their carbon emission reduction potential. Consequently, numerous researchers have conducted research on the operational control of carbon capture systems, focusing on adapting to flue gas disturbances, flexibly adjusting carbon capture rates, and maintaining reboiler temperatures. In practical operation, it is crucial not only to ensure the stability of the system's closed-loop control and the flexibility of operating conditions, but also to consider the economic efficiency of the regulation process. Current economic optimization efforts for carbon capture systems focus on steady-state parameter optimization and steady-state economic scheduling, with limited attention paid to optimizing the economic operation of the system's dynamic processes. However, due to load fluctuations in upstream power plants, carbon capture systems frequently experience operating fluctuations. Furthermore, due to the system's high inertia, the transition time during the dynamic process is long. Therefore, there is an urgent need to improve the dynamic economic efficiency of carbon capture system operation, reducing the operating costs of the dynamic regulation process while ensuring system stability and flexibility. Summary of the Invention

[0004] Purpose of the invention: The present invention provides a decentralized predictive control method for a post-combustion carbon capture system with absorbent storage, which can achieve rapid adjustment of the carbon capture rate to meet the carbon capture rate target, maximize the CO2 production during the dynamic operation of the system and reduce the reboiler extraction steam consumption, and effectively suppress the disturbance effect of upstream flue gas on the carbon capture system.

[0005] Technical Solution: The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage described in the present invention comprises the following steps:

[0006] Step 1: conducting an open-loop step experiment on a post-combustion carbon capture system with absorbent storage to obtain a dynamic characteristic experimental curve of the post-combustion carbon capture system with absorbent storage, wherein the carbon capture system is divided into an absorption side and a desorption side;

[0007] Step 2: On the absorption side, a discrete state space model characterizing the dynamic characteristics of the carbon capture system is identified by a subspace identification method based on data corresponding to the dynamic characteristic experimental curve, and the discrete state space model is expanded into an incremental state space prediction model;

[0008] Step 3: On the desorption side, a nonlinear discrete difference algebraic equation based on the mechanism model is established as a prediction model to ensure sufficient economic optimization accuracy;

[0009] Step 4: Optimize the tracking target on the absorption side and construct a tracking objective function. Optimize the economic target on the desorption side and construct an economic objective function. Combine the prediction model to design the absorption-desorption decentralized predictive controller corresponding to the carbon capture system. The absorption-desorption decentralized predictive controller is used to realize the control mode of tracking control on the desorption side and economic control on the desorption side of the system.

[0010] Furthermore, in step 1, the controlled variable on the absorption side includes the lean liquid flow rate, the controlled variable includes the CO2 capture rate, and the measurable disturbance flue gas flow rate is taken into account; the controlled variable on the desorption side includes the rich liquid flow rate and the reboiler extraction steam flow rate, and the controlled variable includes the reboiler temperature and CO2 production.

[0011] Furthermore, in step 2, the discrete state space model is expressed as:

[0012]

[0013] Among them, u abs (t k )=u lean (t k ),u abs (t k ) represents t k The absorption side control quantity at the moment, u lean (t k ) represents t k The lean solution flow rate at the moment y abs (t k )=y CL (t k ), y abs (t k ) represents t k The controlled quantity on the absorption side at the moment, yCL (t k ) represents t k CO2 capture rate at time d abs (t k )=d fg (t k ), d abs (t k ) represents t k The measurable disturbance on the absorbing side at time d fg (t k ) represents t k Smoke flow at the moment; x abs,0 (t k ) represents t k The absorbing side state vector at time t; A0, B0, C0, D0, E0 and F0 all represent system characteristic matrices.

[0014] Furthermore, in step 2, the incremental state space prediction model is expressed as:

[0015]

[0016]

[0017] Where O represents the zero matrix; I ny Represents the identity matrix.

