Boiler combustion control method and control system based on co measurement

Abstract

The application discloses a boiler combustion control system based on CO measurement, comprising: constructing a full-furnace three-dimensional geometric model from a burner to an economizer outlet; performing batch numerical simulation calculation on a furnace combustion process; measuring CO concentration at the economizer outlet; establishing a CO concentration prediction model at the economizer outlet according to the measurement result; training and debugging the CO concentration prediction model according to the calculation result of the numerical simulation; using the CO prediction model to predict the CO concentration at the economizer outlet according to real-time data of furnace inlet parameters, and adjusting the air volume and the coal powder volume to adjust the furnace combustion condition according to the CO concentration. The application uses tunable spectral absorption technology to measure the CO concentration value, and can accurately obtain the CO concentration at the economizer outlet in real time; the air volume and the coal powder volume are guided by the CO concentration prediction model and the furnace combustion model.

Description

Technical Field

[0001] This invention relates to the technical field of boiler combustion adjustment, specifically to a boiler combustion control method and control system based on CO measurement. Background Technology

[0002] In the combustion of coal-fired power plant boilers, the CO concentration in the flue gas reflects the quality of the pulverized coal and air ratio inside the furnace. Uneven ratios of pulverized coal and air increase the heat loss from incomplete mechanical and chemical combustion during the combustion process. Therefore, the CO concentration in the flue gas is one of the main parameters indicating the incomplete combustion state of combustibles. Excessive CO concentration in the flue gas will reduce the boiler's combustion thermal efficiency and easily cause high-temperature corrosion of the water-cooled walls.

[0003] In addition, the decrease in CO concentration also indicates that the furnace air volume is too high at this time, the oxygen content is sufficient, and the total air volume is too high. The excessive total air volume will increase the flue gas volume, increase the boiler exhaust heat loss, and reduce the boiler combustion thermal efficiency.

[0004] Under certain external conditions (coal quality, burner swirl angle, etc.), the pulverized coal combustion state inside the furnace is determined by the total coal quantity, total air volume, and the air volume distributed to each burner.

[0005] Therefore, the CO concentration at the economizer outlet is necessarily related to the boiler load, air volume, and coal volume. By leveraging this relationship, we can use the CO concentration at the economizer outlet to guide the adjustment of furnace combustion, thereby optimizing the combustion conditions in the furnace. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a boiler combustion control method and control system based on CO measurement, which adjusts combustion by measuring the CO concentration at the economizer outlet and predicting the CO concentration in real time.

[0007] To solve the above-mentioned technical problems, the technical method adopted by the present invention is as follows: The present invention discloses a boiler combustion control method based on CO measurement, comprising the following steps.

[0008] S1. Measure the CO concentration at the economizer outlet and establish a prediction model for the CO concentration at the economizer outlet based on the measurement results;

[0009] S2 trains and debugs the CO concentration prediction model based on the calculation results of the numerical simulation of the CO concentration prediction model.

[0010] S3 uses a CO prediction model to predict the CO concentration at the economizer outlet based on real-time data of furnace inlet parameters, and then adjusts the furnace combustion status by regulating the air volume and pulverized coal volume.

[0011] Furthermore, the CO concentration prediction model is as follows:

[0012]

[0013] In the formula, the Gaussian kernel function α i α i * It is a Lagrange multiplier, b * It is a displacement term; This indicates that the Lagrange multiplier method was used.

[0014] Furthermore, establishing the CO concentration prediction model includes the following steps:

[0015] S21. A Gaussian kernel function was used for mapping to find a mapping that places points in a two-dimensional plane into a three-dimensional plane, i.e., k = (x n ,x),n=1,2,3,4,5,6;

[0016] In the formula, X1: coal feed rate of AF mill, X2: central air opening, X3: primary air opening, X4: internal secondary air opening, X5: external secondary air opening, and X6: burnout air opening;

[0017] S22. Apply the Grammar multiplier method to find the limit value under multiple constraints, i.e.

