Method and system for regulating and controlling NOx concentration of flue gas at SCR inlet of green ammonia blending burning coal-fired unit

Abstract

The embodiment of the invention provides a method and system for regulating and controlling the concentration of NOx in flue gas at an SCR inlet of a green ammonia blending burning coal-fired unit, and belongs to the technical field of clean emission. The method comprises the following steps: constructing a corresponding reaction mechanism model of an ammonia gas and coal combustion reaction based on coal-fired unit design information; on the basis of the reaction mechanism model, an optimization problem is constructed with the minimum NOx concentration ratio of the SCR inlet flue gas as a target, a constraint function is added to the optimization problem, and an optimization model is obtained; based on the optimization model and the long-short-term memory network, analyzing the influence degree of each operating parameter of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet to obtain a NOx concentration prediction model of the coal-fired unit; and based on real-time operation parameters of the coal-fired unit and the NOx concentration prediction model, performing real-time optimization control on the coal-fired unit so as to ensure that the coal-fired unit is discharged at the lowest NOx concentration. According to the scheme, the problems that in a traditional method, ammonia gas flow control is difficult, and NOx concentration fluctuation is large are solved, and therefore the environment-friendly performance and the operation efficiency of a coal-fired unit are improved.

Description

Technical Field

[0001] The present invention relates to the field of clean emission technology, and in particular to a method for controlling the NOx concentration in flue gas at an SCR inlet of a coal-fired unit burning green ammonia and a system for controlling the NOx concentration in flue gas at an SCR inlet of a coal-fired unit burning green ammonia. Background Art

[0002] As global climate change intensifies, the impact on the environment and society becomes more and more significant, and countries have set carbon peak and carbon neutrality as important goals. In this context, the energy industry, especially the coal-fired power generation industry, which is mainly based on coal, is facing tremendous pressure and urgently needs to reduce greenhouse gas emissions and improve energy efficiency through technological innovation and industrial transformation.

[0003] Although new energy technologies have developed significantly in recent years, my country's electricity supply is still dominated by thermal power generation, with coal as the main energy source, due to the instability of natural conditions (such as solar and wind power). Therefore, how to reduce the impact on the environment on the basis of coal-fired power generation has become an urgent problem to be solved in the current energy industry. The boiler of a coal-fired power plant is a core equipment. The large amount of nitrogen oxides (NOx) produced during the combustion process is one of the main air pollutants, posing a serious threat to air quality and human health. NOx emission control is not only closely related to environmental protection policies, but also has an important impact on the economy and thermal efficiency of coal-fired units.

[0004] At present, the blending of green ammonia has attracted attention as a potential solution. Green ammonia is a carbon-free fuel produced through renewable energy, which can reduce carbon emissions while reducing the generation of NOx. However, the existing solutions still face multiple challenges in practical applications, especially during the combustion process, it is difficult to control the opening of the ammonia valve, and it is easy for ammonia to escape excessively or under-react. In addition, the NOx concentration of the flue gas at the inlet of the selective catalytic reduction (SCR) is difficult to control accurately, mainly because the combustion system is affected by a variety of disturbance factors and system dynamic characteristics. The existing technology lacks effective means to deal with these complex factors, and it is difficult to maintain the stability and economy of the combustion system while ensuring that NOx emissions meet the standards. Therefore, how to achieve real-time and accurate control of the NOx concentration at the SCR inlet under complex working conditions and effectively control the flow of ammonia has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a method and system for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit burning green ammonia, so as to at least solve the problem that it is difficult to achieve real-time and accurate control of the NOx concentration at the SCR inlet under the current complex working conditions.

[0006] In order to achieve the above-mentioned purpose, the first aspect of the present invention provides a method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit that burns green ammonia. The method is applied to predicting and controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit that burns green ammonia. The method comprises: constructing a corresponding reaction mechanism model of the ammonia and coal combustion reaction based on the design information of the coal-fired unit; constructing an optimization problem based on the reaction mechanism model with the goal of minimizing the proportion of NOx concentration in the flue gas at the SCR inlet as the goal, and adding a constraint function to the optimization problem to obtain an optimization model; based on the optimization model and a long short-term memory network, analyzing the influence of various operating parameters of the coal-fired unit on the NOx concentration in the flue gas at the SCR inlet to obtain a NOx concentration prediction model for the coal-fired unit; based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model, performing real-time optimization control of the coal-fired unit to ensure that the coal-fired unit maintains the lowest NOx concentration emission; wherein the real-time optimization control of the coal-fired unit at least includes the opening size control of the green ammonia supply valve.

[0007] Optionally, the reaction mechanism model of the corresponding ammonia and coal combustion reaction is constructed based on the coal-fired unit design information, including: constructing a green ammonia production model based on the coal-fired unit design information; based on the green ammonia production model, constructing a reaction mechanism model of the ammonia and coal combustion reaction in the corresponding furnace; wherein the green ammonia production model is constructed by a synthetic ammonia device, a renewable energy power generation device and a water electrolysis hydrogen production device.

