A dry / wet state conversion automatic control method and system for improving deep peak regulation capacity of a thermal power unit

Abstract

The application discloses a dry / wet state conversion automatic control method and system for improving deep peak regulation capacity of a thermal power unit. Firstly, a supercritical combined heat and power unit is taken as a research object and difficulties of dry / wet state conversion of the supercritical combined heat and power unit under deep peak regulation demand are analyzed; then, a dry / wet state conversion automatic control strategy and system are designed for the supercritical combined heat and power unit in combination with a multi-objective model predictive control algorithm. Finally, the feasibility of the dry / wet state conversion automatic control method is verified by relying on a simulation platform. In the analysis of the research object, the supercritical combined heat and power unit in the dry state and the wet state are both simplified into different four-input four-output systems, so that the dynamic characteristics of the supercritical combined heat and power unit are more accurately represented. In addition, the multi-objective model predictive control algorithm is constructed by comprehensively considering the power generation cost, carbon transaction cost, smoothness of the manipulated variable and load tracking error of the unit, so that the unit realizes the improvement of the deep peak regulation capacity on the basis of low carbon, economy and stable operation.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for dry / wet state switching in deep peak shaving of thermal power generating units, and more specifically, to an automatic control method and system for dry / wet state switching to improve the deep peak shaving capability of supercritical cogeneration units. Background Technology

[0002] The penetration of a high proportion of renewable energy has effectively accelerated the transformation of the low-carbon, clean power grid. However, the resulting indirectness has also exacerbated grid fluctuations. Supercritical cogeneration units, capable of simultaneously supplying electricity and heat to users, offer greater operational flexibility than conventional condensing units. Furthermore, the high thermal efficiency resulting from the high parameters of the working fluid in the once-through boiler has made cogeneration units the mainstay of power and heat supply in northern my country. The dry / wet switching effect of the once-through boiler in a supercritical cogeneration unit has a significant impact on the unit's deep peak-shaving capability. Therefore, to fully leverage the peak-shaving and valley-filling functions of cogeneration units and accommodate larger-scale renewable energy grid connections, it is urgently necessary to design an automatic control algorithm and system that promotes smooth dry / wet switching in the once-through boiler.

[0003] Supercritical combined heat and power (CHP) units are a type of thermal power generating unit capable of simultaneously producing electricity and heat. Most existing CHP units in northern my country convert their chemical energy into electricity and heat through coal combustion. The thermal efficiency of pure condensing generator units is generally 25%–30%, while the total thermal efficiency of CHP units is above 45%. Furthermore, CHP units can rapidly alter electricity output by adjusting the heat supplied through steam extraction, thus mitigating fluctuations caused by renewable energy sources. Therefore, deeply exploring the peak-shaving capabilities of CHP units and leveraging their peak-shaving and valley-filling effects can promote the transformation to a clean power grid. As a core component of energy conversion in supercritical CHP units, the steam-water flow of the working fluid inside the once-through boiler is relatively complex. The control effect of this flow directly impacts the thermal efficiency of the once-through boiler. When the unit operates at 30% or higher of its rated load, the steam inside the once-through boiler is typically dry. However, when the unit operates below 30% of its rated load, the steam inside the once-through boiler is typically wet. It is evident that when supercritical cogeneration units participate in deep peak shaving, the dry / wet state conversion of steam within the once-through boiler is unavoidable. If the steam in the once-through boiler can achieve a smooth and rapid dry / wet state conversion, the supercritical cogeneration unit can effectively fulfill its deep peak shaving function. However, the dynamic characteristics of the working fluid in the once-through boiler of a supercritical cogeneration unit are completely different when operating in dry and wet modes, and both exhibit multivariable, nonlinear, and strongly coupled characteristics. Currently, the dry / wet state conversion of supercritical cogeneration units still mainly relies on manual adjustment, which is largely limited by the experience of operators, and the conversion speed and effectiveness are difficult to meet the flexibility requirements. Therefore, this paper takes the supercritical cogeneration unit as the controlled object and designs an advanced automatic control method and system for dry / wet state conversion based on an improved multi-objective model predictive control algorithm, providing an effective reference for improving the deep peak shaving capability of large thermal power generating units.

[0004] Model predictive control (MDC) is considered one of the most advanced control methods due to its superior performance in handling complex industrial processes with large delays, multiple variables, multiple constraints, and instability. The predictive model, rolling optimization, and feedback correction constitute the three stages of MDC. The online optimization characteristic of the rolling optimization stage gives MDC excellent setpoint tracking performance, disturbance suppression performance, and robustness. Furthermore, MDC has been widely and successfully applied in the energy, power, and aerospace fields due to its ability to effectively handle multivariable and multi-constraint problems. Therefore, by incorporating the low-carbon, economical, and stable operation requirements of supercritical cogeneration units under a flexible background into the overall optimization objective, a multi-objective MDC method is constructed, laying a solid foundation for designing an automatic control system for dry / wet switching of once-through boilers to improve the deep peak-shaving capability of the unit. Summary of the Invention

[0005] This invention aims to provide an automatic control method for dry / wet state switching to improve the peak-shaving capacity of thermal power generating units, thereby fully leveraging their role in peak shaving and valley filling in the new power system. This method fully considers the different dynamic characteristics of thermal power generating units operating in dry and wet modes, and combines an improved multi-objective model predictive control algorithm, adept at rolling optimization and handling multiple constraints, to design different dry and wet controllers to form an automatic control system for dry / wet state switching. During the optimal control law solution process of the multi-objective model predictive control algorithm, the power generation cost, carbon emission cost, and smoothness of manipulated variables of the thermal power generating unit are reasonably balanced. Based on the designed automatic control system for dry / wet state switching, the thermal power generating unit can quickly and smoothly complete the automatic dry / wet state switching of the working fluid in the once-through boiler when participating in deep peak shaving, maintaining stable, low-carbon, and economical operation, thus achieving the goal of improving deep peak shaving capacity.