[0018] S23. Obtain the CO concentration prediction model:

[0019] Furthermore, step S1 also includes the following steps:

[0020] S101: Based on the boiler design data, an approximate processing method is used to transform the furnace into a three-dimensional spatial structure at a 1:1 scale;

[0021] S102: Establish a three-dimensional geometric model of the entire furnace based on the three-dimensional spatial structure; the three-dimensional geometric model mainly includes the cold ash hopper area, burner area, burnout air area, flame deflector area, horizontal flue area and vertical flue area;

[0022] The cold ash hopper area is mainly responsible for cooling and collecting ash and slag; the burner area contains 30 swirl burners, each consisting of a central air duct, a primary air duct, an inner secondary air duct, and an outer secondary air duct; the burnout air area contains 10 burnout air burners, each consisting of a central direct-flow air duct and an outer swirl air duct; in the horizontal and vertical flues, heat-receiving surfaces such as superheaters, reheaters, and economizers are constructed with thin walls of no thickness; geometric modeling and structured mesh generation are performed using the pre-processing ICEM software;

[0023] S103: Obtain the boundary conditions for combustion in the furnace; obtain the real-time operating parameters of the boiler according to the distributed control method of the power plant. The main operating parameters include: boiler load, pulverized coal quantity, total air volume, primary air temperature, secondary air temperature, burnout air temperature, main combustion zone temperature, economizer outlet flue gas temperature, economizer outlet oxygen content, etc.; set the emissivity and heat transfer coefficient of the wall surface of each zone according to the actual operating conditions of the boiler.

[0024] S104: Construct a furnace combustion model based on the three-dimensional spatial structure and the boundary conditions; the numerical calculation model of the whole furnace combustion includes basic governing equations, gas-phase turbulence model, gas-solid two-phase flow model, pulverized coal combustion model, and radiation heat transfer model;

[0025] S105: Determine a dynamic database of furnace combustion distribution under different operating conditions based on the three-dimensional geometric model and the furnace combustion model.

[0026] Furthermore, step S2 also includes the following steps:

[0027] S201: The CO concentration value is measured by tunable spectral absorption technology. A certain amount of flue gas is extracted from the economizer outlet, passes through the pretreatment filtration unit, and enters the laser concentration measurement module. The flue gas with the dust filtered out is passed into the CO gas absorption cell, and the CO concentration value is obtained by measuring the changes in the laser spectrum.

[0028] S202: Obtain CO concentration values ​​from multiple measuring points using S201, and establish a prediction model for CO concentration at the economizer outlet.

[0029] S203: A machine learning algorithm is used to obtain a predicted CO concentration value; the machine learning algorithm uses the numerical calculation results in S105 as the training set and the DCS data as the test set to train the CO concentration prediction model based on the training set to obtain a trained CO prediction model, including: pre-setting the learning rate, training rounds and number of samples of the CO concentration prediction model.

[0030] Furthermore, the independent variable of the economizer outlet CO concentration prediction model is the furnace combustion inlet parameter data, and the dependent variable is the furnace outlet CO concentration data; the regression function and the normal vector of the regression function are respectively:

[0031]

[0032]

[0033] In the formula, α i α i* is a Lagrange multiplier, and b is a displacement term.

[0034] Furthermore, step S3 also includes the following steps:

[0035] S301. Based on the data comparison method, the CO concentration prediction model is used to generate the CO prediction value over time.

[0036] S302. Using the CO prediction model, predict the CO concentration at the economizer outlet based on real-time data of furnace inlet parameters, and adjust the air volume and pulverized coal volume accordingly, and adjust the combustion status according to the combustion model.

[0037] The present invention also discloses a boiler combustion control system based on CO measurement, including a CO concentration calculation module 100, a CO concentration prediction model module 200, and a boiler combustion adjustment module 300; the boiler combustion control system based on CO measurement implements the steps of the above-mentioned boiler combustion control method based on CO measurement when executed.

[0038] Beneficial effects:

[0039] Compared to existing technologies, this invention uses a CO concentration prediction model and a furnace combustion model to guide the intake air volume and pulverized coal volume. Based on the real-time furnace inlet parameters under actual operating conditions, it can predict the changes in CO at the furnace outlet when combustion conditions change, i.e., when operating parameters such as air / pulverized coal are adjusted. This provides advance guidance on the internal combustion status of the furnace, reduces heat loss from incomplete combustion in the furnace, improves the thermal efficiency of coal-fired power plants, and is of great significance for improving the economic benefits of power plants. Attached Figure Description

[0040] Figure 1 This is a system structure diagram of boiler combustion control based on CO measurement in this invention.

[0041] Figure 2 This is a schematic diagram of the CO prediction model established by the support vector machine in this invention;

[0042] Figure 3 This is a user interface diagram of the full furnace physical model visualization system in this invention. Detailed Implementation

[0043] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0044] like Figure 1 As shown, this example is a boiler combustion control system based on CO measurement, including a CO concentration calculation module 100, a CO concentration prediction model module 200, and a boiler combustion adjustment module 300.