[0018] Furthermore, in step 3, the nonlinear discrete difference algebraic prediction model is expressed as:

[0019]

[0020] Among them, u str (t k )=[u rich (t k ),u reb (t k )],u str (t k ) represents t k The desorption side control quantity at the moment, u rich (t k ) represents t k The rich liquid flow rate at time u reb (t k ) represents t k Reboiler extraction steam flow at time ; y str (t k ) represents t k The controlled quantity on the desorption side at the moment, y reb (t k ) represents t kThe reboiler temperature at time Indicates t k CO2 production at the time; x str (t k ) represents t k The desorption side differential state vector at time z str (t k ) represents t k The algebraic state vector of the desorption side at time .

[0021] Furthermore, in step 4, the tracking objective optimization and tracking objective function of the absorbing side are expressed as:

[0022]

[0023] stx abs (t+1)=Ax abs (t)+B u Δu abs (t)+B d Δd abs (t)

[0024] y abs (t) = Cx abs (t)

[0025] x abs (t k )=x abs,m (t k )

[0026] y abs,min ≤y abs (t)≤y abs,max

[0027] u abs,min ≤u abs (t)≤u abs,max

[0028] Δu abs,min ≤Δu abs (t)≤Δu abs,max

[0029] Among them, y abs,r Indicates the set value of the controlled quantity on the absorption side; N p represents the prediction time domain; Q represents the error weight; R represents the control weight; x abs,m Indicates the absorption side state quantity x abs The measured value of y abs,min and y abs,max Represent the controlled quantity y on the absorption side respectively abs Upper and lower limits of u abs,min and u abs,max They represent the absorption side control quantity uabs Upper and lower limits of Δu abs,min and Δu abs,max They represent the change rate of the control quantity on the absorption side Δu abs upper and lower limits.

[0030] Furthermore, in step 4, the desorption side economic target optimization is divided into two stages: steady-state target value optimization and dynamic transition process optimization. The steady-state target value optimization objective function is expressed as:

[0031]

[0032] stf(x str,s ,z str,s ,u str,s )=0

[0033] g(x str,s ,z str,s ,u str,s )=0

[0034] y str,s =h(x str,s ,z str,s ,u str,s )

[0035] y str,min ≤y str,s ≤y str,max

[0036] u str,min ≤u str,s ≤u str,max

[0037] Where α1 and α2 represent the CO2 transaction price and reboiler extraction steam price respectively; x str,s and z str,s They represent the differential state and algebraic state at the optimal economic steady-state operating point on the desorption side respectively; u str,s and y str,s They represent the controlled quantity and the controlled quantity at the economically optimal steady-state operating point on the desorption side respectively; u str,min and u str,max Represents u str,s The upper and lower limits of y str,min and y str,max Represents y str,s Upper and lower limits of ;

[0038] The optimization objective function of the dynamic transition process is expressed as:

[0039]

[0040] stx str (t+1)=f(xstr (t),z str (t),u str (t))

[0041] g(x str (t),z str (t),u str (t))=0

[0042] y str (t) = h(x str (t),z str (t),u str (t))

[0043] x str (t k )=x str,m (t k )

[0044] z str (t k )=z str,m (t k )

[0045] y str,min ≤y str (t)≤y str,max

[0046] u str,min ≤u str (t)≤u str,max

[0047] Among them, β1, β2 and β3 represent regularization coefficients respectively; x str,m Represents the differential state quantity x on the desorption side str The measured value of z str,m represents the algebraic state quantity z on the desorption side str The measured value of V f (·) is expressed as the terminal cost function, which ensures that the predicted state of the system at the end of the prediction time domain is within the terminal domain, thereby ensuring the feasibility and stability of the system under EMPC control.

[0048] Furthermore, in step 4, the absorption-desorption distributed predictive controller corresponding to the carbon capture system includes an absorption side MPC controller and a desorption side EMPC controller; the absorption side MPC controller controls the CO2 capture rate on the absorption side, and the desorption side EMPC controller controls the reboiler temperature and CO2 production on the desorption side.