[0008] Optionally, the reaction mechanism model of the combustion reaction of ammonia and coal is:

[0009] in, is the generation rate of NOx; is the NOx generation rate constant; is the mass flow rate of coal; is the mass fraction of nitrogen in coal; is the mass fraction of unburned nitrogen; is the reduction rate of NOx; is the process efficiency value; is the mass flow rate of green ammonia; is the concentration of NOx.

[0010] Optionally, the optimization problem is expressed as:

[0011] in, is the proportion of NOx concentration in the flue gas at the SCR inlet; the constraint functions include: any one or more of the energy balance constraint, the green ammonia proportion constraint of the total fuel blending, the ammonia escape constraint at the boiler outlet, the oxygen concentration constraint of the flue gas at the boiler outlet, the unit output constraint and the non-negative constraint.

[0012] Optionally, based on the optimization model and the long short-term memory network, the influence of various operating parameters of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet is analyzed to obtain the NOx concentration prediction model of the coal-fired unit, including: constructing a long short-term memory network based on the optimization model, wherein the long short-term memory network includes a memory unit, an input gate, a forget gate and an output gate, and a linear regression layer at the end; based on the long short-term memory network, simulating the influence of various parameters on the NOx concentration at the SCR inlet under different operating conditions, and taking the parameters with an influence greater than a preset influence threshold as key parameters; and constructing the NOx concentration prediction model of the coal-fired unit based on the key parameters.

[0013] Optionally, the NOx concentration prediction model for the coal-fired unit is constructed based on key parameters, including: constructing an initial framework of the NOx concentration prediction model based on selected key parameters to clarify the input and output of the model; training the NOx concentration prediction model based on existing historical data and the initial framework of the NOx concentration prediction model to obtain an initial NOx concentration prediction model; verifying and testing the prediction model based on an unused data set, and using the verified model as the NOx concentration prediction model.

[0014] Optionally, the real-time optimal control of the coal-fired unit is performed based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model to ensure that the coal-fired unit maintains the lowest NOx concentration emission, including: collecting the real-time operating parameters of the coal-fired unit, and using the real-time operating parameters as the input parameters of the NOx concentration prediction model, and outputting the NOx concentration prediction value at the SCR inlet based on the model; wherein, in the process of the model outputting the NOx concentration prediction value at the SCR inlet, additional disturbance compensation is performed based on the expanded state observer; based on the NOx concentration prediction value at the SCR inlet, it is judged whether the prediction value is the minimum NOx concentration under the corresponding operating conditions; if not, generating the current optimal control strategy based on the generalized predictive control model; and generating an adjustment plan for each execution component based on the optimal control strategy as an optimal control plan.

[0015] Optionally, the extended state observer is expressed as:

[0016] in, and is the derivative of the extended state variable of the extended state observer; and is the detector gain parameter; is the observer bandwidth; is the state variable of the coal-fired unit; is the control gain coefficient; is the input parameter of the extended state observer; is the output parameter after additional disturbance compensation.

[0017] Optionally, the generalized predictive control model is:

[0018] in, Indicates the control amount increment Make the objective function minimize; is the predicted value of NOx concentration; is the expected value of NOx concentration; T is a matrix; To control the weight of the increment.

[0019] The second aspect of the present invention provides a NOx concentration control system for the SCR inlet flue gas of a coal-fired unit blended with green ammonia, which is applied to the prediction and control of the NOx concentration of the SCR inlet flue gas of a coal-fired unit blended with green ammonia. The system comprises: an initial unit, which is used to construct a corresponding reaction mechanism model of the ammonia and coal combustion reaction based on the design information of the coal-fired unit; a model construction unit, which is used to construct an optimization problem based on the reaction mechanism model with the goal of minimizing the proportion of NOx concentration in the SCR inlet flue gas, and add a constraint function to the optimization problem to obtain an optimization model; a prediction unit, which is used to analyze the influence of various operating parameters of the coal-fired unit on the NOx concentration of the SCR inlet flue gas based on the optimization model and a long short-term memory network, and obtain a NOx concentration prediction model for the coal-fired unit; a control unit, which is used to perform real-time optimization control of the coal-fired unit based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model to ensure that the coal-fired unit maintains the lowest NOx concentration emissions.

[0020] On the other hand, the present invention provides a computer-readable storage medium having instructions stored thereon, which, when executed on a computer, enables the computer to execute the above-mentioned method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit blended with green ammonia.