[0045] The CO concentration calculation module 100 consists of the following processes:

[0046] S101: Based on the boiler design data, an approximate processing method is used to transform the furnace into a three-dimensional spatial structure at a 1:1 scale;

[0047] S102: As Figure 3 As shown, a three-dimensional geometric model of the entire furnace is established based on the three-dimensional spatial structure. The model mainly includes the cold ash hopper area, burner area, burnout air area, flame deflector area, horizontal flue area, and vertical flue area.

[0048] The cold ash hopper area is mainly responsible for cooling and collecting ash and slag; the burner area contains 30 LNASB swirl burners, each burner consisting of a central air duct, a primary air duct, an inner secondary air duct, and an outer secondary air duct; the burnout air area contains 10 burnout air burners, each burner consisting of a central direct air duct and an outer swirl air duct; in the horizontal and vertical flues, heat-receiving surfaces such as superheaters, reheaters, and economizers are constructed with thin walls of no thickness; geometric modeling and structured mesh generation are performed using the pre-processing ICEM software.

[0049] S103: Obtain the boundary conditions for combustion in the furnace; Based on the power plant's Distributed Control System (DCS), obtain the real-time operating parameters of the boiler, including: boiler load, pulverized coal quantity, total air volume, primary air volume, secondary air volume, burnout air volume, main combustion zone temperature, economizer outlet flue gas temperature, economizer outlet oxygen content, etc.; Based on the actual operating conditions of the boiler, set the emissivity and heat transfer coefficient of the wall surface in each zone respectively.

[0050] S104: Construct a furnace combustion model based on the three-dimensional spatial structure and boundary conditions; the numerical calculation model of the whole furnace combustion includes the basic governing equations, gas phase turbulence model, gas-solid two-phase flow model, pulverized coal combustion model, and radiation heat transfer model.

[0051] The specific explanation is as follows:

[0052] The basic governing equations include conservation of mass, conservation of momentum, and conservation of energy.

[0053] The mass conservation equation is:

[0054]

[0055] In the formula, ρ is the density of the flue gas, kg / m³ 3 ; For velocity vectors, m / s; for time, t, s; for speed vectors, S. m It is the mass added to the continuous phase from the dispersed second phase (e.g., due to the evaporation of droplets) and user-defined mass sources.

[0056] The momentum conservation equation is:

[0057]

[0058] In the formula, p is the static pressure of the fluid, in Pa; It is the stress tensor, Pa; For and These are gravitational force and external force, respectively, in N.

[0059] The energy conservation equation is:

[0060]

[0061] In the formula, x j Let m be the displacement of the fluid in the j-direction; u be the displacement of the fluid in the j-direction. j ρ is the velocity of the fluid in the j-direction, m / s; h is the distance of the fluid from the boundary, m; T is the thermodynamic temperature, K; This represents the thermal conductivity term, S. h Φ and Φ represent the dissipation terms of heat source and fluid mechanical energy, respectively.

[0062] The gas-phase turbulence model adopts the Realizable k-ε model, which is more suitable for flows including strong streamline curvature, eddies and rotations, after the modification of the swirling flow.

[0063] The equations for k and ε in the Realizable k-ε model are as follows:

[0064]

[0065]

[0066] In the formula, p represents pressure, Pa; k represents turbulent pulsating kinetic energy, J; ε represents the dissipation rate of turbulent pulsating kinetic energy, %; μ and μ t σt represents the turbulent viscosity of the fluid under standard conditions and at temperature t, respectively, in kg / (m·s); k and σ ε These are Prandtl's constants for k and ε, respectively; G k and G b These are the turbulent kinetic energies generated by the laminar velocity gradient and buoyancy, respectively, m 2 / s 2 ;Y MThe value m represents the contribution of wave expansion in compressible turbulence to the total dissipation rate. 2 / s 2 S k and S ε These are user-defined source terms; C1, C2, C 1ε C 3ε It is a constant.

[0067] The gas-solid two-phase flow model employs a discrete phase model to couple the flow of pulverized coal and gas, treating the gas as a continuous phase and the pulverized coal particles as a discrete phase. The incident pulverized coal is simulated by establishing incident particles, and a discrete random walk model based on the Lagrange-Sun model is used to describe the motion of the incident particles.

[0068] This model simulates the interaction between particles and a series of discrete fluid phase turbulent eddies, where the particle eddy transit time is defined as:

[0069]

[0070] In the formula, τ is the particle relaxation time; L e It is the vortex length scale; |uu p | represents the magnitude of the relative velocity.

[0071] The pulverized coal combustion models include volatile matter analysis models, meteorological combustion models, and coke combustion models.

