CFB boiler NOx emission optimization control method suitable for deep peak regulation

Abstract

The invention discloses a CFB boiler NOx emission optimization control method suitable for deep peak regulation, and belongs to the technical field of green emission reduction of thermal power generating units, and the method comprises the following steps: S1, constructing a key parameter model; s2, constructing a partition NOx emission model; s3, performing simulation verification and dynamic characteristic test; and S4, designing a control strategy. According to the CFB boiler NOx emission optimization control method suitable for deep peak regulation, the dynamic regulation and control precision of the combustion process is improved by establishing the regional reaction dynamic NOx model; the emission characteristics of nitric oxide in the wide-load operation and rapid load changing process are analyzed, and the defect of NOx emission control in the rapid load changing process is overcome; dynamic optimization control is realized by means of multi-step prediction, rolling optimization and online parameter identification, and rapid tracking of pollutant concentration and collaborative optimization of operation cost are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of green emission reduction of thermal power units, and specifically relates to a CFB boiler NO x Emission optimization control method. Background Art

[0002] Circulating fluidized bed (CFB) boilers, with their low-temperature combustion, strong fuel adaptability, and superior pollutant emission control, offer significant advantages in the coal-fired sector and are a key path to clean and efficient coal utilization. Their core approach is to achieve cyclical coal combustion through efficient gas-solid two-phase flow. This technology is particularly well-suited for the clean utilization of low-calorific-value, high-ash fuels, and can operate over a wide load range and rapidly change loads, providing technical support for pollutant emission control in supercritical CFB boilers.

[0003] However, the existing technology has the following shortcomings: the existing technology relies on the ideal working condition data of the laboratory to pass the static NO x Emission modeling and simulation cannot directly guide the real-time control optimization under complex load changes in power plants. In existing technologies, CFB units mainly improve the load change rate of the units by optimizing the coordinated control system, without considering the NO emissions during rapid load changes. x Emission control problems; and PID control has significant limitations when dealing with large lag and multivariable coupling systems.

[0004] Therefore, a new approach is urgently needed. Summary of the Invention

[0005] The purpose of the present invention is to provide a CFB boiler NO suitable for deep peak regulation x The emission optimization control method is to analyze the evolution mechanism of nitrogen component morphology by constructing a multi-region molecular model and establish a regional reaction dynamic NO x The model improves the dynamic control accuracy of the combustion process; analyzes the emission characteristics of nitrogen oxides during wide load operation and rapid load change, and converts NO x Emission control is taken into consideration to compensate for the NO x Insufficient emission control; dynamic optimization control is achieved with the help of multi-step prediction, rolling optimization and online parameter identification, achieving rapid tracking of pollutant concentrations and coordinated optimization of operating costs.

[0006] To achieve the above objectives, the present invention provides a CFB boiler NO suitable for deep peak regulation. x The emission optimization control method comprises the following steps:

[0007] S1. Establish oxygen volume fraction model and bed temperature model in dense phase and dilute phase areas of the furnace, and output key parameter model;

[0008] S2, based on the key parameter model in S1, analyze NO x Generation and removal process, output NO x Emission parameters;

[0009] S3: Select real-time operating data of the unit under different load conditions and input it into the models established in S1 and S2, comparing the predicted values ​​with the measured values. Furthermore, through open-loop step disturbance testing, analyze the pollutant generation response law and obtain dynamic mapping relationships and reliability verification results.

[0010] S4, based on the dynamic mapping relationship in S3, design an improved step-by-step generalized predictive control scheme to achieve NO x Emission optimization control.

[0011] Preferably, the key parameter model in S1 includes: dense phase oxygen volume fraction, dilute phase oxygen volume fraction, flue gas oxygen volume fraction, dense phase temperature, dilute phase temperature, carbon burnt amount, carbon burnt rate and oxygen molar concentration.

[0012] Preferably, the oxygen volume fractions in the dense phase region and the dilute phase region in S1 are calculated as follows:

[0013] The oxygen volume fraction in the dense phase is expressed as:

[0014]

[0015] Among them, F1 is the amount of oxygen input by the primary air, unit is m 3 / s;V m is the volume of the dense phase region, in m 3 ξ B k is the oxygen consumption correction factor of solid fuel; y1 is the correction coefficient of smoke emission in dense phase area; B is the remaining mass of fuel, unit is kg; is the oxygen volume fraction in the dense phase; is the volume fraction of primary air oxygen;

[0016] The oxygen volume fraction in the freeboard zone is expressed as:

[0017]

[0018] in, is the current oxygen volume fraction in the freeboard zone; V x is the volume of the dilute phase, unit is m 3 ; F2 is the secondary air volume flow rate, unit is m 3 / s; is the volume fraction of oxygen in the secondary air; ξ h is the oxygen consumption coefficient of volatile matter; η xhr is the combustion share of volatile matter in the dilute phase zone; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in the dilute phase; F is the total air flow; W c is the coal feeding rate, unit is kg / s; k y2 is the correction coefficient of flue gas emission in the dilute phase zone;

[0019] The oxygen volume fraction in the oxygen measurement point area below the superheater is expressed as:

[0020]

[0021] in, is the oxygen volume fraction; V c is the volume of the area, in m 3 ;k y3 is the correction coefficient of smoke emission in the area; η hr is the volatile matter combustion share of the area; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in this area.

[0022] Preferably, S1 further includes:

[0023] Considering the influence of boiler temperature, the formula for calculating the molar concentration of oxygen is:

[0024]

[0025] Among them, T0 is 273.15 Kelvin; is; γ is the molar volume of gas, unit is L / mol; T i is the temperature of the furnace interior area, in K.

[0026] Preferably, S1 also includes a dense phase temperature model, which is expressed as:

[0027]

[0028] Among them, c cl is the specific heat capacity of the bed material, in J / (kg·K); is the mass of the bed material, in kg; H B is the calorific value of burning carbon, which is 4.722MJ / kg; j is the mass fraction of volatile matter j, in MJ / kg; H j is the combustion calorific value of volatile component j; j is CH4, H2, CO, and tar, and their calorific values ​​are 50.016MJ / kg, 124.238MJ / kg, 10.077MJ / kg, and 37MJ / kg respectively; is the specific heat capacity of primary air, which is 777 J / (m3·K); is the primary air temperature, which is 534.15K;sl,m is the water-cooled wall heat transfer coefficient; S sl,m is the water-cooled wall area in the dense phase zone, in m 2 ;T z Main steam temperature, in K.

