Multivariable Control Method for Fast Load Change of Gas-Supercritical CO2 Thermal Cycle

Abstract

The present invention discloses a multivariable control method for rapid load change of a gas-supercritical CO2 thermodynamic cycle, which relates to the technical field of coordinated control of power generation technology, and solves the technical problems that the supercritical CO2 has poor rapid load change ability and it is difficult to regulate the nonlinear behavior of the multiphase heterogeneous energy flow in the cycle. The key points of the technical solution are as follows: aiming at the nonlinear characteristics of the cycle transient process under rapid load change, the loads of multiple heat-work conversion devices in the cycle are coordinately controlled in a way of integral sliding mode nonlinear feedback and neural network inverse model decoupling collaboration. Thus, the power of the supercritical CO2 turbine can be rapidly compensated, the rapid load change ability of the cycle can be improved, the key problem of difficult regulation of the nonlinear behavior of the multiphase heterogeneous energy flow in the cycle can be solved, and further the development space of intermittent renewable energy such as wind energy and solar energy can be expanded.

Description

Technical Field

[0001] The present application relates to the technical field of coordinated control of power generation technologies, and particularly to a multivariable control method for rapid load change of a gas-supercritical CO2 thermodynamic cycle. Background Art

[0002] The construction of a new energy supply and consumption system based on large-scale wind and solar power bases and supported by clean, efficient, advanced and energy-saving power sources around them is an important development direction for achieving the goals of carbon peak and carbon neutrality. At present, the consumption of grid-connected renewable energy electricity has shown the characteristics of high proportion, which requires coal-fired units in the regional power grid to frequently perform deep peak shaving or even start-stop peak shaving. However, solar power generation has natural uncertainty and intermittency, and wind power output has random volatility and reverse peak shaving characteristics. The large-scale grid-connected consumption of "changeable" wind and solar power requires not only high proportion but also flexibility and rapidity.

[0003] More specifically, gas turbines are strategic equipment in the energy field and national defense construction, with the advantages of clean, efficient, flexible start-stop, and rapid load change. Their load regulation rate can reach 4% / min of the rated power. Adopting "gas-electricity peak shaving" can better suppress the minute-level intermittent power fluctuations of renewable energy. The exhaust gas of gas turbines carries medium-high temperature gas waste heat, and the high-efficiency supply of natural gas power generation is often achieved through the gas-steam combined cycle. However, the rapid load change ability of the steam power bottom cycle is limited, which will limit the ability of natural gas power generation to quickly consume intermittent renewable energy electricity.

[0004] In addition, there are problems in the coordinated operation of rapid load change of the gas-supercritical CO2 thermodynamic cycle, such as the difficulty in regulating the nonlinear behavior of the transient process of the multi-phase heterogeneous energy flow in the cycle. The operating condition change range of the existing research on the transient process is limited, and the wide-load high nonlinearity of the cycle has not been fully reflected in the research on the transient process. Correspondingly, further improvements are urgently needed in this field to better meet the requirements of the construction of a new energy supply and consumption system based on large-scale wind and solar power bases and supported by clean, efficient, advanced and energy-saving power sources around them. Summary of the Invention

[0005] The present application provides a multivariable control method for rapid load change of a gas-supercritical CO2 thermodynamic cycle, and its technical purpose is to achieve rapid power compensation of the supercritical CO2 turbine, improve the rapid load change ability of the cycle, and solve the key problem of difficult regulation of the nonlinear behavior of the multi-phase heterogeneous energy flow in the cycle.

[0006] The above technical purpose of the present application is achieved through the following technical solutions:

[0007] A multivariable control method for rapid load change of a gas-supercritical CO2 thermodynamic cycle includes:

[0008] S1: Construct the inverse model of the T-S fuzzy neural network;

[0009] S2: Design the control strategy through integral sliding mode nonlinear feedback, output the control signal to the inverse model of the T-S fuzzy neural network according to the control strategy, and the inverse model of the T-S fuzzy neural network controls and adjusts the fast variable load multivariable according to the control signal;

[0010] Among them, the integral sliding mode nonlinear feedback is compensated by the power feedforward of the supercritical CO2 turbine to overcompensate for the delay in the carbon dioxide transcritical phase change process.

[0011] Furthermore, the step S1 includes:

[0012] S11: Match the antecedent of the fuzzy rule of the inverse model of the T-S fuzzy neural network through the antecedent network, and infer and generate the consequent of the fuzzy rule of the inverse model of the T-S fuzzy neural network through the consequent network;

[0013] S12: Derive the output matrix of the inverse model of the T-S fuzzy neural network, and further obtain the rank of the Jocobian matrix of the output matrix, and complete the reversibility proof of the supercritical CO2 thermodynamic cycle multi-input multi-output system through the correlation algorithm;

[0014] S13: Use the data of the cyclic transient process simulation working condition library to train and verify the inverse model of the T-S fuzzy neural network until an inverse model of the T-S fuzzy neural network that meets the preset standard is obtained.