[0072] The volatile matter precipitation model uses a two-competing-rates model to simulate the volatile matter precipitation process. The volatile matter precipitation rate is:

[0073]

[0074]

[0075] In the formula, m p For particle mass, kg; m p,0 f represents the initial mass of the particles, in kg; v,0 The initial mass fraction of volatiles in the particles is denoted as ; k is the kinetic rate, s. -1 E is the activation energy, J / mol; A is the pre-exponential factor; R is the molar gas constant.

[0076] The meteorological combustion model uses a non-premixed combustion model, which simplifies thermochemical substances to a single parameter, namely the mixing fraction.

[0077] The coke combustion model selected is a diffusion-kinetic controlled reaction rate model. The surface reaction combustion rate of coke is... for:

[0078]

[0079] In the formula, T p and T ∞ These are the temperatures of the reactant surface and the surrounding medium, respectively, in K; d p m is the surface area where the chemical reaction occurs. 2 ;Y ox M represents the mass fraction of the oxidant in the local gas. w,ox denoted as the molar mass of the oxidant, in g / mol; R is the kinetic reaction constant considering the reaction and diffusion on the inner surface of the coke.

[0080] The radiation heat transfer model adopts the P1 model, which has a radiation flux q. r Represented as:

[0081]

[0082] In the formula, a is the absorption coefficient; σ is the absorption coefficient; s denoted as scattering coefficient; G is incident emissivity; C is the phase function coefficient for anisotropy.

[0083] Then, based on the three-dimensional geometric model and the furnace combustion model, a dynamic database of furnace combustion distribution under different operating conditions is determined 105.

[0084] CO concentration prediction model module 200 includes the following steps:

[0085] S201 uses tunable spectral absorption technology to measure CO concentration. A certain amount of flue gas is extracted from the economizer outlet, passes through the pretreatment filtration unit, and enters the laser concentration measurement module. The flue gas with the dust removed is passed into the CO gas absorption cell, and the CO concentration value is obtained by measuring the changes in the laser spectrum.

[0086] S202 obtains CO concentration values ​​from multiple measuring points from S201 and establishes a CO concentration prediction model 202 at the economizer outlet.

[0087] The independent variable of the economizer outlet CO concentration prediction model 202 is the furnace combustion inlet parameter data, and the dependent variable is the furnace outlet CO concentration data. The regression function and the normal vector of the regression function are as follows:

[0088]

[0089]

[0090] In the formula, α i α i* is a Lagrange multiplier, and b is a displacement term.

[0091] The Gaussian kernel function, which has strong nonlinear mapping capabilities, was selected as the kernel function for the support vector machine to adapt to the complex combustion environment inside the boiler.

[0092] A schematic diagram of the CO prediction model based on support vector machines is shown below. Figure 2 As shown.

[0093] S203 uses a machine learning algorithm to obtain predicted CO concentration values.

[0094] The algorithm is a machine learning algorithm. The numerical calculation results in the dynamic database in S105 are used as the training set, and the DCS data is used as the test set. The CO concentration prediction model is trained based on the training set to obtain a trained CO prediction model. This includes: setting the learning rate, training rounds and number of samples for the CO concentration prediction model in advance.

[0095] The boiler combustion adjustment module 300 includes the following steps:

[0096] S301 uses a data comparison method to generate the time-varying pattern of predicted CO values ​​through a CO concentration prediction model;

[0097] S302 uses a CO prediction model to predict the CO concentration at the economizer outlet based on real-time data of furnace inlet parameters, and adjusts the air volume and pulverized coal volume accordingly, adjusting the combustion status based on the combustion model.

[0098] The adjustment method is linked to the dynamic database in S105. The dynamic database records the relationship between total coal volume, total air volume, primary air volume, secondary air volume, burnout air volume, and CO concentration at the economizer outlet. Based on this relationship, combustion adjustment guidance can be provided. For example, at 50% load, the CO concentration at the economizer outlet decreases as the primary air volume increases. When the CO concentration is too high, the primary air volume can be increased.

[0099] The user interface of the aforementioned prediction system during its specific implementation is as follows: Figure 3 As shown.

[0100] The prediction system can predict the CO concentration at the furnace outlet in real time based on the furnace inlet parameters under actual operating conditions, providing guidance to users.