[0049] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: Through the control structure, the present invention adopts the conventional MPC method on the absorption side to track the carbon capture rate set value to adapt to flue gas fluctuations and meet the emission reduction target, and adopts the EMPC method on the desorption side to improve the economy during dynamic operation, give full play to the flexibility support function of the carbon capture system with absorbent storage, realize rapid adjustment of the carbon capture rate to meet the carbon capture rate target, maximize the CO2 production during the dynamic operation of the system and reduce the reboiler extraction steam consumption, and effectively suppress the disturbance effect of upstream flue gas on the carbon capture system, thereby realizing flexible, economical and efficient operation of the post-combustion carbon capture system with absorbent storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of the system structure of the present invention.

[0051] Figure 2 Schematic diagram of the control structure of the present invention.

[0052] Figure 3 Schematic diagram of the system dynamic response curve of the present invention.

[0053] Figure 4 Schematic diagram of the experimental simulation result curve of scenario 1 of the present invention.

[0054] Figure 5 Schematic diagram of the simulation result curve of scenario 2 experiment of the present invention. DETAILED DESCRIPTION

[0055] like Figure 1 The system shown in Figure 1 is a post-combustion carbon capture system with absorbent storage. The main components of a post-combustion carbon capture system with absorbent storage include an absorber, desorber, reboiler, lean liquid tank, rich liquid tank, and condenser. Other auxiliary equipment includes heat exchangers, lean liquid pumps, rich liquid pumps, piping, and valves. Flue gas enters the absorber from the bottom up, where it comes into countercurrent contact with the lean absorbent solution (low CO2 loading absorbent) entering from the top. After CO2 removal, the absorbent leaves the top of the absorber and is discharged into the atmosphere. Rich absorbent solution (high CO2 loading absorbent), containing a large amount of CO2, leaves the bottom of the absorber and is stored in the rich liquid tank. An interstage cooling process is implemented in the middle section of the absorber to reduce the temperature rise caused by the exothermic heat of CO2 absorption and enhance the reaction driving force. After the rich liquid and the regenerated lean liquid exchange heat in the heat exchanger, they enter the desorber to desorb the CO2. The desorption heat is provided by steam extraction from the generator turbine. The desorbed CO2 leaves the top of the desorber and enters the condenser to complete the subsequent process. The regenerated lean liquid flows into the lean liquid tank for storage and is sprayed into the absorption tower again through the lean liquid pump to start the next cycle.

[0056] In order to improve the economic efficiency, carbon emission reduction benefits and operational flexibility of a post-combustion carbon capture system with absorbent storage, the present invention proposes a decentralized predictive control method for a post-combustion carbon capture system with absorbent storage, which is implemented through a control structure. The control structure implements control through an absorption side MPC controller and a desorption side MPC controller, such as Figure 2 As shown in the figure, the absorption-side MPC controller controls the lean liquid pump of the carbon capture system, while the desorption-side EMPC controller controls the rich liquid pump of the carbon capture system. The absorption-side MPC controller uses the lean liquid flow rate to control the CO2 capture rate, with the flue gas flow rate as a measurable disturbance. The desorption-side EMPC controller uses the rich liquid flow rate and reboiler extraction steam flow rate to control the reboiler temperature and CO2 production.

[0057] The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage includes:

[0058] S1: Conduct an open-loop step experiment on a post-combustion carbon capture system with absorbent storage to obtain a dynamic characteristic experimental curve of the post-combustion carbon capture system with absorbent storage; wherein the carbon capture system is divided into an absorption side and a desorption side; the absorption side control variable includes the lean liquid flow rate, the controlled variable includes the CO2 capture rate, and a measurable disturbance flue gas flow rate is considered; the desorption side control variable includes the rich liquid flow rate and the reboiler extraction steam flow rate, and the controlled variable includes the reboiler temperature and CO2 production;