[0006] The dry / wet state switching automatic control method and system for improving the deep peak shaving capability of thermal power units proposed in this invention consists of the following three steps:

[0007] S1: Analysis of the control challenges of supercritical cogeneration units operating in dry and wet modes;

[0008] S2: Establish an automatic control method and system for dry / wet state switching for deep peak shaving of supercritical cogeneration units;

[0009] S3: Verify and analyze the effectiveness of the designed automatic control method for dry / wet state transition based on the simulation platform.

[0010] S1: The purpose of thermal power generating units participating in deep peak shaving is to provide a margin for the grid connection of large-scale renewable energy power. To improve the deep peak shaving capability and power regulation quality of thermal power generating units, advanced control theory can be used to design an automatic dry / wet switching control system for once-through boilers to meet multi-objective regulation requirements. However, thermal power generating units operating at low load levels for peak shaving have low thermal efficiency, resulting in significant losses. Therefore, the requirements for dry / wet switching of once-through boilers are speed and smoothness, reducing their operating time in dry / wet mixed mode and preventing repeated switching from damaging the equipment. The task of the automatic dry / wet switching control system for once-through boilers is to automatically and quickly complete the dry / wet steam switching while taking into account the environmentally friendly and economical operation of thermal power generating units, without relying on the experience of operators. If the control effect of the automatic dry / wet switching control system meets the requirements of speed, accuracy, and stability, then the thermal power generating unit can play its important role in peak shaving and valley filling. Accurate model construction and appropriate control algorithms are the foundation for designing an automatic dry / wet switching control system. Thermal power generating units consist of two main components: a boiler and a steam turbine, supplemented by a reheat system. A boiler is a large and complex heat supply system, mainly composed of a furnace, water-cooled walls, a steam-water separator, a superheater, a spray desuperheater, a reheater, an economizer, and an air preheater. The operating mode of the unit is related to the state of the working fluid at the steam-water separator outlet. The factors affecting the stable operation of a thermal power generating unit differ significantly between dry and wet operating modes. In dry mode, the working fluid at the separator outlet is steam, and the water level in the storage tank is 0. The enthalpy of the working fluid at the separator outlet, i.e., the midpoint enthalpy, can accurately characterize the state of the working fluid. After being heated by the superheater, the steam flows through the desuperheater and finally enters the steam turbine to perform work and generate electricity. The steam flow rate entering the steam turbine is controlled by the opening of the main steam valve and is positively correlated with the unit's output power. Considering the most critical factors of the steam-water flow and unit operating state in a once-through boiler, a thermal power generating unit in dry operating mode can be simplified as a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W The four output variables are the output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T MIn wet operation mode, the working medium at the separator outlet in the boiler is steam, but the water level in the separator-storage tank is typically 2-7 meters. The water level in the separator-storage tank characterizes the degree of operation in wet mode. If the water level gradually decreases to 0, it indicates a successful switch from wet to dry operation. Therefore, a thermal power generating unit operating in wet mode can be simplified as a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T The four output variables are the output power N. E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W Regardless of whether it is dry or wet operation mode, each output variable and input variable is accompanied by complex dynamic characteristics of strong coupling, time variation, large delay and nonlinearity. In addition, the need for rapid peak shaving and valley filling makes it difficult for the unit to achieve a fast and smooth dry / wet state transition.

[0011] Based on the analysis in S1, it is evident that achieving rapid and smooth dry / wet state switching performance in supercritical cogeneration units is challenging due to the involvement of multiple variables in the control. Therefore, an advanced automatic control method and system for dry / wet state switching needs to be designed, incorporating an improved multi-objective model predictive control algorithm. Considering the requirements for low-carbon, economical, and stable operation under peak-shaving conditions, a multi-objective optimization function is constructed, taking into account the unit's carbon emission costs, power generation costs, and the smoothness of manipulated variables. Subsequently, an automatic control system for dry / wet state switching is built using a model predictive control algorithm with rolling optimization characteristics. By rolling the solution of the optimal control law under multiple constraints, the unit achieves rapid, accurate, and automatic dry / wet state switching while maintaining safe and economical operation. Since the model predictive control algorithm can handle multi-variable and multi-constraint problems, only one multi-objective model predictive controller needs to be designed for each of the unit operating in dry and wet modes. The controller for the unit in dry mode is denoted as the "dry-state multi-objective model predictive controller," and the controller for the unit in wet mode is denoted as the "wet-state multi-objective model predictive controller." Finally, by connecting the dry-state multi-objective model predictive controller and the wet-state multi-objective model predictive controller through the load command judgment logic, the design of the unit's dry / wet state switching automatic control system can be completed. Step S2 can then be specified as follows:

[0012] S2.1: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is greater than 30% of the rated load, the unit must operate in dry mode. Execute the dry-state multi-objective model predictive controller to achieve the goal of quickly and accurately meeting the external load demand while maintaining low-carbon and economical operation.

[0013] S2.2: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is less than 25% of the rated load, execute the wet-state multi-objective model predictive controller to achieve the purpose of quickly shedding load and ensuring the safe and economical operation of the unit.