[0029] Preferably, S1 also includes a freeboard temperature model, which is expressed as:

[0030]

[0031] Among them, T x is the freeboard temperature, in K; c yq is the specific heat capacity of flue gas, in J / (m3·K); is the specific heat capacity of secondary air, which is 534.15K; is the secondary air temperature; S sl,x is the water-cooled wall area in the dilute phase zone, unit: m 2 .

[0032] Preferably, NO in S2 x Emission parameters include: dense phase NO x Emission concentration, dilute phase NO x Emission concentration, NO after denitrification outside the furnace x Emission molar concentration and reduced concentration.

[0033] Preferably, S2 further includes:

[0034] NO x The emission model is expressed as:

[0035]

[0036] in, is the NO self-generation rate, in kg / s; r HCN-NO is the reaction rate of HCN oxidation to generate NO, in kg / s; is the reaction rate of NH3 oxygen to NO, in kg / s; r c_NO is the reaction rate of carbon-nitrogen oxidation to generate NO, in kg / s; M NO is the molar mass of NO, in g / mol; is the oxygen molar concentration in the dense phase, in kmol / m 3 ;k HCN-NO is the reaction kinetic parameter constant for the oxidation of HCN to NO, in m 3 / (kmol·s); C HCN is the HCN molar concentration, in kmol / m 3 ; is the reaction kinetic parameter constant for the oxidation of NH3 to NO, in m 3 / (kmol·s); k c_NO is the kinetic parameter of the reaction of carbon-nitrogen oxidation to NO; w c_N is the carbon-nitrogen mass fraction of coke; R N is the fuel nitrogen conversion rate;

[0037] The mass fraction of nitrogen in coke and the dissociation efficiency of nitrogen in coal are expressed as:

[0038]

[0039]

[0040] Among them, μ N w is the relationship coefficient between the instantaneous carbon nitrogen and the basic nitrogen received by the fuel; N is the received basic nitrogen mass fraction; Represents the function of the mass fraction of each element in coal and the excess air coefficient; C NO,i is the NO molar concentration in kmol / m 3 ;

[0041] The molar concentration model of HCN in the furnace is expressed as:

[0042]

[0043] Among them, w h_N is the mass fraction of nitrogen in the volatile matter; η h_N(HCN) is the proportion of HCN in volatile nitrogen; M N is the molar mass of N in g / mol;

[0044] The molar concentration model of NH3 in the furnace is expressed as:

[0045]

[0046] in, is the ratio of NH3 in volatile nitrogen; is the catalytic oxidation rate of NH3 by CaO, in kg / s.

[0047] Preferably, the improved step-by-step generalized predictive control scheme in S4 is specifically as follows:

[0048] Using the CARIMA model, it is expressed as:

[0049]

[0050] Among them, A(z -1 ) is the output autoregressive polynomial; y(t) is the system output; B(z -1) is the input polynomial; Δu(td) is the input difference; C(z -1 ) is the noise polynomial; ξ(t) is white noise; Δ is the difference operator;

[0051] The improved Adam optimization algorithm is used, which is expressed as:

[0052]

[0053] Among them, θ t is the model parameter at the current time t; clip is the gradient clipping function; θ t-1 is the model parameter at the previous moment; η is the time-varying learning rate including step size attenuation and momentum compensation mechanism; β1 is the momentum attenuation factor, the value is 0.9; β2 is the adaptive factor, the value is 0.999; m t-1 is the first-order momentum; v t-1 is the second-order momentum; g t is the instantaneous gradient; ε is a numerical stability constant, fixed at 10 -8 ;

[0054] Construct a multi-step prediction model and optimize the performance index, expressed as:

[0055]

[0056] Among them, G j is the reference trajectory value of step j; b j is the basic target value of step j; n b is the number of steps in the linear growth segment; N is the length of the prediction time domain;

[0057] Construct a multi-step prediction model, expressed as:

[0058]

[0059]

[0060] in, is the output prediction value at time t for the future time t+j; n a is the order of the output autoregressive term; a i is the output autoregressive coefficient; b i is the input coefficient; m (·) is the predicted output of the model; y(·) is the actual measured output;

[0061]

[0062]

[0063] in, is the output prediction value at time t for the future time t+j; n a is the order of the output autoregressive term; a i is the output autoregressive coefficient; b i is the input coefficient; m (·) is the predicted output of the model; y(·) is the actual measured output;

[0064] The smooth reference trajectory is expressed as:

[0065] y r (t+j)=α j y(t)+(1-α j )y target ;

[0066]

[0067] Among them, y r is the reference trajectory value; α is the attenuation coefficient; y target The target output of the system;

[0068] The optimization performance index of the control system is expressed as:

[0069]

[0070] Among them, J is the objective function value; N is the prediction time domain length; is the output prediction value; y r is the reference trajectory value; λ is the penalty coefficient; N u is the length of the control time domain; Δu(t+j-1) is the rate of change of the control quantity;

[0071] The closed-form optimal solution is expressed as:

[0072] Δu=(G T G+λI) -1 G T (Y r -Y m );

[0073] Where G is the gradient matrix of the prediction model; I is the identity matrix;

[0074] Safety limit mechanism u(t)∈[0.7y target ,1.3y target ].

[0075] The present invention also provides a CFB boiler NO suitable for deep peak regulation x Emission optimization control system, including:

[0076] Key parameter model building module, used to establish oxygen volume fraction models in the dense phase and dilute phase zones, bed temperature models in the furnace, and output key parameter models;

[0077] The emission model building module is connected with the key parameter model building module, and analyzes NO based on the key parameter model. x Generation and removal process, output NO x Emission parameters;

[0078] The model verification and dynamic characteristics research module is connected to the emission model construction module and is used to select real-time operating data of different load conditions of the unit to input the established model and compare the predicted values ​​with the measured values; and through open-loop step disturbance testing, analyze the pollutant generation response law and obtain dynamic mapping relationships and reliability verification results;

[0079] The control strategy design module is connected with the model verification and dynamic characteristics research module, and designs an improved step-by-step generalized predictive control scheme based on the dynamic mapping relationship to achieve NO x Emission optimization control.