[0021] Through the above technical scheme, the scheme of the present invention accurately describes the generation mechanism of nitrogen oxides (NOx) during the combustion process by constructing a reaction mechanism model of the combustion reaction of ammonia and coal, and establishes an optimization problem with the goal of minimizing the NOx concentration of flue gas at the inlet of SCR. By adding constraint functions, an optimization model is generated, which can identify key operating parameters under complex working conditions. Subsequently, the operating parameters of the coal-fired unit are analyzed in combination with the long short-term memory network (LSTM) to predict the changes in the NOx concentration at the inlet of SCR under different operating conditions. Based on this prediction model, the system can optimize and control the real-time operating data of the coal-fired unit, especially adjust the opening of the green ammonia supply valve to ensure that the NOx emission concentration is always maintained at the lowest level. This technology can realize dynamic adjustment of the combustion process, solve the problems of difficult ammonia flow control and large fluctuations in NOx concentration in traditional methods, thereby improving the environmental protection performance and operating efficiency of the coal-fired unit.

[0022] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings: Figure 1 It is a flowchart of the steps of a method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit mixed with green ammonia provided by one embodiment of the present invention; Figure 2 This is a manufacturing flow chart of green ammonia provided by one embodiment of the present invention; Figure 3 It is a flow chart of coal-fired unit control and optimization design provided by one embodiment of the present invention; Figure 4 It is a schematic diagram of a NOx model at the outlet of a coal-fired unit based on LSTM provided by an embodiment of the present invention; Figure 5 It is a schematic diagram of the optimization process of the NOx concentration of the flue gas at the SCR inlet provided by one embodiment of the present invention; Figure 6 is a control structure block diagram provided by an embodiment of the present invention; Figure 7 It is a control effect comparison diagram of IPC and TRD control strategies provided by an embodiment of the present invention under different flue gas NOx concentration setting values; Figure 8 It is a comparison diagram of NOx concentration in flue gas at the SCR inlet under different load conditions provided by one embodiment of the present invention; Fig. 9It is a system structure diagram of a NOx concentration control system for the SCR inlet flue gas of a coal-fired unit with green ammonia combustion provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0024] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0025] Figure 1 The method flow chart of the method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit with green ammonia combustion is provided in one embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a method of xx, the method comprising: Step S10: constructing a corresponding reaction mechanism model of the ammonia and coal combustion reaction based on the coal-fired unit design information.

[0026] Specifically, a green ammonia production model is constructed based on the design information of the coal-fired unit; based on the green ammonia production model, a reaction mechanism model of the ammonia and coal combustion reaction in the furnace is constructed; wherein the green ammonia production model comprises a synthetic ammonia device, a renewable energy power generation device and a water electrolysis hydrogen production device.

[0027] In the embodiment of the present invention, Figure 2 , green ammonia production adopts three core process paths: the traditional Haber-Bosch method, flexible process, and new photocatalytic, plasma or electrochemical processes. Among them, the Haber-Bosch method couples the green hydrogen generated by renewable energy with nitrogen to react to generate green ammonia through the action of iron-based catalysts under thermal catalysis, high temperature and high pressure conditions. The reaction conditions of this process are high pressure of 20~50MPa and high temperature of 350~500℃. The upper part of the device is a contact chamber equipped with an iron-based catalyst, and the lower part is a heat exchanger. Hydrogen and nitrogen are preheated, reacted and cooled in this system to finally generate liquid ammonia. The unreacted hydrogen and ammonia are circulated into the synthesis tower for reuse.

[0028] Furthermore, on this basis, the green ammonia production model combines renewable energy power generation devices (such as wind power or solar power) and water electrolysis hydrogen production devices, using green electricity to drive the water electrolysis reaction to generate green hydrogen, ensuring that the carbon footprint of the ammonia production process is minimized. Through the cooperation of energy storage facilities, the stability of production is ensured, and the impact of new energy fluctuations on hydrogen generation and ammonia synthesis is avoided. The constructed green ammonia production model is then used to model the reaction mechanism of the ammonia and coal combustion reaction in the furnace, especially for the combustion process of ammonia and coal and its reaction characteristics in a high temperature environment. Through this model, the impact of ammonia and coal on NOx generation and reduction during the combustion process can be accurately analyzed, further providing accurate data support for NOx emission control of coal-fired units.

[0029] Based on the scheme of the present invention, the scheme of the present invention integrates the production process of green ammonia, which not only meets the demand for clean production of ammonia, but also provides low-carbon fuel for coal-fired units, effectively reducing greenhouse gas emissions. At the same time, the efficient production and application of green ammonia can reduce the generation of NOx and achieve precise control of NOx emissions.

[0030] Furthermore, in the furnace, the combustion reaction of ammonia and coal can be described by the following chemical reaction to describe the generation and reduction of NOx:

[0031]

[0032]

[0033]

[0034]

[0035] The output of the furnace includes the heat energy produced by combustion, nitrogen in the flue gas, water vapor, unreacted ammonia, remaining NOx and other combustion products. In order to convert these chemical reactions into physical expressions, the following aspects are considered: 1) The heat balance in the furnace can be described by the energy conservation equation:

[0036] in, The heat input from the combustion of fuel, is the heat lost through flue gas and furnace wall, The heat stored in water and steam.

[0037] Furthermore, the mass flow rate is expressed as:

[0038] in, represents the mass flow rate, and the subscripts represent different substances.

