Method and system for monitoring combustion instability of coal-fired boiler based on multi-parameter fusion

Abstract

The invention discloses a coal-fired boiler combustion instability monitoring method and system based on multi-parameter fusion, and the method comprises the steps: constructing a hearth flame imaging system which comprises an image detector, a lens, a stainless steel cavity and a cooling air channel; the flame image intensity is converted into radiation intensity through blackbody furnace calibration, and the flame temperature and emissivity are calculated based on the Planck law; extracting flame characteristic parameters including average radiation intensity, temperature, emissivity and standard deviation thereof; constructing a load-flame characteristic parameter model, and predicting parameter reference values under different loads through multiple linear regression; and calculating a combustion instability index CII, carrying out weighted fusion on the normalized deviation of the real-time measured value and the predicted value, and judging that CII is equal to 1 when the flame area is 0 or the average temperature is lower than 900 DEG C. The flame radiation image and the unit operation parameters are fused, and the limitation of traditional single-parameter monitoring is solved; combustion instability dynamic quantitative evaluation is achieved through the CII index, and the method is suitable for early warning of the deep peak regulation working condition.

Description

Technical Field

[0001] The present invention relates to the technical field of coal-fired boiler combustion monitoring technology, and in particular to a coal-fired boiler combustion instability monitoring method and system based on multi-parameter fusion. Background Art

[0002] Renewable energy generation is currently developing rapidly, with wind and photovoltaic power grid-connected capacity growing rapidly. Meanwhile, while the share of thermal power is gradually declining, supporting the development of new energy, it will still maintain a dominant position and provide peak-shaving support for new energy consumption, thereby compensating for technical issues such as the volatility, intermittency, and weak resistance to disturbances of renewable energy generation. Driven by external factors such as the absorption of clean energy, the participation of thermal power units in deep peak-shaving operations will become a regular practice.

[0003] Under deep peak-shaving operation conditions, the low-load stable combustion problem of the boiler is a key issue that restricts the depth and speed of peak-shaving. If there is a quantitative measurement device for the stability of boiler combustion, it will be possible to quantitatively evaluate the safety of peak-shaving operation to a certain extent, and optimize and adjust potential combustion problems to ensure stable and safe combustion. In addition, the stability of flame combustion is crucial to ensuring the safe, stable and efficient operation of industrial combustion equipment such as boilers, furnaces and engines. Flame instability may lead to reduced combustion efficiency, increased pollutant emissions, and even fire or explosion accidents. In industrial combustion systems, increasing flame stability can be achieved by optimizing burner configuration, optimizing the mixing method of fuel and oxidant, adjusting the flow rate of fuel and oxidant, and implementing advanced control strategies.

[0004] Digital imaging technology is considered one of the most effective methods for measuring flame stability in power plant boiler furnaces, thanks to its system functionality, portability, low cost, and information-rich nature. Wojcik et al. collected pulverized coal flame radiation signals at the KOZIENICE power plant and detected flame stability based on continuous offline wavelet transform. Stable and unstable flame signals had similar high-frequency components in their spectral distribution, but unstable signals contained significant low-frequency components. Smart et al. investigated the stability characteristics of mixed-fuel flames by measuring the uniformity of the two-dimensional brightness distribution of natural gas flames and coal-biomass mixed-combustion flames on a 0.5MW swirl combustion test device. In addition, Matthes et al. measured the stability of a 1MW swirl coal-fired flame, using a modified Otsu threshold algorithm to segment the flame edge to obtain the flame region, and measured flame stability based on the degree of flame shape change. Cheng et al. proposed a method for measuring the flame stability of two burners based on image processing and spectral analysis. They measured the stability of ideal flames, methane flames, and biomass flames on a laboratory combustion device to verify the effectiveness of the method. They also developed a flame detection system and conducted applied research on coal powder flame stability measurement in full-scale furnaces of multiple thermal power plants.