[0080] Therefore, the present invention adopts the above-mentioned CFB boiler NO suitable for deep peak regulation. x Compared with the prior art, the emission optimization control method of the present invention has the following beneficial effects:

[0081] (1) The present invention analyzes the evolution mechanism of nitrogen component morphology by constructing a multi-region molecular model and establishes a regional reaction dynamic NO x The model improves the dynamic control accuracy of the combustion process;

[0082] (2) The present invention analyzes the nitrogen oxide emission characteristics during wide load operation and rapid load change, and x Emission control is included in the analysis to compensate for the NO x inadequate emission controls;

[0083] (3) The present invention realizes dynamic optimization control through multi-step prediction, rolling optimization and online parameter identification, achieving rapid tracking of pollutant concentrations and coordinated optimization of operating costs.

[0084] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 The present invention is a CFB boiler NO adapted to deep peak regulation x Variable load condition data diagram of an embodiment of the emission optimization control method;

[0086] Figure 2The present invention is a CFB boiler NO adapted to deep peak regulation x A data graph showing prediction results of oxygen volume fraction under variable load conditions in an embodiment of an emission optimization control method;

[0087] Figure 3 The present invention is a CFB boiler NO adapted to deep peak regulation x Data graph of dense phase temperature prediction results under three variable load conditions of an embodiment of the emission optimization control method; Figure 3 (a), (b), and (c) correspond to the temperature prediction results of the dense phase region under working conditions 1, 2, and 3, respectively; Figure 3 (d), (e), and (f) correspond to the temperature error results of the dense phase region under working conditions 1, 2, and 3, respectively;

[0088] Figure 4 The present invention is a CFB boiler NO adapted to deep peak regulation x Data graph of the predicted results of the freeboard zone temperature under three variable load conditions of the embodiment of the emission optimization control method; Figure 4 (a), (b), and (c) correspond to the predicted results of the dilute phase temperature under working conditions 1, 2, and 3, respectively; Figure 4 (d), (e), and (f) correspond to the dilute phase temperature error results under working conditions 1, 2, and 3, respectively;

[0089] Figure 5 The present invention is a CFB boiler NO adapted to deep peak regulation x NO emission optimization control method embodiment under variable load conditions x Emission prediction results and error result data graph; Figure 5 (a), (b), and (c) correspond to NO under working conditions 1, 2, and 3, respectively. x Emissions forecast results; Figure 5 (d), (e), and (f) correspond to NO under working conditions 1, 2, and 3, respectively. x Error results from emission model predictions;

[0090] Figure 6 The present invention is a CFB boiler NO adapted to deep peak regulation x NOx emission optimization control method embodiment x Step response data diagram of the emission concentration dynamic model, Figure 6 (a) shows the step response curve after the coal feed rate increases by 5%; Figure 6 (b) shows the step response curve after the primary air volume increases by 5%; Figure 6 (c) shows the step response curve after the secondary air volume increases by 5%; Figure 6(d) shows the step response curve after the urea solution flow rate increases by 20%;

[0091] Figure 7 The present invention is a CFB boiler NO adapted to deep peak regulation x CFB boiler NO emission optimization control method embodiment x Emission optimization control effect data chart, Figure 7 (a), (b), and (c) correspond to the CFB boiler NO under working conditions 1, 2, and 3, respectively. x Emission optimization control effect data chart. DETAILED DESCRIPTION

[0092] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used in the present invention should be the common meanings understood by people with ordinary skills in the field to which the present invention belongs.

[0093] Example 1

[0094] A CFB boiler NO adapted to deep peak regulation of the present invention x The emission optimization control method comprises the following steps:

[0095] S1. Establish oxygen volume fraction model and bed temperature model in dense phase and dilute phase zones in the furnace;

[0096] In this step, the oxygen volume fractions in the dense phase zone, dilute phase zone, and the oxygen measurement point area below the superheater are calculated respectively, and the oxygen volume fraction models of the dense phase zone and dilute phase zone in the furnace are constructed. Specifically,

[0097] When modeling the oxygen volume fraction in different areas of the furnace, while the fuel burnout rate in air is clear, its combustion intensity is closely related to the oxygen volume fraction: combustion is more intense at higher oxygen content, and more gradual at lower oxygen content. Therefore, the fuel combustion intensity is expressed as the ratio of the volume fraction of oxygen in the dense phase to the volume fraction of oxygen in the primary air. In combination with the law of conservation of coal mass, the formula for calculating the remaining fuel mass after volatilization analysis is as follows:

[0098]

[0099] in, is the time rate of change of the remaining fuel mass, in kg / s; B is the remaining fuel mass, in kg; w his the mass fraction of volatile matter; W c is the coal feeding rate, unit is kg / s; is the oxygen volume fraction in the dense phase; t B is the burnout time of the remaining fuel, in seconds; is the volume fraction of primary air oxygen;

[0100] For the oxygen volume fraction in the dense phase, the oxygen input comes from the primary air, and the output needs to consider fuel consumption and flue gas emissions. Since the primary air volume is less than the actual flue gas volume discharged from the dense phase to the dilute phase, a correction coefficient is introduced, and the calculation formula is:

[0101]

[0102] Among them, F1 is the amount of oxygen input by the primary air, unit is m 3 / s;V m is the volume of the dense phase region, in m 3 ξ B k is the oxygen consumption correction factor of solid fuel; y1 is the correction coefficient of flue gas emission in dense phase area;

[0103] For the oxygen volume fraction in the dilute phase, the oxygen input comes from the secondary air, and the output needs to consider the total air supply and fuel volatile consumption. Because the total air volume is less than the flue gas emission, a correction factor is introduced, and the calculation formula is:

[0104]