[0015] Furthermore, in step S2, designing the control strategy through integral sliding mode nonlinear feedback includes:

[0016] Design the continuous integral type sliding mode surface;

[0017] Design the integral sliding mode controller based on the continuous integral type sliding mode surface;

[0018] When the trajectory of the supercritical CO2 thermodynamic cycle multi-input multi-output system reaches the sliding mode surface, introduce nonlinear state feedback, and the integral sliding mode controller designs the control strategy through integral sliding mode nonlinear feedback.

[0019] Furthermore, the acquisition process of the power feedforward of the supercritical CO2 turbine includes:

[0020] In the inverse model of the T-S fuzzy neural network, design the feedforward control link of the control quantity s-CO2 turbine valve opening CV st ;

[0021] Furthermore, the design parameters of the feedforward control link are obtained based on a multiphase heterogeneous energy flow quantitative correlation model, which reveals the energy storage space characteristics and dynamic correlation mechanism of the quadruple time-scale cyclic components.

[0022] Furthermore, when the feedforward control link participates in control regulation, it includes:

[0023] According to the change of the reference value of the load controlled quantity, a rapid action instruction for the supercritical CO2 turbine throttle valve is given;

[0024] According to the change of the parameters of the transcritical phase change endothermic process, the T-S type fuzzy neural network inverse model tracks the control state;

[0025] According to the power change of the transcritical CO2 turbine, the supercritical CO2 turbine throttle valve is slowly restored to complete the full-time rapid compensation of the power during the rapid load change process.

[0026] The beneficial effects of this application are as follows: Aiming at the nonlinear characteristics of the cyclic transient process under rapid load change, a new method of integral sliding mode nonlinear feedback and T-S type fuzzy neural network inverse model decoupling and coordination is used to reveal the multivariable control strategy of the gas-supercritical CO2 thermal cycle, realize the multivariable rapid and error-free regulation of the loads of multiple heat-work conversion devices in the cycle, improve the rapid load change ability of the cycle, solve the key problem of difficult regulation of the nonlinear behavior of the multiphase heterogeneous energy flow in the cycle, and further expand the development space of intermittent renewable energy such as wind energy and solar energy. Description of the Drawings

[0027] Figure 1 is the logic flow chart of the method described in this application;

[0028] Figure 2 is the flow chart of the method described in this application. Detailed Embodiments

[0029] The technical solution of this application will be described in detail below with reference to the drawings. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0030] As Figure 2 shown, the gas-supercritical CO2 thermal cycle rapid load change multivariable control method described in this application is characterized by including:

[0031] S1: Construct a T-S type fuzzy neural network inverse model to decouple multivariable input and multivariable output.

[0032] Specifically, it includes S11: Matching the antecedent of the fuzzy rules of the T-S type fuzzy neural network inverse model through the antecedent network, and inferring and generating the consequent of the fuzzy rules of the T-S type fuzzy neural network inverse model through the consequent network.

[0033] S12: Derive the output matrix of the inverse model of the T-S type fuzzy neural network, and further obtain the rank of the Jacobian matrix of the output matrix, and complete the reversibility proof of the multi-input multi-output system of the supercritical CO2 thermodynamic cycle through the correlation algorithm.

[0034] Specifically, as Figure 1 shown, the output matrix is expressed as:

[0035] Y = (y1, y2) T = (P s , P t ) T ;

[0036] where P s represents the output work of the supercritical CO2 bottom cycle; P t represents the output work of the transcritical CO2 bottom cycle;

[0037]

[0038]

[0039] where p sc represents the compressor inlet pressure of the supercritical CO2 bottom cycle, T sc represents the compressor inlet temperature of the supercritical CO2 bottom cycle, q s represents the circulating working fluid flow rate of the supercritical CO2 bottom cycle, ε s represents the cycle compression ratio of the supercritical CO2 bottom cycle; q t represents the circulating working fluid flow rate of the transcritical CO2 bottom cycle, p tc represents the carbon dioxide condensation pressure of the transcritical CO2 bottom cycle, ε t represents the cycle compression ratio of the transcritical CO2 bottom cycle; k cs represents the supercritical CO2 bottom cycle constant; k ct represents the transcritical CO2 bottom cycle constant; κ represents the adiabatic index; ρ s represents the density of supercritical CO2; ρ t represents the density of transcritical CO2; η st represents the supercritical CO2 turbine efficiency; η se represents the supercritical CO2 generator efficiency; η tt represents the transcritical CO2 turbine efficiency; η te represents the transcritical CO2 generator cycle efficiency; R represents the gas constant;