[0005] Among the existing products, the UvisorFAU810 flame detector developed by ABB is a mature and representative product in the world. This product analyzes the flame based on the full radiation spectrum from ultraviolet to infrared, and obtains the flame quality index according to the three parameters of flame intensity, flicker frequency and AC amplitude. The quality index ranges from 0-100%, where 100% represents the optimal flame state. If the quality index drops slightly, it indicates that the flame is approaching the alarm or trip point, and the operator needs to take preventive measures before the equipment trips. However, this indicator is only based on the three parameters of flame intensity, flicker frequency and AC amplitude, and does not cover spatial distribution parameters. In addition, the detector's narrow field of view (usually 3°-15°) will lead to missed detections, false detections, etc. under the requirements of deep peak regulation.

[0006] However, it should be noted that the stability of flame combustion is affected by multiple factors, and there is a lack of accurate definition of stability and no unified measurement standard for relative measurement values. Although some scholars have conducted preliminary explorations on the measurement of burner flame stability based on various technologies in recent years, there are still some challenges. Due to the harsh furnace environment and complex fuel composition in industrial combustion systems, flame stability measurement methods based on laser measurement are difficult to apply. Existing stability measurement methods based on visible light flame images usually use a certain flame feature alone to reflect the stability of flame combustion. This method of using single angle stability is often difficult to reflect the true flame stability.

[0007] Furthermore, unit operating parameters are crucial. Under both deep peak-shaving conditions and normal operation, various mill combinations and heat loads will affect the combustion flame morphology and flame image characteristic parameters to a certain extent. Therefore, combining unit parameters can, on the one hand, accumulate flame image data under different operating conditions, enabling big data learning of flame image characteristic parameters and improving the accuracy of flame image judgments on stable combustion. On the other hand, unit parameters also effectively supplement the information provided by flame images, making the resulting quantitative combustion stability criteria more accurate. Summary of the Invention

[0008] This invention discloses a method and system for monitoring combustion instability in coal-fired boilers based on multi-parameter fusion. This method addresses the inaccuracy of existing single-parameter criteria and is particularly suitable for early warning of combustion instability under deep peak-shaving conditions. By analyzing flame radiation images, the invention measures the flame temperature and radiation parameters within the furnace. Based on the measured flame temperature and radiation parameters, combined with unit operating parameters, a Combustion Instability Index (CII) is proposed as an important reference indicator for determining boiler combustion stability.

[0009] The method of the present invention comprises the following steps:

[0010] (1) Construct a furnace flame imaging system, including an image detector, a lens, a stainless steel cavity, and a cooling air channel;

[0011] (2) The flame image intensity, exposure time, and radiation intensity are established through blackbody furnace calibration, and the R, G, and B channel monochrome image intensities of the image detector are converted into radiation intensity based on Planck's law;

[0012] (3) Calculate flame characteristic parameters, including average flame radiation intensity, average flame temperature, average emissivity, radiation intensity standard deviation, flame temperature standard deviation, emissivity standard deviation, and flame area;

[0013] (4) A load-flame characteristic parameter dynamic prediction model is constructed based on multiple linear regression. By analyzing the response characteristics of each flame characteristic parameter under different loads, a mapping relationship between the load and each flame characteristic parameter is established to predict the baseline value of the flame characteristic parameter under the target load;

[0014] (5) Based on the real-time flame characteristic parameters calculated in step (3) and the reference values ​​predicted in step (4), the deviations of the parameters are calculated and normalized, and weighted fusion is performed according to preset weight coefficients to generate the CII index, where the sum of the weight coefficients is 1;

[0015] (6) Analyze, evaluate and judge the flame stability based on the combustion instability index.

[0016] Preferably, the calculation of the flame characteristic parameters in step (3) includes:

[0017] (a) Average flame radiation intensity I m :

[0018]

[0019] Among them, I j are the radiation intensity of the jth pixel, and n is the total number of pixels within the flame surface;

[0020] (b) Average flame temperature T m and the average emissivity ε m :

[0021]

[0022] Among them, T j , ε j are the flame temperature and emissivity of the j-th pixel respectively;

[0023] (c) Spatial parameters: Calculate the standard deviation of radiation intensity I s , flame temperature standard deviation T s , emissivity standard deviation ε s and the flame area F a , the formula is as follows:

[0024]

[0025] Where m is the total number of image pixels.