[0039] Based on this, the reaction mechanism model of the ammonia and coal combustion reaction is:

[0040] in, is the generation rate of NOx; is the NOx generation rate constant; is the mass flow rate of coal; is the mass fraction of nitrogen in coal; is the mass fraction of unburned nitrogen; is the reduction rate of NOx; is the process efficiency value; is the mass flow rate of green ammonia; is the concentration of NOx.

[0041] Step S20: Based on the reaction mechanism model, an optimization problem is constructed with the goal of minimizing the proportion of NOx concentration in the flue gas at the SCR inlet, and a constraint function is added to the optimization problem to obtain an optimization model.

[0042] Specifically, the optimization problem is expressed as:

[0043] in, is the proportion of NOx concentration in the flue gas at the SCR inlet; the constraint functions include: any one or more of the energy balance constraint, the green ammonia proportion constraint of the total fuel blending, the ammonia escape constraint at the boiler outlet, the oxygen concentration constraint of the flue gas at the boiler outlet, the unit output constraint and the non-negative constraint.

[0044] In the embodiment of the present invention, the minimum proportion of NOx concentration in the flue gas at the SCR inlet is used as the criterion. That is, given the proportion of green ammonia in the total fuel blending, the ammonia escape at the boiler outlet, the oxygen concentration of the flue gas at the boiler outlet, and the unit output, various constraints are met to minimize the NOx at the flue gas outlet during the control period. Specifically, each constraint is expressed as: 1) Energy balance constraints: ; 2) Constraints on the proportion of green ammonia in total fuel blending: ; 3) Constraints on ammonia escape at boiler outlet: ; 4) Constraints on oxygen concentration in flue gas at boiler outlet ; 5) Unit output constraints: ; 6) Non-negative constraints: The above variables are non-negative.

[0045] in, and The heat of combustion of coal and ammonia; the heat of combustion of coal; The proportion of green ammonia in blending; and is the ammonia escape amount at the boiler outlet and the maximum value allowed; , and are the masses of coal, air volume and ammonia respectively; is the oxygen concentration of flue gas at boiler outlet; , and They are the unit output and the allowed range of constraint values ​​respectively.

[0046] SCR (flue gas denitrification technology) inlet flue gas NOx content setting value: boiler outlet flue gas volume = SCR inlet flue gas volume, expressed as:

[0047] in, Total SCR inlet flue gas volume.

[0048] Step S30: Based on the optimization model and the long short-term memory network, the influence of various operating parameters of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet is analyzed to obtain a NOx concentration prediction model for the coal-fired unit.

[0049] Specifically, a long short-term memory network is constructed based on the optimization model, wherein the long short-term memory network includes a memory unit, an input gate, a forget gate, an output gate, and a linear regression layer at the end; based on the long short-term memory network, the influence of various parameters on the NOx concentration at the SCR inlet under different operating conditions is simulated, and the parameters with an influence greater than a preset influence threshold are used as key parameters; and a NOx concentration prediction model for coal-fired units is constructed based on the key parameters.

[0050] In the embodiment of the present invention, LSTM (Long Short-Term Memory Network) is a special RNN that learns long-term dependency information while avoiding the gradient vanishing problem. LSTM adds a structure called a memory cell to the neurons in the hidden layer of RNN to remember past information, and adds three gate structures (input gate, forget gate, output gate) to control the use of historical information.

[0051] Assume the input sequence is , the hidden layer state is , then at time t we have:

[0052]

[0053]

[0054]

[0055]

[0056] in, represents the sigmoid activation function, represents the gate activation function (usually tanh), represents the Hadamard product (element-wise product), W and b represent weight and bias respectively. Indicates the hidden state, Indicates input, , , and They represent the update gate, forget gate, output gate and unit state respectively. In order to make LSTM meet the purpose of prediction, a linear regression layer needs to be added, namely:

[0057] in, Output representing the final prediction result; Represents the threshold of the linear regression layer.

[0058] In the embodiment of the present invention, in the application scenario of NOx concentration prediction, the long short-term memory network (LSTM) can effectively process the time series data of the operating parameters of the coal-fired unit, especially the problem of NOx concentration changes in complex dynamic systems. LSTM can avoid the problem of gradient disappearance while maintaining the memory of historical data by introducing memory units and the structure of input gates, forget gates and output gates.

[0059] During the application process, the various operating parameters of the coal-fired unit are first input, such as the combustion temperature, oxygen concentration, fuel flow rate, etc. of the boiler. These parameters are input into the LSTM network as a time series, and the input of each time step is combined with the hidden state of the previous moment. LSTM determines how much of the current input information is updated to the memory unit through the input gate, and controls whether the past information needs to be discarded through the forget gate. The memory unit can effectively process long-term dependent inputs through the dynamic adjustment of these gates.