[0059] S2: The absorption side identifies and obtains a discrete state space prediction model characterizing the dynamic characteristics of the carbon capture system using a subspace identification method based on the data corresponding to the dynamic characteristic experimental curve, and expands the discrete state space model into an incremental state space prediction model;

[0060] Specifically, the discrete state space model is expressed as:

[0061]

[0062] Among them, u abs (t k )=u lean (t k ),u abs (t k ) represents t k The absorption side control quantity at the moment, u lean (t k ) represents t k The lean solution flow rate at the moment y abs (t k )=y CL (t k ), y abs (t k ) represents t kThe controlled quantity on the absorption side at the moment, y CL (t k ) represents t k CO2 capture rate at time d abs (t k )=d fg (t k ), d abs (t k ) represents t k The measurable disturbance on the absorbing side at time d fg (t k ) represents t k Smoke flow at the moment; x abs,0 (t k ) represents t k The absorbing side state vector at time t; A0, B0, C0, D0, E0 and F0 all represent system characteristic matrices;

[0063] The incremental state space model is expressed as:

[0064]

[0065] Where O represents the zero matrix; I ny Represents the identity matrix.

[0066] S3: On the desorption side, a nonlinear discrete difference algebraic equation based on the mechanism model is established as a prediction model to ensure sufficient economic optimization accuracy;

[0067] Specifically, the nonlinear discrete difference algebraic prediction model is expressed as:

[0068]

[0069] S4: Optimize the tracking target on the absorption side and construct a tracking objective function, optimize the economic target on the desorption side and construct an economic objective function, and design the absorption-desorption decentralized prediction controller corresponding to the carbon capture system in combination with the prediction model. The control mode of the system desorption side tracking control and the desorption side economic control is realized through the absorption-desorption decentralized prediction controller, thereby realizing flexible, economical and efficient operation of the post-combustion carbon capture system with absorbent storage.

[0070] Specifically, the absorbing side tracking target optimization and tracking objective function are expressed as:

[0071]

[0072] stx abs (t+1)=Ax abs (t)+B u Δu abs (t)+Bd Δd abs (t)

[0073] y abs (t) = Cx abs (t)

[0074] x abs (t k )=x abs,m (t k )

[0075] y abs,min ≤y abs (t)≤y abs,max

[0076] u abs,min ≤u abs (t)≤u abs,max

[0077] Δu abs,min ≤Δu abs (t)≤Δu abs,max (5)

[0078] Among them, y abs,r Indicates the set value of the controlled quantity on the absorption side; N p represents the prediction time domain; Q represents the error weight; R represents the control weight; x abs,m Indicates the absorption side state quantity x abs The measured value of y abs,min and y abs,max Represent the controlled quantity y on the absorption side respectively abs Upper and lower limits of u abs,min and u abs,max They represent the absorption side control quantity u abs Upper and lower limits of Δu abs,min and Δu abs,max They represent the change rate of the control quantity on the absorption side Δu abs upper and lower limits.

[0079] The desorption side economic target optimization is divided into two stages: steady-state target value optimization and dynamic transition process optimization. The steady-state target value optimization objective function is expressed as:

[0080]

[0081] stf(x str,s ,z str,s ,u str,s )=0

[0082] g(x str,s ,z str,s ,u str,s )=0

[0083] y str,s =h(x str,s ,z str,s ,u str,s )

[0084] y str,min ≤y str,s ≤y str,max

[0085] u str,min ≤u str,s ≤u str,max (6)

[0086] Where α1 and α2 represent the CO2 transaction price and reboiler extraction steam price respectively; x str,s and z str,s They represent the differential state and algebraic state at the optimal economic steady-state operating point on the desorption side respectively; u str,s and y str,s They represent the controlled quantity and the controlled quantity at the economically optimal steady-state operating point on the desorption side respectively; u str,min and u str,max Represents u str,s The upper and lower limits of y str,min and y str,max Represents y str,s upper and lower limits.