[0026] Preferably, the flame temperature and emissivity of the pixel in step (3) are solved by the equation system:

[0027] I(λ j ,T)=(a0+a1λ j )I b (λ j ,T),j=R,G,B (8)

[0028] Among them, I(λ j ,T) is the flame at wavelength λ j , radiation intensity at temperature T, I b (λ j ,T) is the wavelength λ j , blackbody radiation intensity at temperature T; a0 and a1 are the coefficients to be solved for the linear emissivity model; the radiation intensity of the R, G, and B channels is obtained by calibration in step (2); when the radiation intensity and wavelength are known, the equation group is solved by the nonlinear least squares method to obtain the optimal solution, and the flame temperature T, a0 and a1 are obtained; the emissivity ε=a0+a1*λ j.

[0029] Preferably, step (4) specifically includes:

[0030] (4a) Collect flame characteristic parameters under different load conditions;

[0031] (4b) Analyze the response characteristics of each flame characteristic parameter and load;

[0032] (4c) Taking load as the independent variable and each flame characteristic parameter as the dependent variable, a multivariate linear regression relationship model between load and each parameter is established through polynomial fitting;

[0033] (4d) Based on the regression relationship model, the target load value is input and the baseline value of each flame characteristic parameter under the load is predicted.

[0034] Preferably, in step (5), when CII is closer to 0, the flame instability index is lower and the flame is more stable; when CII is closer to 1, the flame is more unstable; when the measured flame area is 0 or the average flame temperature is lower than 900°C, CII is directly 1.

[0035] Preferably, the method further comprises step (7): executing a combustion control strategy after calculating the CII index:

[0036] If CII is ≥0.8 or CII=1, an audible and visual alarm is triggered and no more than two oil guns in adjacent layers are put into operation. At the same time, the primary air volume is increased to stabilize combustion. After 60 seconds, recheck CII. If it is still ≥0.8, perform MFT.

[0037] If 0.5≤CII<0.8, adjust the pulverizer output within the range of ±5% to balance the coal powder concentration, and at the same time adjust the secondary air damper opening by +5%.

[0038] The present invention also provides a coal-fired boiler combustion instability monitoring system based on multi-parameter fusion based on the method, comprising:

[0039] Imaging module: Contains image detector, lens, stainless steel cavity and cooling air channel, used to collect furnace flame images in real time;

[0040] Calibration module: converts flame image intensity into radiation intensity through a blackbody furnace;

[0041] Feature extraction module: calculate flame characteristic parameters;

[0042] Modeling module: Construct load-flame characteristic parameter model through multiple linear regression;

[0043] CII calculation module: integrates real-time feature parameters and model prediction values ​​to output weighted CII index;

[0044] Judgment module: compares the CII index with the CII threshold to judge the flame stability;

[0045] Control execution module: displays the CII index and combustion status evaluation results in real time; triggers different control strategies according to the numerical range of the CII index.

[0046] Beneficial effects:

[0047] The present invention proposes a method and system for monitoring combustion instability of coal-fired boilers by combining multi-parameter fusion of flame radiation images and unit operating parameters. By fusing the intensity parameters (such as average temperature, radiation intensity, emissivity) of the flame radiation image with spatial parameters (such as temperature standard deviation, emissivity standard deviation, flame area), and combining the unit operating parameters to construct a combustion instability index (CII), it achieves accurate monitoring and quantitative evaluation of the combustion stability of coal-fired boilers, and has significant technical advantages and engineering value. First, the multi-parameter fusion strategy breaks through the limitations of traditional single feature monitoring. By analyzing the synergistic characteristics of average temperature fluctuations and emissivity standard deviations under low-load conditions, it can more comprehensively capture the precursors of combustion instability, which is a significant improvement over the narrow field of view and high false detection rate of traditional detectors. Second, the CII index forms a quantitative indicator by normalizing the parameter change rate and weighted summing it, thus achieving dynamic quantitative evaluation of the combustion state. Third, the non-contact measurement solution based on digital imaging technology avoids the application limitations of laser measurement in harsh furnace environments and solves the limitations of traditional single characteristic parameter monitoring. Verified by the implementation examples, the warning accuracy of the CII index in the 85MW low-load condition of a coal-fired boiler is more than 30% higher than that of the traditional single-parameter method. It is particularly suitable for early warning of combustion instability under deep peak-shaving conditions of boilers. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of a coal-fired boiler combustion instability monitoring method based on multi-parameter fusion;