[0060] Next, LSTM processes the hidden state of each time step according to the prediction task and outputs the predicted value of NOx concentration. The output gate plays a key role in this process, which determines how much the hidden state of the current time step can affect the predicted value. In order to make the model more in line with the prediction requirements, a linear regression layer is finally added at the end of the LSTM network to map the processed hidden state to the predicted output of NOx concentration.

[0061] Based on the solution of the present invention, the application of LSTM in NOx concentration prediction can effectively predict the changing trend of NOx concentration at the SCR inlet by capturing the dynamic changes of various operating parameters in the coal-fired unit. Compared with traditional linear models or simple time series models, LSTM has a stronger ability to learn complex time dependencies, can process information over a long time span, and effectively avoids the gradient vanishing problem in traditional recurrent neural networks (RNNs). In this way, LSTM can accurately predict NOx concentration, thereby providing important decision-making basis for the optimization control of coal-fired units, helping to minimize NOx emissions and meet environmental protection standards.

[0062] Step S40: Based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model, the coal-fired unit is subjected to real-time optimization control to ensure that the coal-fired unit maintains the lowest NOx concentration emission.

[0063] Specifically, the real-time operating parameters of the coal-fired unit are collected and used as input parameters of the NOx concentration prediction model, and the NOx concentration prediction value at the SCR inlet is output based on the model; wherein, in the process of the model outputting the NOx concentration prediction value at the SCR inlet, additional disturbance compensation is performed based on the expanded state observer; based on the NOx concentration prediction value at the SCR inlet, it is judged whether the prediction value is the minimum NOx concentration under the corresponding operating condition; if not, the current optimal control strategy is generated based on the generalized predictive control model; based on the optimal control strategy, an adjustment plan for each execution component is generated as an optimal control plan.

[0064] In the embodiment of the present invention, Figure 4 First, the real-time parameters of the coal-fired unit are collected, including the combustion temperature of the boiler, air and fuel flow, etc. These parameters are passed as input to the NOx concentration prediction model, which uses a combination of historical data and real-time data to predict the NOx concentration at the SCR (selective catalytic reduction) inlet.

[0065] During the prediction process, the extended state observer (ESO) is introduced to compensate for disturbances and dynamic changes not considered in the model. ESO ensures that the prediction results are more accurate and robust by monitoring the changes in external disturbances in real time and compensating for them. At this time, the system will determine whether the predicted value of NOx concentration output by the model is the minimum value of NOx under the current operating conditions. If the predicted value is not the optimal value, the generalized predictive control (GPC) model is introduced to generate the optimal control strategy based on the current system state.

[0066] Furthermore, generalized predictive control predicts future operating conditions by combining current status and historical data, and calculates the optimal control solution to minimize NOx concentration. This control solution generates a specific adjustment plan for the execution components to adjust the combustion conditions in the boiler, such as fuel input, air volume, ammonia flow, etc., to minimize NOx emissions.

[0067] Based on the scheme of the present invention, Figure 5 , by combining the NOx concentration prediction model, extended state observer and generalized predictive control (GPC), real-time optimization control of coal-fired units under complex operating conditions is achieved. First, the extended state observer effectively compensates for the impact of external disturbances on the prediction model to ensure the accuracy and stability of the prediction. Secondly, the generalized predictive control generates the optimal control strategy based on the prediction results to ensure that the NOx concentration is always at the lowest level under various operating conditions. Through this real-time optimization control, not only can NOx emissions be effectively reduced and environmental protection requirements be met, but the overall operating efficiency of the coal-fired unit can also be optimized, and the economy and stability of the system can be improved.

[0068] In one possible implementation, the extended state observer strategy is as follows: Reorganize the system into a first-order form as an uncertain system, and the expression is:

[0069] In the formula: g is the unknown dynamic characteristic; b is the key gain, which indicates the effect of the control quantity u on the controlled quantity y.

[0070] When modeling, b is difficult to obtain accurately, and the formula is as follows:

[0071] Where: b 0 ——approximation of b; ——Total disturbance, including unknown external disturbances and internal dynamic characteristics.

[0072] Let y = x 1 , f = x 2 , the expanded state space expression is:

[0073] Its extended state observer (ESO) can be designed as:

[0074] Where: ωo is the observer bandwidth. The larger the ωo is, the closer the observed value of the system is to the actual value. However, if it is too large, it will produce noise influence, so it should be considered comprehensively during adjustment.

[0075] Furthermore, the controlled autoregressive integrated moving average (CARIMA) model of generalized predictive control (GPC) describes the controlled object as:

[0076] in, and are the control quantity and output of the system at time t respectively; and They represent the backshift operator of and polynomial of order, and is the first polynomial; is white noise with zero mean and bounded variance; k is the minimum delay of the system; is the difference operator.

[0077] According to the GPC principle and based on the Diophantine equation and the controlled object, the optimal prediction of the system output at time t for the future time t+k+i can be obtained as follows:

[0078] The known quantities in the optimal prediction are:

[0079] in, and is the Diophantine equation For the polynomial In the range, p is the prediction step length, is the lth coefficient of the plant step response, which can be written in matrix form as , the actual output is , E is the error vector, given by Determined together with the Diophantine equation.