[0105] in, is the current oxygen volume fraction in the freeboard zone; V x is the volume of the dilute phase, unit is m 3 ; F2 is the secondary air volume flow rate, unit is m 3 / s; is the volume fraction of oxygen in the secondary air; ξ h is the oxygen consumption coefficient of volatile matter; η xhr is the combustion share of volatile matter in the dilute phase zone; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in the dilute phase; F is the total air flow; k y2 is the correction coefficient of flue gas emission in the dilute phase zone;

[0106] For the oxygen volume fraction in the oxygen measurement area below the superheater, the oxygen content in this area is affected by factors such as primary and secondary air and coal feed rate. The calculation formula is:

[0107]

[0108] in, is the oxygen volume fraction; V c is the volume of the area, in m3 ;k y3 is the correction coefficient of smoke emission in the area; η hr is the volatile matter combustion share of the area; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in the region;

[0109] Based on the theory of immediate carbon, the coal feed is divided into four parts: combustion heat release, bed material accumulation, and fly ash and bottom ash loss. The storage of immediate carbon in the dense and dilute phases is described using a "ring-core" structure. The temperatures of the dense and dilute phases are calculated based on the combustion rate. A bed temperature model is constructed, specifically as follows:

[0110] During the combustion process of a CFB boiler, the fuel blown in by the primary air is divided into four parts: combustion heat, accumulation in the bed material, unburned fly ash loss, and unburned bottom ash loss. The calculation formula is:

[0111]

[0112] Among them, B c is the total mass of the burning carbon in the furnace; w is the mass fraction of the received carbon; r c,i is the combustion rate of the burning carbon; W Z is the ash discharge rate; w cZ is the carbon content in the ash; W F is the fly ash emission rate; w cF is the carbon content in fly ash;

[0113] Under the action of fluidizing air, the "ready-to-burn carbon" in the bed material flows in the furnace in a "ring-core" structure, filling the entire furnace space. The storage capacity in the dense phase area and the dilute phase area is calculated as follows:

[0114]

[0115] Among them, B c,i is the total mass of combustible carbon in region i, i = m is the dense phase region, i = x is the dilute phase region; is the height distribution function of the burning carbon in area i; L is the height coordinate of the furnace;

[0116] Combining the historical operation curve with mathematical statistical methods, the calculation formula for relevant parameters is:

[0117]

[0118] Among them, M Z is the ash emission; P THA is the heat receiving load of the steam turbine; w cZ is the mass fraction of carbon in the ash; Q ar,net To receive the base low heat; MF is the fly ash emission; w cF is the mass fraction of carbon in fly ash; T m is the temperature of dense phase region;

[0119] The combustion rate of the SCFB boiler is determined by the mass of carbon burned and the oxygen concentration, and the calculation formula is:

[0120]

[0121] Among them, k c,i is the combustion rate constant; is the molar concentration of oxygen; d c is the average diameter of the burning carbon particles; ρ c is the density of the burning carbon particles;

[0122] Considering the influence of boiler temperature, the formula for calculating the molar concentration of oxygen is:

[0123]

[0124] Among them, T0 is 273.15 Kelvin; is; γ is the molar volume of gas, unit is L / mol; T i is the temperature of the internal area of ​​the furnace, in K;

[0125] In CFB boilers, the firing rate directly affects the bed temperature and pollutant generation, and thus provides input to the bed temperature model.

[0126] The temperature of the dense phase zone affects the combustion rate, heat transfer coefficient and chemical reaction rate in the furnace. It needs to comprehensively consider the heat released by the burning carbon and coal volatiles, the heat transported by the primary air and the heat absorbed by the water-cooled wall. The calculation formula is:

[0127]

[0128] Among them, c cl is the specific heat capacity of the bed material, in J / (kg·K); is the mass of the bed material, in kg; H B is the calorific value of burning carbon, which is 4.722MJ / kg; j is the mass fraction of volatile matter j, in MJ / kg; H j is the combustion calorific value of volatile component j; j is CH4, H2, CO, and tar, and their calorific values ​​are 50.016MJ / kg, 124.238MJ / kg, 10.077MJ / kg, and 37MJ / kg respectively; is the specific heat capacity of primary air, which is 777 J / (m3·K); is the primary air temperature, which is 534.15K; sl,mis the water-cooled wall heat transfer coefficient; S sl,m is the water-cooled wall area in the dense phase zone, in m 2 ;T z is the main steam temperature, in K;

[0129] The main components of coal volatiles are CH4, H2, CO and tar. The fitting function of coal volatile calorific value is:

[0130]

[0131] in, is the mass fraction of methane; is the mass fraction of hydrogen; w CO is the mass fraction of carbon monoxide; w tar is the mass fraction of tar;

[0132] The temperature of the dilute phase zone needs to take into account the heat released by the burning carbon, the heat carried by the primary and secondary air, and the heat absorbed by the water-cooled wall. The calculation formula is:

[0133]

[0134] Among them, T x is the freeboard temperature, in K; c yq is the specific heat capacity of flue gas, in J / (m3·K); is the specific heat capacity of secondary air, which is 534.15K; is the secondary air temperature; S sl,x is the water-cooled wall area in the dilute phase zone, unit: m 2 ;

[0135] S2, based on the key parameter model of S1, build NO x Emission model; analysis of NO x The generation and removal process of NO was established by combining the key parameter model and using the mechanism modeling method. x Emission models, including calculation models for generation rate, molar concentration of related substances, removal rate, etc.;

[0136] In this step, the partitioned NO is constructed by combining the oxygen volume fraction and bed temperature model using a mechanism modeling method. x The emission model is expressed as:

[0137]

[0138] in, is the NO self-generation rate, in kg / s; r HCN-NO is the reaction rate of HCN oxidation to generate NO, in kg / s; is the reaction rate of NH3 oxygen to NO, in kg / s; r c_NO is the reaction rate of carbon-nitrogen oxidation to generate NO, in kg / s; M NO is the molar mass of NO, in g / mol; is the oxygen molar concentration in the dense phase, in kmol / m 3 ;k HCN-NO is the reaction kinetic parameter constant for the oxidation of HCN to NO, in m 3 / (kmol·s); C HCN is the HCN molar concentration, in kmol / m 3 ; is the reaction kinetic parameter constant for the oxidation of NH3 to NO, in m 3 / (kmol·s); k c_NO is the kinetic parameter of the reaction of carbon-nitrogen oxidation to NO; w c_N is the carbon-nitrogen mass fraction of coke; R N is the fuel nitrogen conversion rate;