[0049] Figure 2 This is a schematic diagram of the blackbody furnace radiation calibration system;

[0050] Figure 3 This is the load variation of a 330MW coal-fired power generation unit over two days;

[0051] Figure 4 The flame images under 285MW load, (a) is camera 1, (b) is camera 2, (c) is camera 3, (d) is camera 4,

[0052] (e) is camera 5, (f) is camera 6, (g) is camera 7, and (h) is camera 8;

[0053] Figure 5Flame images under 85MW load, (a) is camera 1, (b) is camera 2, (c) is camera 3, and (d) is camera 4.

[0054] (e) is camera 5, (f) is camera 6, (g) is camera 7, and (h) is camera 8;

[0055] Figure 6 Flame radiation images, (a) is camera 1, 85MW; (b) is camera 2, 85MW; (c) is camera 1, 285MW; (d) is camera 2, 285MW;

[0056] Figure 7 The measurement results of 285MW load, (a) is the flame temperature of camera 1, (b) is the flame temperature of camera 2, (c) is the emissivity of camera 1, and (d) is the emissivity of camera 2;

[0057] Figure 8 The measurement results of 85MW load, (a) is the flame temperature of camera 1, (b) is the flame temperature of camera 2, (c) is the emissivity of camera 1, and (d) is the emissivity of camera 2;

[0058] Figure 9 is the change of average radiation energy intensity, (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4;

[0059] Figure 10 is the average flame temperature change, (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4;

[0060] Figure 11 The average flame emission rate changes, (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4;

[0061] Figure 12 The standard deviation of flame radiation intensity changes, (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4;

[0062] Figure 13 is the standard deviation of flame temperature, (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4;

[0063] Figure 14 The standard deviation of flame emissivity is shown in Figure 2. (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4.

[0064] Figure 15 The flame area changes are shown in Figure 1. (a) is nozzle No. 1, (b) is nozzle No. 2, (c) is nozzle No. 3, and (d) is nozzle No. 4.

[0065] Figure 16 is the load-radiation intensity response relationship;

[0066] Figure 17 To compare the measured radiation intensity with the optimal radiation intensity;

[0067] Figure 18 Response relationships: (a) is the load-flame temperature response relationship, and (b) is the load-flame emissivity response relationship.

[0068] Figure 19 For comparison between measured and fitted values, (a) is the flame temperature, (b) is the flame emissivity;

[0069] Figure 20 are response relationships, (a) is the load-radiation intensity standard deviation response relationship, (b) is the load-emissivity standard deviation response relationship;

[0070] Figure 21 For comparison between measured and fitted values, (a) is the standard deviation of radiation intensity, and (b) is the standard deviation of emissivity;

[0071] Figure 22 The CII changes during two days of operation. DETAILED DESCRIPTION

[0072] The following is a further detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the following embodiments may be combined with each other without conflict.

[0073] like Figure 1 As shown, the present invention provides a method for monitoring combustion instability of a coal-fired boiler based on multi-parameter fusion, comprising the following steps:

[0074] Step 1: Construct a furnace flame imaging system: The furnace flame imaging system includes: an image detector, a lens, a stainless steel cavity and a cooling air channel.