[0087] The dynamic transition process optimization objective function is expressed as:

[0088]

[0089] stx str (t+1)=f(x str (t),z str (t),u str (t))

[0090] g(x str (t),z str (t),u str (t))=0

[0091] y str (t) = h(x str (t),z str (t),u str (t))

[0092] x str (t k )=x str,m (t k )

[0093] y str,min ≤ystr (t)≤y str,max

[0094] u str,min ≤u str (t)≤u str,max (7)

[0095] Among them, β1, β2 and β3 represent regularization coefficients respectively; x str,m Represents the differential state quantity x on the desorption side str The measured value of V f (·) is expressed as the terminal cost function, which ensures that the predicted state of the system at the end of the prediction time domain is within the terminal domain, thereby ensuring the feasibility and stability of the system under EMPC control.

[0096] The control structure of the decentralized predictive control algorithm is as follows Figure 2 As shown in the figure, specifically, the MPC controller on the absorption side uses the lean liquid flow rate to control the CO2 capture rate, and the flue gas flow rate is used as a measurable disturbance; the EMPC controller on the desorption side uses the rich liquid flow rate and the reboiler extraction steam flow rate to control the reboiler temperature and CO2 production.

[0097] The following calculation is based on the actual system as an example. The specific process is as follows:

[0098] First, based on the established mechanism model of the post-combustion carbon capture system with absorbent storage, an open-loop step experiment is performed to obtain the corresponding dynamic characteristic experimental curve, which is then compared with the dynamic characteristic experimental curve of the post-combustion carbon capture system without absorbent storage, such as Figure 3 The introduction of the absorbent storage tank eliminates the influence of the lean liquid flow rate on the desorption process, weakens and delays the influence of the reboiler extraction steam flow rate on the absorption process, and basically realizes the decoupling between the absorption and desorption processes of the system.

[0099] Subsequently, using a subspace identification method based on the corresponding dynamic characteristics experimental data with a sampling period of 20 seconds, an incremental state-space model of the absorption side was obtained after transformation. A nonlinear discrete difference algebraic equation based on the mechanism model was established as the prediction model of the desorption side with a sampling period of 20 seconds.

[0100] Subsequently, according to equations (5) to (7), the optimization problem of MPC control on the absorption side and EMPC control on the desorption side of the carbon capture system can be constructed, and the optimization solution can be performed to control the post-combustion carbon capture system with absorbent storage.

[0101] The basic parameters and constraint settings of the controller of the decentralized predictive control method are shown in Table 1. In order to verify the superiority of the decentralized predictive control algorithm proposed in this embodiment, it is compared with the tracking MPC control algorithm adopted on both the absorption and desorption sides.

[0102] Table 1 Decentralized predictive controller parameters and constraint settings

[0103]

[0104] The simulation results are as follows Figure 4 and Figure 5 As shown. Among them, Figure 4 The following is a comparison chart of the control performance of the two control methods. Figure 5 This is a comparison chart of the control performance of the two control methods under scenario 2.

[0105] In the above scenario, the absorption side needs to meet a higher CO2 capture rate requirement, and the desorption side needs to meet a given CO2 production requirement. Assume that the initial operating conditions of the carbon capture system are as shown in the left column of Table 2, the target optimized operating conditions are as shown in the right column of Table 2, and the flue gas flow rate remains unchanged during the simulation. Figure 4 As shown, at 200 seconds, the CO2 capture rate setpoint and CO2 production setpoint become 95% and 30 kg / s, respectively. On the absorption side, both the decentralized predictive controller and the comparative controller are able to quickly and steadily control the CO2 capture rate to a new steady-state operating point by adjusting the lean liquid flow rate. On the desorption side, the comparative controller's control objective is to quickly track the given CO2 production and the new reboiler temperature. Therefore, both the rich liquid flow rate and the reboiler extraction steam flow rate increase rapidly to quickly meet the higher CO2 production demand. The decentralized predictive controller's control objective is to minimize reboiler heat consumption during the dynamic process of meeting the higher CO2 production target, thereby reducing operating costs. Therefore, the reboiler extraction steam flow rate increases slightly more slowly than the comparative controller. At the same time, to offset the impact of the increased reboiler extraction steam flow rate and quickly reach the CO2 production target, the rich liquid flow rate is significantly increased to achieve higher dynamic operating economic benefits. Economic comparison results show that during the primary dynamic transition period (200 seconds to 800 seconds), the decentralized predictive controller achieves a 1.23% improvement in economics over the comparative controller.