[0014] S2.3: The external load demand signal is sent to the load command judgment logic. If the load demand on the unit decreases from above 30% of the rated load to below 25% of the rated load, the unit is required to participate in deep peak shaving. The unit needs to switch from dry-state operation mode to wet-state operation mode, which involves the dry / wet conversion of the once-through boiler. The control system needs to automatically switch from a dry-state multi-objective model predictive controller to a wet-state multi-objective model predictive controller. During the process of the unit's output load decreasing from 30% of the rated load to 25% of the rated load, the separator-storage tank water level shows a very obvious and rapid downward trend. When the unit's output load reaches below 25% of the rated load, the separator-storage tank water level drops to 0 meters, quickly and successfully switching to wet-state operation mode.

[0015] After determining the framework of the automatic control system for dry / wet switching of thermal power generating units based on multi-objective model predictive control algorithms, the design steps of the dry-state multi-objective model predictive controller and the wet-state multi-objective model predictive controller are specified in S2.4:

[0016] S2.4.1: Design of a predictive controller for a dry-state multi-objective model.

[0017] The three main components of a model predictive control algorithm are the predictive model, rolling optimization, and feedback correction. For an incremental model predictive control algorithm that reconstructs the state, its state prediction model is:

[0018] (1)

[0019] The state vector in the prediction time domain P has the following form:

[0020] (2)

[0021] (3)

[0022] (4)

[0023] The predicted output model is:

[0024] (5)

[0025] in,

[0026] (6)

[0027] The controlled variable is the unit's output variable: output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T M That is, the controlled variable matrix is A, B, and C are all coefficient matrices of the unit's state-space model; , and These are the augmented matrices of A, B, and C, respectively;

[0028] (7)

[0029] (8)

[0030] The control law obtained by the rolling optimization solution is in the following form:

[0031] (10)

[0032] Among them, the control quantity is the input variable of the unit: fuel quantity u B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W That is, the control matrix is The incremental optimal control law is:

[0033] (10)

[0034] (11)

[0035] (12)

[0036] To achieve deep peak-shaving capability of thermal power generating units while ensuring their low-carbon, economical, and stable operation, it is necessary to construct a system that encompasses the unit's output... Carbon emission costs Electricity generation cost C e and state stability The multi-objective optimization function J is used; the current optimal control law increment ΔU of the unit is obtained by rolling minimization of the multi-objective function J. Therefore, the optimization objectives are as follows:

[0037] (13)

[0038] The sum of the weighting factors c1, c2 and c3 is 1.

[0039] J1 is based on unit output The optimization is expressed as:

[0040] (14)

[0041] Where w(k) is the setpoint matrix of the unit output.

[0042] J2 is based on the unit's carbon emission cost. Electricity generation cost C e The optimization is expressed as:

[0043] (15)

[0044] Carbon emission costs of thermal power generating units The calculation formula is:

[0045] (16)

[0046] Wherein, Ω represents the unit's basic carbon emission allowance, and any excess must be purchased from the carbon trading market; α and β represent the carbon trading price and carbon emissions per unit load, respectively.

[0047] The power generation cost C of thermal power generating units e The calculation formula is:

[0048] (17)

[0049] Where θ is the coal price; and η is the coal consumption rate related to the unit's output load. c The expression is:

[0050] (18)

[0051] Among them, a2 and b2 are constant factors, which can be obtained by fitting the actual operating data of the unit.

[0052] J3 is based on unit state stability. Optimization, its expression is:

[0053] (19)

[0054] Among them, Q y R y Q c R c Q x and R x All are weight matrices.

[0055] The constraints are:

[0056] (20)

[0057] Among them, u max and u min These are the upper and lower bounds of the control law matrix, respectively; Δu max and Δu min These are the upper and lower limits of the control law increment matrix, respectively.

[0058] Next, the optimal control law of the multi-objective model predictive control algorithm is derived by combining J1, J2 and J3 and solved using the quadratic programming method.

[0059] Will Substituting into equation (14) and retaining only the terms relevant to optimization, then

[0060] (twenty one)

[0061] Rewrite equation (21) in the form of a quadratic programming problem. ,but

[0062] (twenty two)

[0063] (twenty three)

[0064] Will Substituting into equation (15) and retaining only the terms relevant to optimization, then

[0065] (twenty four)

[0066] in,

[0067] (25)

[0068] Rewrite equation (24) in the form of a quadratic programming problem. ,but

[0069] (26)

[0070] (27)

[0071] Will Substituting into equation (19) and retaining only the terms relevant to optimization, then

[0072] (28)

[0073] Rewrite equation (28) in the form of a quadratic programming problem. ,but

[0074] (29)

[0075] (30)

[0076] In summary, by combining equations (31) and (32), quadratic programming is used to solve the optimal control law based on the multi-objective model predictive control algorithm. This enables the unit to operate in a low-carbon, economical, and stable manner under dry operating conditions.

[0077] (31)

[0078] (32)

[0079] The design of the dry-state multi-objective model predictive controller for thermal power generating units has now been completed.

[0080] S2.4.2: Design of a predictive controller for a wet-state multi-objective model.

[0081] The design process for a multi-objective model predictive controller (MAD) for unit operation in wet mode is similar to that for a dry-state multi-objective model predictive controller. When the unit is in wet operation mode, its four output variables are: output power N... E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T The controlled variable and control variable matrices in equations (6) and (9) are rewritten as follows: and The design of the wet-state multi-objective model predictive controller can be completed by executing steps

[0017] to

[0078] .

[0082] Based on the dry-state multi-objective model predictive controller and wet-state multi-objective model predictive controller obtained in steps S2.4.1 and S2.4.2, an automatic control system for dry / wet state switching is built for the unit. In step S3, the effectiveness of the designed automatic control system for dry / wet state switching is verified and analyzed using a simulation platform. The specific process is as follows:

[0083] S3.1: Select 10,000 sets of actual historical operating data of supercritical cogeneration units at 20% and 70% rated operating points respectively, and use the subspace identification method to obtain the state space model of the unit in wet and dry operating modes.