[0075] Step 2: Calibration of furnace flame imaging system: Multi-parameter calibration includes flame image intensity, exposure time and radiation intensity, such as Figure 2 Blackbody furnace calibration is used to establish a conversion relationship between the flame image intensity and radiation intensity captured by the image detector. The flame image is then used to obtain the flame intensity image for the corresponding channel. Based on the results of the blackbody furnace radiation calibration, the monochrome image intensity of the industrial camera's R, G, and B channels in the image detector can be converted into radiation intensity. According to Planck's law, blackbody radiation intensity is a function of temperature and wavelength:

[0076]

[0077] In formula (1), I b (λ,T) is the blackbody radiation intensity, W / m 3 ε(λ) is the spectral emissivity; h is Planck's constant, J·s; c is the speed of light, m / s; k is the Boltzmann constant, J / K; λ is the wavelength, m; and T is the temperature of the radiating object. The furnace flame temperature detection system, after being calibrated with a blackbody furnace's radiation intensity, can convert image intensity into radiation intensity.

[0078] Step 3: Calculate the flame radiation intensity, flame temperature, flame emissivity, flame radiation intensity standard deviation, and flame emissivity standard deviation: A direct manifestation of combustion instability is the dramatic fluctuations in the flame's spontaneous radiation in the visible light band, including fluctuations in time, space, and radiation intensity. The radiation intensity corresponding to each pixel in the flame radiation image is related to the flame temperature and emissivity corresponding to that pixel. To quantify flame stability, the flame radiation image is preprocessed to obtain flame characteristic parameters. In the present invention, flame characteristic parameters are divided into two categories: spatial parameters of radiation and intensity parameters. The flame characteristic parameters are processed into spatial parameters of radiation and intensity parameters based on the flame radiation image.

[0079] For actual thermal radiation objects such as pulverized coal flames, the emissivity is a function of wavelength, and its radiation intensity can be expressed as:

[0080] I(λ, T)=ε(λ)I b (λ, T) (2)

[0081] For thermal radiation objects such as solid surfaces and continuous spectra in the visible light band of flames, their spectral emissivity can be expressed as a polynomial function of wavelength:

[0082] ε(λ)=a0+a1λ+a2λ 2 +…+a m λ m (3)

[0083] In formula (3), m is the order of the polynomial. In this preferred embodiment, m=1, that is, a linear emissivity model is adopted. Then formula (2) can be expressed as:

[0084] I(λ, T)=(a0+a1λ)I b (λ, T) (4)

[0085] According to the results of the blackbody furnace radiation calibration, the monochrome image intensity S of the three channels of the industrial camera R, G and B in the image detector can be R 、S G 、S B Converted to radiation intensity I(λR ,T)、I(λ G ,T)、I(λ B , T), we can get the equation group:

[0086] I(λ j , T)=(a0+a1λ j )I b (λ j , T), j=R, G, B (5)

[0087] In formula (5), λ j is the central wavelength corresponding to the three channels R, G, and B. There are three unknown quantities a0, a1, and T in the equation system. The equation system is positive definite and the optimal solution can be obtained according to the nonlinear least squares method. Finally, the flame temperature T and emissivity ε can be obtained.

[0088] The temperature T directly reflects the temperature distribution of the furnace flame, while the emissivity reflects the distribution of the relative concentration of pulverized coal. This is because the emissivity is the result of the accumulation of the absorption coefficient over the line of sight, and the absorption coefficient is proportional to the soot concentration. Therefore, the emissivity can be regarded as the result of the relative soot concentration distribution over the line of sight.

[0089] Table 1 gives the names and symbols of the spatial parameters and intensity parameters of radiation.

[0090] Table 1 Flame characteristic parameters

[0091]

[0092] The calculation process of the radiation intensity, flame temperature and emissivity in the intensity parameters has been given. The average radiation intensity, average flame temperature and average emissivity are calculated as follows:

[0093]

[0094] Among them, I j 、T j , ε j are the radiation intensity, temperature and emissivity of the j-th pixel respectively, and n is the total number of pixels within the flame surface.