[0080] The set value is softened according to the following formula:

[0081] in, is the set value at time t, is the setting value after softening, is the softening factor, .

[0082] The optimal control strategy adopted by GPC is to minimize the following quadratic objective function:

[0083] in, , To control the step size, , To control the weight of the increment.

[0084] Preferably, during the combustion process of the boiler, the emission of nitrogen oxides (NOx) not only has a significant impact on the environment, but also affects the operating efficiency of the boiler. Therefore, in order to reduce the emission of NOx and improve the combustion efficiency, a comprehensive intelligent control and optimization system is introduced. The core of the system is an optimization model based on the long short-term memory network (LSTM), which can effectively process complex time series data during the operation of the boiler. The LSTM network accurately predicts the emission of NOx under different operating conditions by memorizing and analyzing the changing trends of historical operating parameters. Its unique input gate, forget gate and output gate structure enables LSTM to handle long-term dependency problems and avoid gradient vanishing phenomena, thereby accurately predicting NOx emissions. Through the LSTM model, the system can identify and establish a mathematical model in the boiler combustion process and accurately simulate the influence of boiler operating parameters on NOx generation. This mathematical model provides a basis for the subsequent design of controllers and control methods. The controller adjusts the operating parameters of the boiler in real time according to the model prediction results, such as fuel input, air flow, and ammonia flow, thereby optimizing the combustion process.

[0085] Based on the scheme of the present invention, by introducing the LSTM model, accurate prediction and control of NOx emissions during boiler operation are achieved, significantly reducing NOx emissions. At the same time, the optimization control process also improves combustion efficiency, allowing the boiler to maintain environmental performance while saving fuel. Through the establishment of a mathematical model, the control strategy of the entire system is more flexible and efficient, and can adapt to changes in various working conditions, providing a guarantee for the stability and economy of boiler operation.

[0086] Preferably, in the control strategy, an intelligent feedforward controller uses the set value and measured value of the inlet NOx content to adjust the opening of the green ammonia control valve in real time through an extended state observer and a NOx concentration prediction model. This feedforward control strategy ensures accurate control of the NOx concentration at the boiler outlet. At the same time, the generalized predictive control (GPC) controller optimizes the control strategy based on the information of the prediction model and the observer to adapt to different operating conditions and maintain low NOx emissions.

[0087] In the embodiment of the present invention, in the control strategy, the intelligent feedforward controller uses the NOx set value and actual measurement value at the SCR inlet in real time, combined with the extended state observer (ESO) and the NOx concentration prediction model, to accurately control the opening of the green ammonia valve. The role of this feedforward controller is to predict and compensate for external disturbances and system dynamic characteristics in advance, so that the system can still maintain the stability of NOx concentration when facing a complex operating environment. The extended state observer can detect unmodeled disturbances in real time and compensate for the impact of NOx concentration changes through dynamic adjustment, further improving the robustness of the system.

[0088] In addition, the generalized predictive control (GPC) controller also plays a vital role in this strategy. Based on the output of the prediction model and the information fed back by the ESO, the GPC optimizes the calculation of the changing trend of NOx concentration under different operating conditions to ensure that the system can maintain low NOx emissions under different load and combustion conditions. By combining feedforward control and GPC control, this dual control strategy ensures the system's rapid response and predictive capabilities to disturbances.

[0089] Based on the scheme of the present invention, precise control and optimization of NOx concentration in coal-fired boilers are achieved. The intelligent feedforward controller reduces the fluctuation of NOx concentration and maintains stability under dynamic conditions by predicting and adjusting the green ammonia valve in advance. The extended state observer further enhances the system's ability to resist external disturbances, making the control more precise. Generalized predictive control (GPC) continuously optimizes the control strategy according to the actual operating conditions to ensure that the system always maintains the lowest NOx emission level under different conditions. This scheme not only improves environmental benefits, but also improves combustion efficiency and system operation stability.

[0090] Preferably, Figure 6 , the NOx concentration controller works with the generalized predictive control (GPC) controller to form an intelligent and automated coal-fired boiler control system. The NOx concentration controller monitors the NOx concentration at the boiler outlet in real time, and adjusts the system's operating parameters in real time by comparing the actual emissions with the set environmental standards. In particular, for the opening size of the gaseous green ammonia valve, the controller dynamically adjusts according to the actual operating conditions to ensure that the NOx emissions remain within the specified range. At the same time, the GPC controller further optimizes the control strategy based on the prediction model of the combustion process and the data fed back by the expanded state observer (ESO) to ensure that the combustion system can adapt to changing operating conditions. Through the close cooperation of the NOx concentration controller and the GPC controller, not only can the NOx emissions be accurately controlled, but also the various parameters in the combustion process can be optimized, thereby improving energy utilization efficiency. The control system adopts advanced control algorithms and observers, which can intelligently manage the boiler's combustion process, achieve real-time monitoring and adjustment, and enable the boiler to maintain the best operating state under different loads and combustion conditions.