[0084] S3.2: Based on the model obtained in step S3.1, design a dry-state multi-objective model predictive controller and a wet-state multi-objective model predictive controller for the unit, and build an automatic control system for dry / wet state switching.

[0085] S3.3: External load commands are sent to the unit's dry / wet state switching automatic control system to determine the total load demand on the unit. If the total load demand exceeds 30% of the unit's rated load, the unit executes the dry-state multi-objective model predictive controller to respond quickly and accurately to the load command. If the total load demand is less than 25% of the unit's rated load, the unit executes the wet-state multi-objective model predictive controller to ensure deep peak shaving quality while maintaining the unit's safe and stable operation. If the total load demand decreases from more than 30% of the unit's rated load to below 25% of the rated load, the unit needs to switch from dry-state operation mode to wet-state operation mode. The process of switching the unit from the dry-state multi-objective model predictive controller to the wet-state multi-objective model predictive controller needs to balance the speed, low carbon footprint, and economy of load regulation.

[0086] S3.4: Observe the four controlled variables of the unit: output power N E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W .

[0087] S3.5: Water level H in the computer group's water storage tank W The time required for the unit to descend from its initial height to 0 meters; determine the unit's output power N. E Does it meet external requirements? Observe the main steam pressure P of the unit. T Main steam enthalpy T M Whether the changes are smooth. Analyze the control performance of the unit's automatic control system for dry / wet state switching.

[0088] Beneficial effects of this invention:

[0089] This invention responds to the transformation needs of clean power systems. From the perspective of control strategy design, it combines a multi-objective model predictive controller to design an automatic control method and system for dry / wet state switching of supercritical cogeneration units, thereby improving the unit's deep peak shaving capability and providing margin for absorbing larger-scale renewable energy power grid connection.

[0090] This invention fully considers the different dynamic characteristics and control difficulties of supercritical cogeneration units operating in dry and wet modes in the design of the dry / wet state switching automatic control system. It combines the low-carbon economic operation requirements under the background of deep peak shaving with the model predictive control algorithm that is good at handling multiple variables and constraints and has rolling optimization characteristics, so that the designed dry / wet state switching automatic control system can meet the deep peak shaving requirements of the new power system for thermal power generating units.

[0091] This invention simplifies supercritical cogeneration units operating in dry and wet modes into different four-input, four-output coupled systems, accurately describing the operating characteristics of supercritical thermal power generating units. Furthermore, it establishes separate dry-state and wet-state multi-objective model predictive controllers for the supercritical cogeneration units operating in dry and wet modes. In addition, a load command judgment logic bridge effectively connects the dry-state and wet-state multi-objective model predictive controllers, successfully constructing an automatic control system for dry / wet state switching of the unit. By rapidly, smoothly, and accurately completing the automatic dry / wet state switching of the working fluid in the supercritical unit's once-through boiler, it effectively improves the unit's deep peak-shaving capability while ensuring its low-carbon, economical, and stable operation. Attached Figure Description

[0092] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0093] Figure 1 This is a schematic diagram of the structure of an automatic control system for dry / wet switching of a supercritical cogeneration unit according to an embodiment of this application. Detailed Implementation

[0094] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0095] Please refer to the accompanying drawings in the instruction manual. Figure 1 , Figure 1 The automatic control system for dry / wet switching of the supercritical cogeneration unit involved in this invention mainly consists of a dry-state multi-objective model predictive controller, a wet-state multi-objective model predictive controller, and total load demand judgment logic. The supercritical cogeneration unit in dry and wet operating modes is simplified into different four-input four-output coupled systems. A multi-objective model predictive control algorithm is constructed by integrating load regulation accuracy, carbon emission cost, and power generation cost. Advanced dry-state and wet-state multi-objective model predictive controllers are designed for the supercritical cogeneration unit in dry and wet operating modes respectively. By accurately and rapidly switching between dry and wet controllers, a smooth dry / wet switching of the working fluid in the unit's once-through boiler is achieved, thereby improving the unit's deep peak-shaving capability.

[0096] S1: Analysis of the control challenges of supercritical cogeneration units operating in dry and wet modes;

[0097] S2: Establish an automatic control method and system for dry / wet state switching for deep peak shaving of supercritical cogeneration units;

[0098] S3: Verify and analyze the effectiveness of the designed automatic control method for dry / wet state transition based on the simulation platform.