[0040] p sc 、T sc 、q s 、ε s 、qt , p tc and ε t respectively correspond to the control variables q ad , q sa , CV st , n sc , CV tt , q l and n tp ; where q ad represents the supplementary working fluid flow rate; q sa represents the s-CO2 cooler air volume; CV st represents the s-CO2 turbine valve opening; n sc represents the s-CO2 compressor speed; CV tt represents the t-CO2 turbine governor valve opening; q l represents the t-CO2 condenser LNG flow rate; n tp represents the t-CO2 working fluid pump speed;

[0041] Then the multivariable nonlinear coupling control problem of the supercritical CO2 power cycle is expressed as:

[0042]

[0043]

[0044] Among them, the state variables are expressed as: X = (x1, x2, x3, x4, x5, x6, x7) T = (p sc , T sc , q s , ε s , q t , p tc , ε t ) T ;

[0045] The input variables are expressed as: U = (u1, u2, u3, u4, u5, u6, u7) T = (q ad , q sa , CV st , n sc , CV tt , q l , n tp ) T ;

[0046] The output variables are: Y = (y1, y2) T = (P s , P t ) T .

[0047] S13: Use the data in the operating condition library to simulate the cyclic transient process to train and verify the inverse model of the T-S type fuzzy neural network until an inverse model of the T-S type fuzzy neural network that meets the preset criteria is obtained.

[0048] S2: Design the control strategy through integral sliding mode nonlinear feedback, output the control signal to the inverse model of the T-S type fuzzy neural network according to the control strategy, and the inverse model of the T-S type fuzzy neural network controls and adjusts the fast load-changing multivariable according to the control signal.

[0049] Among them, the integral sliding mode nonlinear feedback is compensated by the power feedforward of the supercritical CO2 turbine to overcompensate for the delay in the carbon dioxide transcritical phase change process.

[0050] Specifically, designing the control strategy through integral sliding mode nonlinear feedback includes:

[0051] (1) Design the continuous integral type sliding mode surface to make the sliding mode of the cyclic controlled system (i.e., the supercritical CO2 thermal cycle multi-input multi-output system) asymptotically stable;

[0052] (2) Design the integral sliding mode controller based on the continuous integral type sliding mode surface, so that the state trajectory of the cyclic controlled system is located on the sliding mode surface at the initial moment, ensuring the robustness of the integral sliding mode controller during fast load-changing operation;

[0053] (3) When the trajectory of the supercritical CO2 thermal cycle multi-input multi-output system reaches the sliding mode surface, introduce nonlinear state feedback, and the integral sliding mode controller designs the control strategy through integral sliding mode nonlinear feedback.

[0054] As a specific embodiment, the acquisition process of the power feedforward of the supercritical CO2 turbine includes: in the inverse model of the T-S type fuzzy neural network, design the feedforward control link of the control quantity s-CO2 turbine valve opening CV st and, based on the multi-phase heterogeneous energy flow quantitative correlation model, reveal the energy storage space characteristics and dynamic correlation mechanism of the four-time-scale cyclic components, and then obtain the design parameters of the feedforward control link.

[0055] The feedforward control link is divided into three stages when participating in control regulation: First, according to the change of the reference value of the load controlled quantity, give a fast action instruction for the supercritical CO2 turbine control valve; Second, according to the change of the parameters of the transcritical phase change endothermic process, the inverse model of the T-S type fuzzy neural network accurately tracks the control state; Finally, according to the power change of the transcritical CO2 turbine, slowly restore the supercritical CO2 turbine control valve to complete the full-time fast compensation of the power during the fast load-changing process.

[0056] The above are only the preferred implementation solutions of the present invention, and other effective implementation solutions are also possible. For other technical personnel in the same technical field, if they put forward effective improvements based on the present invention, these improvements should also be regarded as within the protection scope of the present invention.