[0091] Based on the scheme of the present invention, the environmental protection and energy efficiency performance of coal-fired boilers are significantly improved by integrating NOx concentration controller and GPC controller. It can accurately control NOx emissions under dynamic conditions, ensure that the boiler meets environmental protection standards, and improve energy utilization efficiency. The system design achieves efficient combustion and stable operation of the boiler by intelligently adjusting the opening of the green ammonia valve, reducing energy waste and significantly reducing pollution to the environment. This integrated control system has strong adaptability and can cope with combustion challenges under various working conditions, ensuring the best results in energy saving and environmental protection.

[0092] Example: In this example, the coal-fired unit of the 600MW supercritical unit boiler of the thermal power plant is subjected to different load conditions. The model information is obtained by using the LSTM model under the conditions of coal powder mixed with ammonia and without ammonia. The model information is introduced into the ammonia-mixed combustion control method based on intelligent prediction and control (IPC) and compared with the traditional coal powder mixed with ammonia combustion. The control effect comparison results of the two control strategies are shown in the figure below. Figure 7 The corresponding control indicators are given, including rise time tr, adjustment time ts, decay rate φd and overshoot σ. The comparison of control algorithm performance indicators under different working conditions is shown in Table 1. It can be seen from the table that when the NOx concentration of the SCR inlet flue gas is controlled and compared under two different set values, IPC is better than TRD.

[0093] Table 1 Comparison of control algorithm performance indicators under different setting values

[0094] Test 2: This example uses a 600MW supercritical coal-fired boiler as the object for control and optimization design research. By comparing the ammonia combustion of intelligent prediction and control (IPC) and the traditional pulverized coal (TRD) combustion, the NOx concentration of the flue gas at the SCR inlet is measured and compared. As the load increases and the proportion of pulverized coal increases, the NOx concentration of the flue gas at the SCR inlet shows different trends. The overall performance is that the NOx concentration of the flue gas after ammonia mixing is lower than that of the traditional pulverized coal combustion, such as Figure 8 As shown. When ammonia is blended with pulverized coal, the presence of ammonia can inhibit the formation of thermal NOx and can react with fuel NOx to reduce the NOx concentration in the flue gas. The ammonia blending ratio and combustion conditions need to be carefully controlled to ensure effective NOx reduction while avoiding excessive ammonia escaping into the flue gas. This usually involves real-time monitoring of NOx and ammonia concentrations, and adjusting the blending ratio to optimize the reaction efficiency.

[0095] Fig. 9 This is a system structure diagram of a NOx concentration control system for the SCR inlet flue gas of a coal-fired unit with green ammonia combustion provided by an embodiment of the present invention. Fig. 9 As shown, an embodiment of the present invention provides a system for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit blended with green ammonia, the system comprising: an initial unit, for constructing a corresponding reaction mechanism model of the ammonia and coal combustion reaction based on the design information of the coal-fired unit; a model construction unit, for constructing an optimization problem based on the reaction mechanism model with the goal of minimizing the proportion of NOx concentration in the flue gas at the SCR inlet, and adding a constraint function to the optimization problem to obtain an optimization model; a prediction unit, for analyzing the degree of influence of various operating parameters of the coal-fired unit on the NOx concentration in the flue gas at the SCR inlet based on the optimization model and a long short-term memory network, and obtaining a NOx concentration prediction model for the coal-fired unit; a control unit, for performing real-time optimization control of the coal-fired unit based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model, so as to ensure that the coal-fired unit maintains the lowest NOx concentration emissions.

[0096] An embodiment of the present invention also provides a computer-readable storage medium, which stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit blended with green ammonia.

[0097] Those skilled in the art will understand that all or part of the steps in the method for implementing the above-mentioned embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a single-chip microcomputer, a chip or a processor execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0098] The optional embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the technical concept of the embodiments of the present invention, the technical scheme of the embodiments of the present invention can be subjected to a variety of simple modifications, and these simple modifications all belong to the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0099] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.

Claims

1. A method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit burning green ammonia, the method being applied to predicting and controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit burning green ammonia, characterized in that: The method comprises: Based on the design information of the coal-fired unit, a corresponding reaction mechanism model of the ammonia and coal combustion reaction is constructed; Based on the reaction mechanism model, an optimization problem is constructed with the goal of minimizing the proportion of NOx concentration in the flue gas at the SCR inlet, and a constraint function is added to the optimization problem to obtain an optimization model; Based on the optimization model and the long short-term memory network, the influence of various operating parameters of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet is analyzed to obtain a NOx concentration prediction model for the coal-fired unit; Based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model, the coal-fired unit is controlled in real time to ensure that the coal-fired unit maintains the lowest NOx concentration emission; wherein, The real-time optimization control of the coal-fired unit at least includes the opening size control of the green ammonia supply valve.