[0101] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method of boiler combustion control based on CO measurement, characterized by: Includes the following steps S1. Measure the CO concentration at the economizer outlet and establish a prediction model for the CO concentration at the economizer outlet based on the measurement results; S2. Based on the calculation results of the numerical simulation of the CO concentration prediction model, train and debug the CO concentration prediction model; S3. Using a CO prediction model, the CO concentration at the economizer outlet is predicted based on real-time data of furnace inlet parameters. Based on this, the combustion status of the furnace is adjusted by regulating the air volume and pulverized coal volume. Step S1 further includes the following steps: S101: Based on the boiler design data, an approximate processing method is used to transform the furnace into a three-dimensional spatial structure at a 1:1 scale; S102: Establish a three-dimensional geometric model of the entire furnace based on the three-dimensional spatial structure; the three-dimensional geometric model mainly includes the cold ash hopper area, burner area, burnout air area, flame deflector area, horizontal flue area and vertical flue area; The cold ash hopper area is mainly responsible for cooling and collecting ash and slag; the burner area contains 30 swirl burners, each consisting of a central air duct, a primary air duct, an inner secondary air duct, and an outer secondary air duct; the burnout air area contains 10 burnout air burners, each consisting of a central direct-flow air duct and an outer swirl air duct; in the horizontal and vertical flues, heat-receiving surfaces such as superheaters, reheaters, and economizers are constructed with thin walls of no thickness; geometric modeling and structured mesh generation are performed using the pre-processing ICEM software; S103: Obtain the boundary conditions for combustion in the furnace; obtain the real-time operating parameters of the boiler according to the distributed control method of the power plant. The main operating parameters include: boiler load, pulverized coal quantity, total air volume, primary air temperature, secondary air temperature, burnout air temperature, main combustion zone temperature, economizer outlet flue gas temperature, economizer outlet oxygen content, etc.; set the emissivity and heat transfer coefficient of the wall surface of each zone according to the actual operating conditions of the boiler. S104: Construct a furnace combustion model based on the three-dimensional spatial structure and the boundary conditions; the numerical calculation model of the whole furnace combustion includes basic governing equations, gas-phase turbulence model, gas-solid two-phase flow model, pulverized coal combustion model, and radiation heat transfer model; S105: Determine a dynamic database of the furnace combustion distribution under different operating conditions based on the three-dimensional geometric model and the furnace combustion model; Establishing the CO concentration prediction model includes the following steps: S21. A Gaussian kernel function was used for mapping to find a mapping that places points in a two-dimensional plane into a higher-dimensional space, i.e. n=1,2,3,4,5,6; In the formula, X1: coal feed rate of AF mill, X2: central air opening, X3: primary air opening, X4: internal secondary air opening, X5: external secondary air opening, and X6: burnout air opening; S22. Apply the Grammar multiplier method to find the limit value under multiple constraints, i.e. ; S23. Obtain the CO concentration prediction model: ; Step S2 further includes the following steps: S201: The CO concentration value is measured by tunable spectral absorption technology. A certain amount of flue gas is extracted from the economizer outlet, passes through the pretreatment filtration unit, and enters the laser concentration measurement module. The flue gas with the dust filtered out is passed into the CO gas absorption cell, and the CO concentration value is obtained by measuring the changes in the laser spectrum. S202: Obtain CO concentration values ​​from multiple measuring points using S201, and establish a prediction model for CO concentration at the economizer outlet; S203: A machine learning algorithm is used to obtain a predicted CO concentration value; the machine learning algorithm uses the numerical calculation results in S105 as the training set and the DCS data as the test set to train the CO concentration prediction model based on the training set to obtain a trained CO prediction model, including: pre-setting the learning rate, training rounds and number of samples of the CO concentration prediction model. Step S3 further includes the following steps: S301. Based on the data comparison method, the CO concentration prediction model is used to generate the CO prediction value over time. S302. Using the CO prediction model, predict the CO concentration at the economizer outlet based on real-time data of furnace inlet parameters, and adjust the air volume and pulverized coal volume accordingly, and adjust the combustion status according to the combustion model.

2. The boiler combustion control method based on CO measurement according to claim 1, characterized in that: The CO concentration prediction model is as follows: ; In the formula, the Gaussian kernel function ;α i α i * It is a Lagrange multiplier, b * It is a displacement term; This indicates that the Lagrange multiplier method was used.

3. The boiler combustion control method based on CO measurement according to claim 1, characterized in that: The independent variable of the economizer outlet CO concentration prediction model is the furnace combustion inlet parameter data, and the dependent variable is the furnace outlet CO concentration data; the regression function and the normal vector of the regression function are respectively: ； ； where α i , α i* are Lagrange multipliers and b is the displacement term.

4. A boiler combustion control system based on CO measurement, characterized in that: It includes a CO concentration calculation module (100), a CO concentration prediction model module (200), and a boiler combustion adjustment module (300); when the control system is executed, it implements the steps of the method as described in any one of claims 1-3.

Images