[0139] The mass fraction of nitrogen in coke and the dissociation efficiency of nitrogen in coal are expressed as:

[0140]

[0141]

[0142] Among them, μ N w is the relationship coefficient between the instantaneous carbon nitrogen and the basic nitrogen received by the fuel; N is the received basic nitrogen mass fraction; Represents the function of the mass fraction of each element in coal and the excess air coefficient; C NO,i is the NO molar concentration in kmol / m 3 ;

[0143] During the volatilization process, the nitrogen in the coal decomposes, releasing part of the nitrogen in the volatile matter as HCN, which is then oxidized to NO. In addition, part of the HCN is discharged from the dense phase by the primary air. Therefore, the molar concentration model of HCN in the furnace is expressed as:

[0144]

[0145] Among them, w h_N is the mass fraction of nitrogen in the volatile matter; η h_N(HCN) is the proportion of HCN in volatile nitrogen; M N is the molar mass of N in g / mol;

[0146] The mass fraction of nitrogen in the volatile matter can be expressed as:

[0147]

[0148] Another portion of the nitrogen in the volatile matter is released as NH3, which is then oxidized to NO. Furthermore, under the catalytic action of CaO, NH3 is also oxidized to NO. Finally, a portion of NH3 is blown out of the furnace by the primary air. Therefore, the molar concentration model of NH3 in the furnace can be determined and expressed as:

[0149]

[0150] in, is the ratio of NH3 in volatile nitrogen; is the catalytic oxidation rate of NH3 by CaO, in kg / s;

[0151] CaO has significant catalytic activity for various nitrogen-containing reactions and is the main catalyst for the NO reduction caused by limestone desulfurization in CFB boilers. x The key factor in the increase of emissions. The oxidation rate of NH3 catalyzed by CaO is expressed as:

[0152]

[0153] in, represents the CaO-catalyzed NH3 oxidation rate coefficient; is the molar mass of NH3, in g / mol; s CaO is the specific surface area of ​​the effective reacting limestone particles, in m 2 ; and θ NO,m Represents O2, NH3 and NO occupying the active sites on the CaO surface; k NO,CaO represents the CaO-catalyzed NO oxidation rate coefficient;

[0154] and θ NO,m The calculation method is as follows:

[0155]

[0156] in, K ad(NO) are the adsorption rate constants of O2, NH3 and NO, respectively, in m 3 / kmol; C NO,m is the NO molar concentration in the dense phase, in kmol / m 3 ;

[0157] The removal of NO is divided into CO reduction of NO and coke adsorption reduction. The reaction equation is expressed as:

[0158]

[0159] in, represents the removal rate of NO, in kg / s; r NO-CO is the reaction rate of NO reduction by CO, in kg / s; r NO-c,i is the reaction rate of NO reduction by coke, in kg / s; k NO-CO is the rate constant of NO reduction by CO, in m 3 / (kmol·s); C CO is the molar concentration of CO in kmol / m 3 ;k NO-c,i is the rate constant of NO reduction by coke, in kg -1 ·s -1 ;

[0160] k NO-CO and k NO-c,i Gray catalytic activity ζ hui The calculation method is:

[0161]

[0162] Among them, hui Indicates gray catalytic activity; χ CaO / MgO , are the mass fractions of SiO2, Al2O3, CaO / MgO, and Fe2O3 respectively;

[0163] In the complex reaction process within the CFB furnace, it is assumed that CO and CO2 are generated simultaneously. In addition, CaO also catalyzes the CO oxidation reaction. The amount of CO retained in the furnace, after deducting the CO discharged from the total gas flow, is expressed as:

[0164]

[0165] Among them, M CO is the molar mass of CO, in g / mol; is the CO oxidation rate constant; φ is the mechanical factor of the chemical reaction, which is strongly related to the bed temperature and the diameter of the burning carbon particles and is calculated as follows;

[0166]

[0167] The reaction rate of CO oxidation catalyzed by CaO can be expressed as:

[0168]

[0169] Among them, r CO,CaO is the catalytic oxidation rate of CO by CaO, in kg / s; K ad(CO)represents the adsorption rate constant of CO2, in m 3 / kmol;

[0170] The lumped parameter equation for the average concentration of in-situ NO emissions in the furnace is expressed as:

[0171]

[0172] in, is the original emission concentration of NO, in kmol / m 3 ;

[0173] The molar concentration of NO in the dense phase and dilute phase is expressed as:

[0174]

[0175] C NO,m NO x Dense phase emission concentration, in kmol / m 3 ; NO x Emission concentration in the dilute phase zone, in kmol / m 3 ;

[0176] For denitrification outside the furnace, the denitrification reaction rate r sncr Expressed as:

[0177]

[0178] Among them, r sncr Indicates the denitrification reaction rate in the SNCR reactor, in kg / s; k sncr is the kinetic equation parameter of the denitrification reaction in the SNCR reactor; C NO,rsncr is the NO molar concentration in the SNCR reactor, in kmol / m 3 ; is the molar concentration of NH3 in the SNCR reactor, in kmol / m 3 ;T sncr is the temperature inside the SNCR reactor, in K;

[0179] When the denitrification reducing agent is injected into the reactor, ammonia molecules are generated immediately, assuming that the reaction time is negligible. By subtracting the NH3 consumed in the denitrification process and the NH3 discharged from the total airflow of the SNCR reactor, the NH3 balance equation model in the SNCR reactor can be established based on the conservation of mass, which is expressed as:

[0180]

[0181] Among them, V sncr is the volume of the SNCR reactor, in m 3;k ns Q is the coefficient of urea solution generating NH3; ns is the urea volume flow rate, in m 3 / h;ρ ns is the mass density of 20% urea solution, in kg / m 3 η ns is the ratio of urea precipitated from urea solution; is the molar mass of (NH2)2CO in g / mol;

[0182] The lumped parameter equation for the average molar concentration of NO emissions is expressed as:

[0183]

[0184] Among them, C NO Represents the molar concentration of NO emissions, in kmol / m 3 ;

[0185] The converted NO emission concentration prediction value is expressed as:

[0186]

[0187] in, NO x Converted emission concentration, unit is kg / m 3 ;

[0188] S3. Select real-time operating data from different load conditions of the unit and input them into the model for comparative verification. Combined with system identification methods, open-loop step disturbance tests are conducted on key operating variables to analyze the pollutant generation response patterns and construct dynamic mapping relationships.