[0099] S1: The purpose of thermal power generating units participating in deep peak shaving is to provide a margin for the grid connection of large-scale renewable energy power. To improve the deep peak shaving capability and power regulation quality of thermal power generating units, advanced control theory can be used to design an automatic dry / wet switching control system for once-through boilers to meet multi-objective regulation requirements. However, thermal power generating units operating at low load levels for peak shaving have low thermal efficiency, resulting in significant losses. Therefore, the requirements for dry / wet switching of once-through boilers are speed and smoothness, reducing their operating time in dry / wet mixed mode and preventing repeated switching from damaging the equipment. The task of the automatic dry / wet switching control system for once-through boilers is to automatically and quickly complete the dry / wet steam switching while taking into account the environmentally friendly and economical operation of thermal power generating units, without relying on the experience of operators. If the control effect of the automatic dry / wet switching control system meets the requirements of speed, accuracy, and stability, then the thermal power generating unit can play its important role in peak shaving and valley filling. Accurate model construction and appropriate control algorithms are the foundation for designing an automatic dry / wet switching control system. Thermal power generating units consist of two main components: a boiler and a steam turbine, supplemented by a reheat system. A boiler is a large and complex heat supply system, mainly composed of a furnace, water-cooled walls, a steam-water separator, a superheater, a spray desuperheater, a reheater, an economizer, and an air preheater. The operating mode of the unit is related to the state of the working fluid at the steam-water separator outlet. The factors affecting the stable operation of a thermal power generating unit differ significantly between dry and wet operating modes. In dry mode, the working fluid at the separator outlet is steam, and the water level in the storage tank is 0. The enthalpy of the working fluid at the separator outlet, i.e., the midpoint enthalpy, can accurately characterize the state of the working fluid. After being heated by the superheater, the steam flows through the desuperheater and finally enters the steam turbine to perform work and generate electricity. The steam flow rate entering the steam turbine is controlled by the opening of the main steam valve and is positively correlated with the unit's output power. Considering the most critical factors of the steam-water flow and unit operating state in a once-through boiler, a thermal power generating unit in dry operating mode can be simplified as a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W The four output variables are the output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T MIn wet operation mode, the working medium at the separator outlet in the boiler is steam, but the water level in the separator-storage tank is typically 2-7 meters. The water level in the separator-storage tank characterizes the degree of operation in wet mode. If the water level gradually decreases to 0, it indicates a successful switch from wet to dry operation. Therefore, a thermal power generating unit operating in wet mode can be simplified as a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T The four output variables are the output power N. E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W Regardless of whether it is dry or wet operation mode, each output variable and input variable is accompanied by complex dynamic characteristics of strong coupling, time variation, large delay and nonlinearity. In addition, the need for rapid peak shaving and valley filling makes it difficult for the unit to achieve a fast and smooth dry / wet state transition.

[0100] Based on the analysis in S1, it is evident that achieving rapid and smooth dry / wet state switching performance in supercritical cogeneration units is challenging due to the involvement of multiple variables in the control. Therefore, an advanced automatic control method and system for dry / wet state switching needs to be designed, incorporating an improved multi-objective model predictive control algorithm. Considering the requirements for low-carbon, economical, and stable operation under peak-shaving conditions, a multi-objective optimization function is constructed, taking into account the unit's carbon emission costs, power generation costs, and the smoothness of manipulated variables. Subsequently, an automatic control system for dry / wet state switching is built using a model predictive control algorithm with rolling optimization characteristics. By rolling the solution of the optimal control law under multiple constraints, the unit achieves rapid, accurate, and automatic dry / wet state switching while maintaining safe and economical operation. Since the model predictive control algorithm can handle multi-variable and multi-constraint problems, only one multi-objective model predictive controller needs to be designed for each of the unit operating in dry and wet modes. The controller for the unit in dry mode is denoted as the "dry-state multi-objective model predictive controller," and the controller for the unit in wet mode is denoted as the "wet-state multi-objective model predictive controller." Finally, by connecting the dry-state multi-objective model predictive controller and the wet-state multi-objective model predictive controller through the load command judgment logic, the design of the unit's dry / wet state switching automatic control system can be completed. Step S2 can then be specified as follows:

[0101] S2.1: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is greater than 30% of the rated load, the unit must operate in dry mode. Execute the dry-state multi-objective model predictive controller to achieve the goal of quickly and accurately meeting the external load demand while maintaining low-carbon and economical operation.

[0102] S2.2: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is less than 25% of the rated load, execute the wet-state multi-objective model predictive controller to achieve the purpose of quickly shedding load and ensuring the safe and economical operation of the unit.

[0103] S2.3: The external load demand signal is sent to the load command judgment logic. If the load demand on the unit decreases from above 30% of the rated load to below 25% of the rated load, the unit is required to participate in deep peak shaving. The unit needs to switch from dry-state operation mode to wet-state operation mode, which involves the dry / wet conversion of the once-through boiler. The control system needs to automatically switch from a dry-state multi-objective model predictive controller to a wet-state multi-objective model predictive controller. During the process of the unit's output load decreasing from 30% of the rated load to 25% of the rated load, the separator-storage tank water level shows a very obvious and rapid downward trend. When the unit's output load reaches below 25% of the rated load, the separator-storage tank water level drops to 0 meters, quickly and successfully switching to wet-state operation mode.

[0104] After determining the framework of the automatic control system for dry / wet switching of thermal power generating units based on multi-objective model predictive control algorithms, the design steps of the dry-state multi-objective model predictive controller and the wet-state multi-objective model predictive controller are specified in S2.4:

[0105] S2.4.1: Design of a predictive controller for a dry-state multi-objective model.

[0106] The three main components of a model predictive control algorithm are the predictive model, rolling optimization, and feedback correction. For an incremental model predictive control algorithm that reconstructs the state, its state prediction model is:

[0107] (33)

[0108] The state vector in the prediction time domain P has the following form:

[0109] (34)

[0110] (35)

[0111] (36)

[0112] The predicted output model is:

[0113] (37)

[0114] in,

[0115] (38)

[0116] The controlled variable is the unit's output variable: output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T M That is, the controlled variable matrix is A, B, and C are all coefficient matrices of the unit's state-space model; , and These are the augmented matrices of A, B, and C, respectively;

[0117] (39)

[0118] (40)

[0119] The control law obtained by the rolling optimization solution is in the following form:

[0120] (41)

[0121] Among them, the control quantity is the input variable of the unit: fuel quantity u B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W That is, the control matrix is The incremental optimal control law is:

[0122] (42)

[0123] (43)

[0124] (44)

[0125] To achieve deep peak-shaving capability of thermal power generating units while ensuring their low-carbon, economical, and stable operation, it is necessary to construct a system that encompasses the unit's output... Carbon emission costs Electricity generation cost C e and state stability The multi-objective optimization function J is used; the current optimal control law increment ΔU of the unit is obtained by rolling minimization of the multi-objective function J. Therefore, the optimization objectives are as follows:

[0126] (45)

[0127] The sum of the weighting factors c1, c2 and c3 is 1.