[0106] Table 2 System operating conditions for scenario 1

[0107] variable Initial operating conditions Target optimized operating conditions u [0.19,0.19,59.33] [0.24,0.46,114.79] y [83.91,391.51,14.78] [95.00,390.95,30.00] d 81.97 81.97

[0108] In the second scenario, the absorption side is affected by the upstream flue gas disturbance and needs to demonstrate its anti-disturbance performance. The desorption side needs to adapt to the load changes of the upstream power plant, so the reboiler extraction steam flow is fixed. Assume that the initial operating conditions of the carbon capture system are as shown in the left column of Table 3, and the target optimized operating conditions are as shown in the right column of Table 3. Figure 5As shown in the figure, at the 200th second, the flue gas flow rate increased by 3%. For the absorption side, when the flue gas flow rate increases, the CO2 capture rate will decrease, but both the controller and the comparison controller adjust the lean liquid flow rate so that the CO2 capture rate eventually returns to the initial value and tends to be stable. For the desorption side, since the value of the reboiler steam extraction flow rate is fixed, the control process depends on the change of the rich liquid flow rate. The comparison controller gradually increases the rich liquid flow rate to achieve the final steady-state optimization result. Under the action of the decentralized predictive controller, since the reboiler heat consumption cost is fixed, the rich liquid flow rate initially increases rapidly and then gradually decreases to a stable value to increase the CO2 production during dynamic operation and improve economic benefits. The economic comparison results show that in the main dynamic transition process (200s-800s), the predictive controller is 1.28% more economical than the comparison controller.

[0109] Table 3 System operating conditions under scenario 2

[0110] variable Initial operating conditions Target optimized operating conditions u [0.19,0.19,59.33] [0.20,0.22,59.33] y [83.91,391.51,14.78] [83.91,391.10,15.44] d 81.97 84.43

[0111] In summary, the decentralized predictive control method for a post-combustion carbon capture system with absorbent storage proposed in the present invention can achieve a more low-carbon, economical, flexible and safe operation of the carbon capture system.

Claims

1. A decentralized predictive control method for a post-combustion carbon capture system with absorbent storage, characterized in that: The steps include: Step 1: conducting an open-loop step experiment on a post-combustion carbon capture system with absorbent storage to obtain a dynamic characteristic experimental curve of the post-combustion carbon capture system with absorbent storage, wherein the carbon capture system is divided into an absorption side and a desorption side; Step 2: On the absorption side, a discrete state space model characterizing the dynamic characteristics of the carbon capture system is identified by a subspace identification method based on data corresponding to the dynamic characteristic experimental curve, and the discrete state space model is expanded into an incremental state space prediction model; Step 3: On the desorption side, a nonlinear discrete difference algebraic equation based on the mechanism model is established as a prediction model to ensure sufficient economic optimization accuracy; Step 4: Optimize the tracking target on the absorption side and construct a tracking objective function. Optimize the economic target on the desorption side and construct an economic objective function. Combine the prediction model to design the absorption-desorption decentralized predictive controller corresponding to the carbon capture system. The absorption-desorption decentralized predictive controller is used to realize the control mode of tracking control on the desorption side and economic control on the desorption side of the system.

2. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, characterized in that: In step 1, the controlled quantity on the absorption side includes the lean liquid flow rate, the controlled quantity includes the CO2 capture rate, and the measurable disturbance flue gas flow rate is taken into account; The controlled variables on the desorption side include the rich liquid flow rate and the reboiler extraction steam flow rate, and the controlled variables include the reboiler temperature and CO2 production.

3. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 2, the discrete state space model is expressed as: Among them, u abs (t k )=u lean (t k ),u abs (t k ) represents t k The absorption side control quantity at the moment, u lean (t k ) represents t k The lean solution flow rate at the moment y abs (t k )=y CL (t k ), y abs (t k ) represents t k The controlled quantity on the absorption side at the moment, y CL (t k ) represents t k CO2 capture rate at time d abs (t k )=d fg (t k ), d abs (t k ) represents t k The measurable disturbance on the absorbing side at time d fg (t k ) represents t k Smoke flow at the moment; x abs,0 (t k ) represents t k The absorbing side state vector at time t; A0, B0, C0, D0, E0 and F0 all represent system characteristic matrices.

4. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 2, the incremental state space prediction model is expressed as: Where O represents the zero matrix; I ny Represents the identity matrix.

5. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 3, the nonlinear discrete difference algebraic prediction model is expressed as: Among them, u str (t k )=[u rich (t k ),u reb (t k )],u str (t k ) represents t k The desorption side control quantity at the moment, u rich (t k ) represents t k The rich liquid flow rate at time u reb (t k ) represents t k Reboiler extraction steam flow at time ; y str (t k ) represents t k The controlled quantity on the desorption side at the moment, y reb (t k ) represents t k The reboiler temperature at time Indicates t k CO2 production at the time; x str (t k ) represents t k The desorption side differential state vector at time z str (t k ) represents t k The algebraic state vector of the desorption side at time .

6. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 4, the tracking objective optimization and tracking objective function of the absorbing side are expressed as: Among them, y abs,r Indicates the set value of the controlled quantity on the absorption side; N p represents the prediction time domain; Q represents the error weight; R represents the control weight; x abs,m Indicates the absorption side state quantity x abs The measured value of y abs,min and y abs,max Represent the controlled quantity y on the absorption side respectively abs Upper and lower limits of u abs,min and u abs,max They represent the absorption side control quantity u abs Upper and lower limits of Δu abs,min and Δu abs,max They represent the change rate of the control quantity on the absorption side Δu abs upper and lower limits.

7. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 4, the desorption side economic target optimization is divided into two stages: steady-state target value optimization and dynamic transition process optimization. The steady-state target value optimization objective function is expressed as: Where α1 and α2 represent the CO2 transaction price and reboiler extraction steam price respectively; x str,s and z str,s They represent the differential state and algebraic state at the optimal economic steady-state operating point on the desorption side respectively; u str,s and y str,s They represent the controlled quantity and the controlled quantity at the economically optimal steady-state operating point on the desorption side respectively; u str,min and u str,max Represents u str,s The upper and lower limits of y str,min and y str,max Represents y str,s Upper and lower limits of ; The optimization objective function of the dynamic transition process is expressed as: Among them, β1, β2 and β3 represent regularization coefficients respectively; x str,m Represents the differential state quantity x on the desorption side str The measured value of z str,m represents the algebraic state quantity z on the desorption side str The measured value of V f (·) is expressed as the terminal cost function, which ensures that the predicted state of the system at the end of the prediction time domain is within the terminal domain, thereby ensuring the feasibility and stability of the system under EMPC control.

8. The decentralized predictive control method for a post-combustion carbon capture system with absorbent storage according to claim 1, wherein: In step 4, the absorption-desorption distributed predictive controller corresponding to the carbon capture system includes an absorption side MPC controller and a desorption side EMPC controller; the absorption side MPC controller controls the CO2 capture rate on the absorption side, and the desorption side EMPC controller controls the desorption side reboiler temperature and CO2 production.

Images