[0189] like Figure 1 As shown in the figure, the three-stage historical operation data of a 350MW CFB boiler distributed control system (DCS) was used for verification. The specific operating load conditions are:

[0190] Condition 1: The load regulation range is reduced from 270MW to 115MW, with a total load reduction of 45%. The average load reduction rate between 7305s and 12810s is 1.7MW / min.

[0191] Condition 2: The load regulation range increases from 140MW to 300MW, accounting for 46% of the total load. The average load growth rate between 16500s and 19320s is 1.2MW / min;

[0192] Condition 3: The load regulation range decreases from 270MW to 120MW, accounting for 43% of the total load. The average load growth rate between 11115s and 12930s is 2.8MW / min.

[0193] The prediction results of the oxygen volume fraction model under different load conditions are as follows: Figure 2 As shown in (a), (b) and (c) in the figure, the corresponding error results are as follows Figure 2 As shown in (d), (e) and (f).

[0194] The error calculation results of the oxygen volume fraction model under different load conditions, including the mean absolute percentage error (MAPE), maximum absolute error (MAE), root mean square error (RMSE), and mean square error (MSE), are shown in Table 1:

[0195] Table 1 Error results of oxygen volume fraction model under different load conditions

[0196]

[0197] Figure 3 (a), (b), (c) and Figure 4 (a), (b), and (c) in the figure respectively give the test results of the temperature of the dense phase region and the dilute phase region under variable load conditions. Figure 3 (d), (e), (f) and Figure 4 (d), (e), and (f) in the figure give the corresponding error results respectively;

[0198] The error calculation results of the temperature prediction model of the two regions under different load conditions are shown in Tables 2 and 3:

[0199] Table 2 Error results of dense phase temperature prediction model under different load conditions

[0200]

[0201] Table 3 Error results of the dilute phase temperature prediction model under different load conditions

[0202]

[0203] The MAPE of the two zone temperatures under all load conditions is less than 1.87%.

[0204] NO x The emission concentration model output results and the measured NO under different working conditions x Emission concentration comparison, such as Figure 5 As shown in (a), (b), and (c), the error results are as follows Figure 5 As shown in (d), (e), and (f); NO under different load conditionsx The error results of the emission prediction model are shown in Table 4:

[0205] Table 4 NO under different load conditions x Error results table of emission prediction model

[0206]

[0207]

[0208] The results show that the model established in this embodiment is suitable for different load conditions of SCFB units, and the prediction time is about 150 seconds;

[0209] like Figure 6 As shown, in order to capture the dynamic characteristics of the prediction model, open-loop step experiments are carried out on each input under 120MW low load and steady-state load conditions;

[0210] Figure 6 (a) shows the step response curve after the coal feed rate increases by 5%. When the air supply rate remains unchanged, the air-coal ratio decreases, increasing the reducing atmosphere and the amount of coke, thereby reducing NO x emissions. Ultimately, NO x The concentration is lower than the initial value.

[0211] Figure 6 (b) and (c) show the step response curves after the primary and secondary air volumes increase by 5%. The increase in oxygen concentration strengthens the oxidizing atmosphere and weakens the NO x The self-reduction effect increases the NO x Emissions of NO x The response to changes in air supply volume is faster than the response to changes in coal feed volume.

[0212] Figure 6 (d) shows the step response curve after the urea solution flow rate increases by 20%. The higher NH3 concentration in the SNCR system increases the ammonia nitrogen molar ratio, enhances the denitrification effect, and reduces NO x Emissions. NO x The response to changes in urea solution is faster than the response to changes in coal and air supply. These analyses confirm that the dynamic model accurately reflects the dynamic characteristics of pollutant emissions;

[0213] S4: Based on the dynamic mapping relationship in S3, an improved stepped GPC scheme is designed. Dynamic optimization control is achieved through five core improvements, including Adam optimizer parameter identification. This includes modules such as controller parameter initialization, online parameter identification, dynamic predictive control matrix generation, and related simulation verification.

[0214] In this step, the CARIMA model is used to initialize the controller parameters through the function. Among them: the model order is set to: n a =3 (output memory item), n b =3(input memory item, corresponding to B(z -1 ) polynomial order), n c =2 (noise model order), system delay d = 1. Pollutant concentration setting value is dynamically configured according to the control object. Sampling period T s It is set to 90 seconds to match the simulation environment and is expressed as:

[0215]

[0216] Among them, A(z -1 ) is the output autoregressive polynomial; y(t) is the system output; B(z -1 ) is the input polynomial; Δu(td) is the input difference; C(z -1 ) is the noise polynomial; ξ(t) is white noise; Δ is the difference operator;

[0217] The parameter online identification module uses the improved Adam optimization algorithm to replace the traditional recursive least squares method to improve the robustness of parameter estimation and solve the problem that the traditional GPC fixed matrix has poor adaptability to time-varying systems. Its comprehensive momentum gradient formula is expressed as;

[0218]

[0219] Among them, θ t is the model parameter at the current time t; clip is the gradient clipping function; θ t-1 is the model parameter at the previous moment; η is the time-varying learning rate including step size attenuation and momentum compensation mechanism; β1 is the momentum attenuation factor, the value is 0.9; β2 is the adaptive factor, the value is 0.999; m t-1 is the first-order momentum; v t-1 is the second-order momentum; g t is the instantaneous gradient; ε is a numerical stability constant, fixed at 10 -8 ;

[0220] The adaptive learning rate adjustment formula is expressed as:

[0221]

[0222] According to the identified B(z -1 ) polynomial coefficients to generate the dynamic predictive control matrix G, and process the unmodeled dynamic characteristics through gain adjustment and attenuation factors. Let the identification parameter vector be b=[b1,b2,...,b nb ] T , then the dynamic control matrix G∈RN ×1 Expressed as:

[0223]