[0128] J1 is based on unit output The optimization is expressed as:

[0129] (46)

[0130] Where w(k) is the setpoint matrix of the unit output.

[0131] J2 is based on the unit's carbon emission cost. Electricity generation cost C e The optimization is expressed as:

[0132] (47)

[0133] Carbon emission costs of thermal power generating units The calculation formula is:

[0134] (48)

[0135] Wherein, Ω represents the unit's basic carbon emission allowance, and any excess must be purchased from the carbon trading market; α and β represent the carbon trading price and carbon emissions per unit load, respectively.

[0136] The power generation cost C of thermal power generating units e The calculation formula is:

[0137] (49)

[0138] Where θ is the coal price; and η is the coal consumption rate related to the unit's output load. c The expression is:

[0139] (50)

[0140] Among them, a2 and b2 are constant factors, which can be obtained by fitting the actual operating data of the unit.

[0141] J3 is based on unit state stability. Optimization, its expression is:

[0142] (51)

[0143] Among them, Q y R y Q c R c Q x and R x All are weight matrices.

[0144] The constraints are:

[0145] (52)

[0146] Among them, u max and u min These are the upper and lower bounds of the control law matrix, respectively; Δu max and Δu min These are the upper and lower limits of the control law increment matrix, respectively.

[0147] Next, the optimal control law of the multi-objective model predictive control algorithm is derived by combining J1, J2 and J3 and solved using the quadratic programming method.

[0148] Will Substituting into equation (46) and retaining only the terms relevant to optimization, then

[0149] (53)

[0150] Rewrite equation (53) in the form of a quadratic programming problem. ,but

[0151] (54)

[0152] (55)

[0153] Will Substituting into equation (47) and retaining only the terms relevant to optimization, then

[0154] (56)

[0155] in,

[0156] (57)

[0157] Rewrite equation (56) in the form of a quadratic programming problem. ,but

[0158] (58)

[0159] (59)

[0160] Will Substituting into equation (51) and retaining only the terms relevant to optimization, then

[0161] (60)

[0162] Rewrite equation 60) in the form of a quadratic program. ,but

[0163] (61)

[0164] (62)

[0165] In summary, by combining equations (61) and (62), quadratic programming is used to solve the optimal control law based on the multi-objective model predictive control algorithm. This enables the unit to operate in a low-carbon, economical, and stable manner under dry operating conditions.

[0166] (63)

[0167] (64)

[0168] The design of the dry-state multi-objective model predictive controller for thermal power generating units has now been completed.

[0169] S2.4.2: Design of a predictive controller for a wet-state multi-objective model.

[0170] The design process for a multi-objective model predictive controller (MAD) for unit operation in wet mode is similar to that for a dry-state multi-objective model predictive controller. When the unit is in wet operation mode, its four output variables are: output power N... E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T The controlled variable and control variable matrices in equations (38) and (41) are rewritten as follows: and The design of the wet-state multi-objective model predictive controller can be completed by executing steps

[0017] to

[0078] .

[0171] Based on the dry-state multi-objective model predictive controller and wet-state multi-objective model predictive controller obtained in steps S2.4.1 and S2.4.2, an automatic control system for dry / wet state switching is built for the unit. In step S3, the effectiveness of the designed automatic control system for dry / wet state switching is verified and analyzed using a simulation platform. The specific process is as follows:

[0172] This embodiment is based on the 350MW supercritical cogeneration unit of Zhuozhou Thermal Power Plant, and the method steps include:

[0173] The operating conditions for the unit at 20% and 70% load points in this example are as follows:

[0174] Table 1 Operating conditions of a 350MW supercritical cogeneration unit under dry-state model

[0175] Operating conditions <![CDATA[N E (MW)]]> <![CDATA[T I ( o C)]]> <![CDATA[P T (MPa)]]> <![CDATA[T M ( o C)]]> <![CDATA[u B (t / h)]]> <![CDATA[u W (t / h)]]> <![CDATA[μ T (%)]]> <![CDATA[D W (t / h)]]> 70% load condition point 245.102 376.29 19.60 559.79 117.32 809.54 69.73 11.81

[0176] Table 1 Operating conditions of a 350MW supercritical cogeneration unit under wet state model

[0177] Operating conditions <![CDATA[N E (MW)]]> <![CDATA[T M ( o C)]]> <![CDATA[P T (MPa)]]> <![CDATA[H W (m)]]> <![CDATA[u B (t / h)]]> <![CDATA[u W (t / h)]]> <![CDATA[μ T (%)]]> <![CDATA[μ H (%)]]> 20% load condition point 70.19 439.04 7.99 7.50 49.64 381.50 30.64 14.83

[0178] S3.1: Select 10,000 sets of actual historical operating data of supercritical cogeneration units at 20% and 70% rated operating points respectively, and use the subspace identification method to obtain the state space model of the unit in wet and dry operating modes.

[0179] S3.2: Based on the model obtained in step S3.1, design a dry-state multi-objective model predictive controller and a wet-state multi-objective model predictive controller for the unit, and build an automatic control system for dry / wet state switching.