Claims

1. A multivariable control method for rapid load change of a gas-supercritical CO2 power cycle, characterized in that, Including: S1: Construct the inverse model of the T-S fuzzy neural network; S2: Design the control strategy through integral sliding mode nonlinear feedback, output the control signal to the inverse model of the T-S fuzzy neural network according to the control strategy, and the inverse model of the T-S fuzzy neural network controls and adjusts the fast variable load multivariable according to the control signal; Among them, the integral sliding mode nonlinear feedback is compensated by the power feedforward of the supercritical CO2 turbine to overcompensate for the delay of the carbon dioxide transcritical phase change process; Among them, the step S1 includes: S11: Match the antecedent of the fuzzy rule of the inverse model of the T-S fuzzy neural network through the antecedent network, and infer and generate the consequent of the fuzzy rule of the inverse model of the T-S fuzzy neural network through the consequent network; S12: Derive the output matrix of the inverse model of the T-S fuzzy neural network, and further obtain the rank of the Jocobian matrix of the output matrix, and complete the reversibility proof of the multi-input multi-output system of the supercritical CO2 thermodynamic cycle through the correlation algorithm; S13: Use the data of the cyclic transient process simulation working condition library to train and verify the inverse model of the T-S fuzzy neural network until an inverse model of the T-S fuzzy neural network that meets the preset standard is obtained; In step S2, the design of the control strategy through integral sliding mode nonlinear feedback includes: Design the continuous integral type sliding mode surface to make the sliding mode of the cyclic controlled system asymptotically stable; Design the integral sliding mode controller based on the continuous integral type sliding mode surface so that the state trajectory of the cyclic controlled system is located on the sliding mode surface at the initial moment; When the trajectory of the multi-input multi-output system of the supercritical CO2 thermodynamic cycle reaches the sliding mode surface, introduce nonlinear state feedback, and the integral sliding mode controller designs the control strategy through integral sliding mode nonlinear feedback; among them, the cyclic controlled system is the multi-input multi-output system of the supercritical CO2 thermodynamic cycle.

2. The control method according to claim 1, wherein, The output matrix of the inverse model of the T-S fuzzy neural network is expressed as: Y = (y1, y2) T = (P s , P t ) T ; Among them, P s represents the output work of the supercritical CO2 bottom cycle; P t represents the output work of the transcritical CO2 bottom cycle; Among them, p sc represents the compressor inlet pressure of the supercritical CO2 bottom cycle, T sc represents the compressor inlet temperature of the supercritical CO2 bottom cycle, q s represents the circulating working fluid flow rate of the supercritical CO2 bottom cycle, ε s represents the cycle compression ratio of the supercritical CO2 bottom cycle; q t represents the circulating working fluid flow rate of the transcritical CO2 bottom cycle, p tc represents the carbon dioxide condensation pressure of the transcritical CO2 bottom cycle, ε t represents the cycle compression ratio of the transcritical CO2 bottom cycle; k cs represents the supercritical CO2 bottom cycle constant; k ct represents the transcritical CO2 bottom cycle constant; κ represents the adiabatic index; ρ s represents the density of supercritical CO2; ρ t represents the density of transcritical CO2; η st represents the supercritical CO2 turbine efficiency; η se represents the supercritical CO2 generator efficiency; η tt represents the transcritical CO2 turbine efficiency; η te represents the transcritical CO2 generator cycle efficiency; R represents the gas constant; p sc 、T sc ,q s , ε s ,q t 、p tc and ε t The corresponding control quantity is q ad ,q sa 、CV st 、n sc 、CV tt ,q l and n tp ; Among them, q ad represents the flow rate of supplementary working fluid; q sa Indicates the air volume of s-CO2 cooler; CV st Indicates the opening of the s-CO2 turbine valve; n sc Indicates the speed of the s-CO2 compressor; CV tt Indicates the opening degree of the t-CO2 turbine valve; q l Indicates t-CO2 condenser LNG flow; n tp Indicates the speed of t-CO2 working fluid pump; Then the multi-variable nonlinear coupling control problem of the supercritical CO2 thermodynamic cycle is expressed as: Among them, the state variable is expressed as: X = (x1, x2, x3, x4, x5, x6, x7) T = (p sc , T sc , q s , ε s , q t , p tc , ε t ) T ; The input variable is expressed as: U = (u1, u2, u3, u4, u5, u6, u7) T = (q ad , q sa , CV st , n sc , CV tt , q l , n tp ) T ; The output variable is: Y = (y1, y2) T = (P s , P t ) T .

3. The control method according to claim 2, characterized in that, The acquisition process of the power feedforward of the supercritical CO2 turbine includes: In the inverse model of the T-S fuzzy neural network, the feedforward control link of the control variable, the opening degree CV of the s-CO2 turbine valve, is designed. st ​ 4. The control method according to claim 3, wherein The design parameters of the feedforward control link are obtained based on the multi-phase heterogeneous energy flow quantitative correlation model, revealing the energy storage space characteristics and dynamic correlation mechanism of the four-time-scale cyclic components.

5. The control method according to claim 4, wherein When the feedforward control link participates in control and adjustment, it includes: Give a fast action command for the supercritical CO2 turbine throttle according to the change of the reference value of the load controlled quantity; The inverse model of the T-S fuzzy neural network tracks the control state according to the change of the parameters of the transcritical phase change endothermic process; Slowly restore the supercritical CO2 turbine throttle according to the power change of the transcritical CO2 turbine to complete the full-time fast compensation of the power during the fast variable load process.

Images