[0224] Among them, G j is the reference trajectory value of step j; b j is the basic target value of step j; n b is the number of steps in the linear growth segment; N is the length of the prediction time domain;

[0225] A multi-step prediction model is constructed to calculate the predicted value for the next 10 steps through rolling calculation of historical data. The prediction recursive equation and historical data splicing rules are expressed as follows:

[0226]

[0227]

[0228] in, is the output prediction value at time t for the future time t+j; n a is the order of the output autoregressive term; a i is the output autoregressive coefficient; b i is the input coefficient; m (·) is the predicted output of the model; y(·) is the actual measured output;

[0229] The smooth reference trajectory is expressed as:

[0230] y r (t+j)=α j y(t)+(1-α j )y target ;

[0231]

[0232] Among them, y r is the reference trajectory value; α is the attenuation coefficient; y target The target output of the system;

[0233] Combined with the actual operation process of the unit, the optimization performance index of the control system is expressed as:

[0234]

[0235] Among them, J is the objective function value; N is the prediction time domain length; is the output prediction value; y r is the reference trajectory value; λ is the penalty coefficient; N u is the length of the control time domain; Δu(t+j-1) is the rate of change of the control quantity;

[0236] The closed-form optimal solution is expressed as:

[0237] Δu=(G T G+λI) -1 G T (Y r -Y m );

[0238] Where G is the gradient matrix of the prediction model; I is the identity matrix;

[0239] In addition, a safety limit mechanism u(t)∈[0.7y target ,1.3y target ].

[0240] In order to verify the NO x The effectiveness of emission optimization control is verified by using the proposed prediction model to simulate the SNCR denitrification process of the unit. In the simulation process, a long period of operation data is selected, and the data such as coal feed rate and air supply volume are used as the input of the SGPC controller. x The concentration setting values ​​are 20, 23, and 20 mg / m3 respectively under three working conditions. The optimization control simulation results are as follows: Figure 7 shown.

[0241] Without considering the on-site disturbance, SGPC has a significant impact on NO x The regulation effect of emission concentration is better and the fluctuation is smaller. The average values ​​after control are lower than those before control. The specific data are shown in Table 5. This is because the SGPC algorithm optimizes the control variables based on the predicted values ​​of nitrogen oxide emissions in the future, advances the adjustment process of the injection amount of ammonia, and avoids NO x Large fluctuations in emission concentrations.

[0242] Table 5 NO x Emission Control Performance Comparison Table

[0243]

[0244] Therefore, the present invention adopts the above-mentioned CFB boiler NO x The emission optimization control method is to analyze the evolution mechanism of nitrogen component morphology by constructing a multi-region molecular model and establish a regional reaction dynamic NO x The model improves the dynamic control accuracy of the combustion process; analyzes the emission characteristics of nitrogen oxides during wide load operation and rapid load change, and converts NO x Emission control is taken into consideration to compensate for the NO xInsufficient emission control; dynamic optimization control is achieved with the help of multi-step prediction, rolling optimization and online parameter identification, achieving rapid tracking of pollutant concentrations and coordinated optimization of operating costs.

[0245] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A CFB boiler suitable for deep peak regulation x The emission optimization control method is characterized in that: The following steps are involved: S1. Establish oxygen volume fraction model and bed temperature model in dense phase and dilute phase areas of the furnace, and output key parameter model; S2, based on the key parameter model in S1, analyze NO x Generation and removal process, output NO x Emission parameters; S3: Select real-time operating data of the unit under different load conditions and input it into the models established in S1 and S2, comparing the predicted values ​​with the measured values. Furthermore, through open-loop step disturbance testing, analyze the pollutant generation response law and obtain dynamic mapping relationships and reliability verification results. S4, based on the dynamic mapping relationship in S3, design an improved step-by-step generalized predictive control scheme to achieve NO x Emission optimization control.

2. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: The key parameter model in S1 includes: oxygen volume fraction in dense phase, oxygen volume fraction in dilute phase, oxygen volume fraction in flue gas, dense phase temperature, dilute phase temperature, carbon burnt, carbon burn rate and oxygen molar concentration.

3. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: The oxygen volume fraction calculation of the dense phase and dilute phase in S1 is as follows: The oxygen volume fraction in the dense phase is expressed as: Among them, F1 is the amount of oxygen input by the primary air, unit is m 3 / s;V m is the volume of the dense phase region, in m 3 ξ B k is the oxygen consumption correction factor of solid fuel; y1 is the correction coefficient of smoke emission in dense phase area; B is the remaining mass of fuel, unit is kg; is the oxygen volume fraction in the dense phase; is the volume fraction of primary air oxygen; The oxygen volume fraction in the freeboard zone is expressed as: in, is the current oxygen volume fraction in the freeboard zone; V x is the volume of the dilute phase, unit is m 3 ; F2 is the secondary air volume flow rate, unit is m 3 / s; is the volume fraction of oxygen in the secondary air; ξ h is the oxygen consumption coefficient of volatile matter; η xhr is the combustion share of volatile matter in the dilute phase zone; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in the dilute phase; F is the total air flow; W c is the coal feeding rate, unit is kg / s; k y2 is the correction coefficient of flue gas emission in the dilute phase zone; The oxygen volume fraction in the oxygen measurement point area below the superheater is expressed as: in, is the oxygen volume fraction; V c is the volume of the area, in m 3 ;k y3 is the correction coefficient of smoke emission in the area; η hr is the volatile matter combustion share of the area; is the theoretical oxygen volume fraction of volatile combustion; is the actual oxygen volume fraction in this area.

4. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: S1 also includes: Considering the influence of boiler temperature, the formula for calculating the molar concentration of oxygen is: Among them, T0 is 273.15 Kelvin; is; γ is the molar volume of gas, unit is L / mol; T i is the temperature of the furnace interior area, in K.

5. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: S1 also includes the dense phase temperature model, which is expressed as: Among them, c cl is the specific heat capacity of the bed material, in J / (kg·K); is the mass of the bed material, in kg; H B is the calorific value of burning carbon, which is 4.722MJ / kg; j is the mass fraction of volatile matter j, in MJ / kg; H j is the combustion calorific value of volatile component j; j is CH4, H2, CO, and tar, and their calorific values ​​are 50.016MJ / kg, 124.238MJ / kg, 10.077MJ / kg, and 37MJ / kg respectively; is the specific heat capacity of primary air, which is 777 J / (m3·K); is the primary air temperature, which is 534.15K; sl,m is the water-cooled wall heat transfer coefficient; S sl,m is the water-cooled wall area in the dense phase zone, in m 2 ;T z Main steam temperature, in K.

6. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: S1 also includes the freeboard temperature model, which is expressed as: Among them, T x is the freeboard temperature, in K; c yq is the specific heat capacity of flue gas, in J / (m3·K); is the specific heat capacity of secondary air, which is 534.15K; is the secondary air temperature; S sl,x is the water-cooled wall area in the dilute phase zone, unit: m 2 .

7. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: S2 NO x Emission parameters include: dense phase NO x Emission concentration, dilute phase NO x Emission concentration, NO after denitrification outside the furnace x Emission molar concentration and reduced concentration.

8. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: S2 also includes: NO x The emission model is expressed as: in, is the NO self-generation rate, in kg / s; r HCN-NO is the reaction rate of HCN oxidation to generate NO, in kg / s; is the reaction rate of NH3 oxygen to NO, in kg / s; r c_NO is the reaction rate of carbon-nitrogen oxidation to generate NO, in kg / s; M NO is the molar mass of NO, in g / mol; is the oxygen molar concentration in the dense phase, in kmol / m 3 ;k HCN-NO is the reaction kinetic parameter constant for the oxidation of HCN to NO, in m 3 / (kmol·s); C HCN is the HCN molar concentration, in kmol / m 3 ; is the reaction kinetic parameter constant for the oxidation of NH3 to NO, in m 3 / (kmol·s); k c_NO is the kinetic parameter of the reaction of carbon-nitrogen oxidation to NO; w c_N is the carbon-nitrogen mass fraction of coke; R N is the fuel nitrogen conversion rate; The mass fraction of nitrogen in coke and the dissociation efficiency of nitrogen in coal are expressed as: Among them, μ N w is the relationship coefficient between the instantaneous carbon nitrogen and the basic nitrogen received by the fuel; N is the received basic nitrogen mass fraction; Represents the function of the mass fraction of each element in coal and the excess air coefficient; C NO,i is the NO molar concentration in kmol / m 3 ; The molar concentration model of HCN in the furnace is expressed as: Among them, w h_N is the mass fraction of nitrogen in the volatile matter; η h_N(HCN) is the proportion of HCN in volatile nitrogen; M N is the molar mass of N in g / mol; The molar concentration model of NH3 in the furnace is expressed as: in, is the ratio of NH3 in volatile nitrogen; is the catalytic oxidation rate of NH3 by CaO, in kg / s.

9. A CFB boiler NO adapted for deep peak regulation according to claim 1 x The emission optimization control method is characterized in that: The improved step-by-step generalized predictive control scheme in S4 is as follows: Using the CARIMA model, it is expressed as: Among them, A(z -1 ) is the output autoregressive polynomial; y(t) is the system output; B(z -1 ) is the input polynomial; Δu(td) is the input difference; C(z -1 ) is the noise polynomial; ξ(t) is white noise; Δ is the difference operator; The improved Adam optimization algorithm is used, which is expressed as: Among them, θ t is the model parameter at the current time t; clip is the gradient clipping function; θ t-1 is the model parameter at the previous moment; η is the time-varying learning rate including step size attenuation and momentum compensation mechanism; β1 is the momentum attenuation factor, the value is 0.9; β2 is the adaptive factor, the value is 0.999; m t-1 is the first-order momentum; v t-1 is the second-order momentum; g t is the instantaneous gradient; ε is a numerical stability constant, fixed at 10 -8 ; Construct a multi-step prediction model and optimize the performance index, expressed as: Among them, G j is the reference trajectory value of step j; b j is the basic target value of step j; n b is the number of steps in the linear growth segment; N is the length of the prediction time domain; Construct a multi-step prediction model, expressed as: in, is the output prediction value at time t for the future time t+j; n a is the order of the output autoregressive term; a i is the output autoregressive coefficient; b i is the input coefficient; m (·) is the predicted output of the model; y(·) is the actual measured output; in, is the output prediction value at time t for the future time t+j; n a is the order of the output autoregressive term; a i is the output autoregressive coefficient; b i is the input coefficient; m (·) is the predicted output of the model; y(·) is the actual measured output; The smooth reference trajectory is expressed as: y r (t+j)=α j y(t)+(1-α j )y target ; Among them, y r is the reference trajectory value; α is the attenuation coefficient; y target The target output of the system; The optimization performance index of the control system is expressed as: Among them, J is the objective function value; N is the prediction time domain length; is the output prediction value; y r is the reference trajectory value; λ is the penalty coefficient; N u is the length of the control time domain; Δu(t+j-1) is the rate of change of the control quantity; The closed-form optimal solution is expressed as: Δu=(G T G+λI) -1 G T (Y r -Y m ); Where G is the gradient matrix of the prediction model; I is the identity matrix; Safety limit mechanism u(t)∈[0.7y target ,1.3y target ].

10. A CFB boiler suitable for deep peak regulation x Emission optimization control system, applied to a CFB boiler NO adapted for deep peak regulation as described in any one of claims 1-9 x The emission optimization control method is characterized in that: include: Key parameter model building module, used to establish oxygen volume fraction models in the dense phase and dilute phase zones, bed temperature models in the furnace, and output key parameter models; The emission model building module is connected with the key parameter model building module, and analyzes NO based on the key parameter model. x Generation and removal process, output NO x Emission parameters; The model verification and dynamic characteristics research module is connected to the emission model construction module and is used to select the real-time operating data of different load conditions of the unit to input the established model and compare the predicted values ​​with the measured values; Through open-loop step disturbance testing, the pollutant generation response law is analyzed to obtain dynamic mapping relationships and reliability verification results; The control strategy design module is connected with the model verification and dynamic characteristics research module, and designs an improved step-by-step generalized predictive control scheme based on the dynamic mapping relationship to achieve NO x Emission optimization control.

Images