[0180] S3.3: External load commands are sent to the unit's dry / wet state switching automatic control system to determine the total load demand on the unit. If the total load demand exceeds 30% of the unit's rated load, the unit executes the dry-state multi-objective model predictive controller to respond quickly and accurately to the load command. If the total load demand is less than 25% of the unit's rated load, the unit executes the wet-state multi-objective model predictive controller to ensure deep peak shaving quality while maintaining the unit's safe and stable operation. If the total load demand decreases from more than 30% of the unit's rated load to below 25% of the rated load, the unit needs to switch from dry-state operation mode to wet-state operation mode. The process of switching the unit from the dry-state multi-objective model predictive controller to the wet-state multi-objective model predictive controller needs to balance the speed, low carbon footprint, and economy of load regulation.

[0181] S3.4: Observe the four controlled variables of the unit: output power N E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W .

[0182] S3.5: Water level H in the computer group's water storage tank W The time required to descend from the initial height to 0 meters; calculate the load tracking error and steam superheat to determine the unit's output power N. E Does it meet external requirements? Observe the main steam pressure P of the unit. T Main steam enthalpy T M Whether the fluctuations are relatively stable. Analyze the control performance of the unit's dry / wet switching automatic control system.

[0183] Statistical results show that the dry / wet state switching automatic control system for deep peak shaving of supercritical cogeneration units proposed in this invention can enable the working fluid in the once-through boiler to complete the rapid and smooth switching between dry and wet states, laying a solid foundation for the unit to play its role in peak shaving and valley filling.

Claims

1. An automatic control method for dry / wet state switching to improve the deep peak-shaving capability of thermal power units, characterized in that: Includes the following steps: S1: Analysis of the control challenges of supercritical cogeneration units operating in dry and wet modes; S2: Establish an automatic control method for dry / wet state switching for deep peak shaving in supercritical cogeneration units; including: S2.1: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is greater than 30% of the rated load, the unit must operate in dry mode. Execute the dry-state multi-objective model predictive controller to achieve the goal of quickly and accurately meeting the external load demand while maintaining low-carbon and economical operation. S2.2: Send the external load demand signal into the load command judgment logic. If the current load demand of the unit is less than 25% of the rated load, execute the wet-state multi-objective model predictive controller to achieve the purpose of quickly shedding load and ensuring the safe and economical operation of the unit. S2.3: The external load demand signal is sent to the load command judgment logic. If the load demand on the unit decreases from above 30% of the rated load to below 25% of the rated load, the unit is required to participate in deep peak shaving. The unit needs to switch from dry operation mode to wet operation mode, which involves the dry / wet conversion of the once-through boiler. The control system needs to automatically switch from a dry-state multi-objective model predictive controller to a wet-state multi-objective model predictive controller. During the process of the unit's output load decreasing from 30% of the rated load to 25% of the rated load, the water level in the separator-storage tank shows a very obvious and rapid downward trend. When the unit's output load reaches below 25% of the rated load, the water level in the separator-storage tank drops to 0 meters, and the switch to wet operation mode is quickly and successfully completed. After determining the framework of the automatic control method for dry / wet conversion of thermal power generating units based on the multi-objective model predictive control algorithm, the design steps of the dry-state multi-objective model predictive controller and the wet-state multi-objective model predictive controller are specified in S2.4: S2.4.1: Design of a predictive controller for a dry-state multi-objective model; The three main components of a model predictive control algorithm are the predictive model, rolling optimization, and feedback correction. For an incremental model predictive control algorithm that reconstructs the state, its state prediction model is: (1) The state vector in the prediction time domain P has the following form: (2) (3) (4) The predicted output model is: (5) in, (6) The controlled variable is the unit's output variable: output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T M That is, the controlled variable matrix is A, B, and C are all coefficient matrices of the unit's state-space model. , and These are augmented matrices for A, B, and C, respectively. (7) (8) The control law obtained by the rolling optimization solution is in the following form: (9) Among them, the control quantity is the input variable of the unit: fuel quantity u B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W That is, the control matrix is The incremental optimal control law is: (10) (11) (12) To achieve deep peak-shaving capability of thermal power generating units while ensuring their low-carbon, economical, and stable operation, it is necessary to construct a system that encompasses the unit's output... Carbon emission costs Electricity generation cost C e and state stability The multi-objective optimization function J is used; the current optimal control law increment ΔU of the unit is obtained by rolling minimization of the multi-objective function J; then, the optimization objectives are as follows: (13) Among them, the sum of weighting factors c1, c2 and c3 is 1; J1 is based on unit output The optimization is expressed as: (14) Where w(k) is the setpoint matrix of the unit output; J2 is based on the unit's carbon emission cost. Electricity generation cost C e The optimization is expressed as: (15) Carbon emission costs of thermal power generating units The calculation formula is: (16) Where Ω represents the unit's basic carbon emission allowance, and any excess must be purchased from the carbon trading market; α and β represent the carbon trading price and carbon emissions per unit load, respectively. The power generation cost C of thermal power generating units e The calculation formula is: (17) Where θ is the coal price; and η is the coal consumption rate related to the unit's output load. c The expression is: (18) Among them, a2 and b2 are constant factors, which can be obtained by fitting actual operating data of the unit; J3 is based on unit state stability. Optimization, its expression is: (19) Among them, Q y R y Q c R c Q x and R x Both are weight matrices; The constraints are: (20) Among them, u max and u min These are the upper and lower bounds of the control law matrix, respectively; Δu max and Δu min These are the upper and lower limits of the control law increment matrix, respectively; Next, the optimal control law of the multi-objective model predictive control algorithm is derived by combining J1, J2 and J3 and solved using the quadratic programming method; Will Substituting into equation (14) and retaining only the terms relevant to optimization, then (21) Rewrite equation (21) in the form of a quadratic programming problem. ,but (22) (23) Will Substituting into equation (15) and retaining only the terms relevant to optimization, then (24) in, (25) Rewrite equation (24) in the form of a quadratic programming problem. ,but (26) (27) Will Substituting into equation (19) and retaining only the terms relevant to optimization, then (28) Rewrite equation (28) in the form of a quadratic programming problem. ,but (29) (30) In summary, by combining equations (31) and (32), quadratic programming is used to solve the optimal control law based on the multi-objective model predictive control algorithm. This enables the unit to operate in a low-carbon, economical, and stable manner under dry operating conditions; (31) (32) The design of the dry-state multi-objective model predictive controller for the thermal power generating unit has now been completed. S2.4.2: Design of a predictive controller for a wet-state multi-objective model; The design process for a multi-objective model predictive controller for unit operation in wet mode is similar to that for a dry-state multi-objective model predictive controller; when the unit is in wet operation mode, its four output variables are: output power N E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T The controlled variable and control variable matrices in equations (6) and (9) are rewritten as follows: and By executing step S2.4.1, the design of the wet-state multi-objective model predictive controller can be completed. S3: Verify and analyze the effectiveness of the designed automatic control method for dry / wet state transition based on the simulation platform.