2. The method according to claim 1, characterized in that The reaction mechanism model of the corresponding ammonia and coal combustion reaction is constructed based on the design information of the coal-fired unit, including: Construct a green ammonia production model based on coal-fired unit design information; Based on the green ammonia production model, a reaction mechanism model of ammonia and coal combustion reaction in the furnace is constructed; wherein: The green ammonia production model comprises a synthetic ammonia device, a renewable energy power generation device and a water electrolysis hydrogen production device.

3. The method according to claim 1, characterized in that The reaction mechanism model of the ammonia and coal combustion reaction is: in, is the generation rate of NOx; is the NOx generation rate constant; is the mass flow rate of coal; is the mass fraction of nitrogen in coal; is the mass fraction of unburned nitrogen; is the reduction rate of NOx; is the process efficiency value; is the mass flow rate of green ammonia; is the concentration of NOx.

4. The method according to claim 1, characterized in that: The optimization problem is expressed as: in, is the proportion of NOx concentration in the flue gas at the SCR inlet; Constraint functions include: Any one or more of the following: energy balance constraint, green ammonia ratio constraint in total fuel blending, ammonia escape constraint at boiler outlet, oxygen concentration constraint in flue gas at boiler outlet, unit output constraint and non-negative constraint.

5. The method according to claim 1, characterized in that: Based on the optimization model and the long short-term memory network, the influence of various operating parameters of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet is analyzed to obtain the NOx concentration prediction model of the coal-fired unit, including: A long short-term memory network is constructed based on the optimization model, wherein: The long short-term memory network includes a memory unit, an input gate, a forget gate, an output gate, and a linear regression layer at the end; Based on the long short-term memory network, the influence of various parameters on the NOx concentration at the SCR inlet under different working conditions is simulated, and the parameters with an influence greater than the preset influence threshold are taken as key parameters; A NOx concentration prediction model for coal-fired units is constructed based on key parameters.

6. The method according to claim 1, characterized in that The NOx concentration prediction model of the coal-fired unit is constructed based on key parameters, including: Based on the selected key parameters, the initial framework of the NOx concentration prediction model is constructed to clarify the input and output of the model; Training the NOx concentration prediction model based on the existing historical data and the initial framework of the NOx concentration prediction model to obtain an initial NOx concentration prediction model; The prediction model is verified and tested based on an unused data set, and the verified model is used as the NOx concentration prediction model.

7. The method according to claim 1, characterized in that The real-time optimization control of the coal-fired unit is performed based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model to ensure that the coal-fired unit maintains the lowest NOx concentration emission, including: Collect the real-time operating parameters of the coal-fired unit, and use the real-time operating parameters as the input parameters of the NOx concentration prediction model, and output the NOx concentration prediction value at the SCR inlet based on the model; wherein, In the process of the model outputting the predicted value of NOx concentration at the SCR inlet, additional disturbance compensation is performed based on the extended state observer; Determining, based on the predicted value of NOx concentration at the SCR inlet, whether the predicted value is a minimum value of NOx concentration under the corresponding operating condition; If not, the current optimal control strategy is generated based on the generalized predictive control model; Based on the optimal control strategy, an adjustment plan for each execution component is generated as an optimal control plan.

8. The method according to claim 7, characterized in that The extended state observer is expressed as: in, and is the derivative of the extended state variable of the extended state observer; and is the detector gain parameter; is the observer bandwidth; is the state variable of the coal-fired unit; is the control gain coefficient; is the input parameter of the extended state observer; is the output parameter after additional disturbance compensation.

9. The method according to claim 7, characterized in that: The generalized predictive control model is: in, Indicates the control amount increment Make the objective function minimize; is the predicted value of NOx concentration; is the expected value of NOx concentration; T is a matrix; To control the weight of the increment.

10. A NOx concentration control system for the SCR inlet flue gas of a coal-fired unit burning green ammonia, which is applied to the prediction and control of the NOx concentration of the SCR inlet flue gas of a coal-fired unit burning green ammonia, characterized in that: The system comprises: An initial unit, used to construct a corresponding reaction mechanism model of ammonia and coal combustion reaction based on the design information of the coal-fired unit; A model building unit is used to build an optimization problem based on the reaction mechanism model with the goal of minimizing the proportion of NOx concentration in the flue gas at the SCR inlet, and to add a constraint function to the optimization problem to obtain an optimization model; A prediction unit, for analyzing the influence of various operating parameters of the coal-fired unit on the NOx concentration of the flue gas at the SCR inlet based on the optimization model and the long short-term memory network, and obtaining a NOx concentration prediction model for the coal-fired unit; The control unit is used to perform real-time optimization control of the coal-fired unit based on the real-time operating parameters of the coal-fired unit and the NOx concentration prediction model to ensure that the coal-fired unit maintains the lowest NOx concentration emission.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the method for controlling the NOx concentration in the flue gas at the SCR inlet of a coal-fired unit blended with green ammonia as described in any one of claims 1 to 9.

Images