2. The automatic control method for dry / wet state switching to improve the deep peak-shaving capability of thermal power units according to claim 1, characterized in that: The specific challenges in controlling the operation of the supercritical cogeneration unit under dry and wet conditions, as analyzed in step S1, are as follows: S1: A thermal power generating unit consists of two main components: a boiler and a steam turbine, supplemented by a reheat system. The boiler is a large and complex heat supply system, mainly composed of a furnace, water-cooled walls, a steam-water separator, a superheater, a spray desuperheater, a reheater, an economizer, and an air preheater. The unit's operating mode is related to the state of the working fluid at the steam-water separator outlet. The factors affecting the stable operation of a thermal power generating unit differ significantly between dry and wet operating modes. In dry mode, the working fluid at the separator outlet in the boiler is steam, and the water level in the storage tank is 0. The enthalpy of the working fluid at the separator outlet, i.e., the midpoint enthalpy, can accurately characterize the state of the working fluid. After being heated by the superheater, the steam flows through the desuperheater and finally enters the steam turbine to perform work and generate electricity. The steam flow rate entering the steam turbine is controlled by the opening of the main steam valve and is positively correlated with the unit's output power. Considering the most critical factors of the steam-water flow in the once-through boiler and the unit's operating state, the thermal power generating unit in dry operating mode can be simplified into a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Main steam valve opening degree μ T Total desuperheating water flow rate D W Four The output variable is the output power N. E Enthalpy T at the midpoint I Main steam pressure P T Enthalpy of main steam T M In wet mode, the working medium at the separator outlet in the boiler is steam, but the water level in the separator-storage tank is generally 2 to 7 meters. The water level in the separator-storage tank in the boiler can characterize the degree of operation in wet mode. If the water level in the separator-storage tank gradually decreases to 0, it means that the unit's operating mode has successfully switched from wet to dry. Therefore, the thermal power generating unit in wet operation mode can be simplified as a coupled system with four inputs and four outputs. The four input variables are fuel quantity u. B Total water supply u W Water level control valve opening degree μ in water storage tank H Main steam valve opening degree μ T Four The output variable is the output power N. E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W Regardless of whether it is in dry or wet operation mode, each output variable and input variable is accompanied by complex dynamic characteristics of strong coupling, time variation, large delay and nonlinearity. In addition, the need for rapid peak shaving and valley filling makes it difficult for the unit to achieve a fast and smooth dry / wet transition.

3. The automatic control method for dry / wet state switching to improve the deep peak-shaving capability of thermal power units according to claim 1, characterized in that: Based on the dry / wet state switching automatic control method for deep peak shaving of supercritical cogeneration units designed in step S2, step S3 verifies and analyzes the effectiveness of the designed dry / wet state switching automatic control method using a simulation platform, including: S3.1: Select 10,000 sets of actual historical operating data of supercritical cogeneration units at 20% and 70% rated operating points respectively, and use the subspace identification method to obtain the state space model of the unit in wet and dry operating modes. S3.2: Based on the model obtained in step S3.1, design a dry-state multi-objective model predictive controller and a wet-state multi-objective model predictive controller for the unit, and build an automatic control system for dry / wet state switching; S3.3: External load commands are sent to the unit's dry / wet state switching automatic control system to determine the total load demand on the unit. If the total load demand is greater than 30% of the unit's rated load, the unit executes the dry-state multi-objective model predictive controller to respond quickly and accurately to the load command. If the total load demand is less than 25% of the unit's rated load, the unit executes the wet-state multi-objective model predictive controller to ensure deep peak shaving quality while maintaining the unit's safe and stable operation. If the total load demand decreases from more than 30% of the unit's rated load to less than 25% of the rated load, the unit needs to switch from dry-state operation mode to wet-state operation mode. The process of switching the unit from the dry-state multi-objective model predictive controller to the wet-state multi-objective model predictive controller needs to take into account the speed, low carbon emissions, and economy of load regulation. S3.4: Observe the four controlled variables of the unit: output power N E Enthalpy of main steam T M Main steam pressure P T Water level H in the storage tank W ; S3.5: Water level H in the computer group's water storage tank W The time required to descend from the initial height to 0 meters; calculate the load tracking error and steam superheat to determine the unit's output power N. E Does it meet external requirements? Observe the main steam pressure P of the unit. T Main steam enthalpy T M Whether the fluctuations are relatively stable; analyze the control performance of the unit's dry / wet state switching automatic control system.

Images