[0095] Spatial parameters reflect the spatial distribution characteristics of the combustion flame within the furnace. Due to the influence of various external factors, the flame temperature distribution is uneven, and even localized high temperatures may occur. Similarly, the pulverized coal concentration is also uneven due to various factors, which also greatly affects combustion stability. The specific calculation process of spatial parameters is shown in the following formula:

[0096]

[0097]

[0098] Where m is the total number of image pixels.

[0099] Step 4. Construct a load-flame characteristic parameter model: According to formulas (6) to (12), respectively calculate and analyze the intensity parameters such as average radiation intensity, temperature, emissivity, and spatial parameters such as standard deviation and flame area, as well as the response characteristics of these parameters under different load conditions, and establish a mapping relationship between load and each parameter to provide thermodynamic and data support for the construction of the combustion instability index, thereby solving the limitations of single parameter monitoring. Among them, a polynomial model is used to construct the response relationship between flame characteristic parameters and load. According to the flame characteristic parameter-load response model obtained by fitting, the flame characteristic parameters under different load requirements are calculated. The results are as follows: Figures 16 to 21 Then the error between the measured value and the calculated value is obtained and normalized. Finally, the normalized flame characteristic parameters are weighted and summed to form the flame combustion instability index (CII). The specific calculation process of CII is shown in formula (13), where w1, w2, w3, w4, w5, w6, and w7 are weight coefficients, and their sum is 1, I' m , T' m 、ε' m 、I' s , T' s , ε'3 and F' a They are the errors between the measured and calculated values ​​of the average flame radiation intensity, average flame temperature, average emissivity, standard deviation of radiation intensity, standard deviation of temperature, standard deviation of emissivity, and flame area.

[0100] CII=w1·I' m +w2·T m '+w3·ε' m +w4·I s '+w5·T s '+w6·ε s '+w7·F a ' (13)

[0101] Step 5: Predict the flame characteristic parameters using the load-flame characteristic parameter prediction model according to the target load.

[0102] Step 6: Calculate the flame instability index based on the deviation between the measured and predicted values: Calculate the deviation between the measured and predicted values ​​based on the real-time monitored flame radiation intensity, temperature, emissivity, radiation intensity standard deviation, and emissivity standard deviation. The deviations of these five flame characteristic parameters are normalized and weighted together, with the sum of the weight coefficients being 1. This is the flame instability index (CII).

[0103] Step 7: Determine flame instability: Analyze and evaluate the constructed combustion instability index. When the CII value approaches 0, the flame instability index is lower and the combustion state is more stable. Conversely, when the CII value approaches 1, the combustion state is more unstable. Additional criteria are added to determine combustion instability. When the measured flame area is 0 or the average flame temperature is below 900°C, the CII is directly 1, indicating extreme instability.

[0104] Example 1

[0105] The present embodiment is a coal-fired boiler combustion instability monitoring method based on multi-parameter fusion, and its research object is a 330MW coal-fired power generation unit. Figure 3 As shown, this embodiment uses the continuous operation data of a 330MW coal-fired power generation unit from May 22 to 23, 2025 as the test object. Since the flame image acquisition system collects data once every 1 minute, the load changes in the DCS system are also collected at a frequency, that is, each time sampling point corresponds to one minute. Taking May 22 as an example, first of all, it can be seen from the figure that in the sampling point 400 to 900 section, that is, from about 7 am to 3 pm, the unit load is relatively low, and the overall load is less than 100MW. When the time comes to the afternoon and evening time period, that is, the time sampling point 900 to 1200 section, the electricity demand ushers in a peak, and the unit load is also adjusted to the peak value. Specifically including the following steps:

[0106] Step 1: Build the furnace flame imaging system: The furnace flame imaging system consists of an image detector, a lens, a stainless steel cavity, and a cooling air duct. To more comprehensively capture the flame parameters within the furnace, eight industrial cameras were used as image detectors. Cameras 1 through 4 capture flame images at nozzles 1 through 4, respectively, while cameras 5 through 8 capture images of the flame within the furnace.

[0107] Step 2: Calibration of the furnace flame imaging system: Multi-parameter calibration includes flame image intensity, exposure time, and radiation intensity. Use a blackbody furnace to calibrate the image detector, compare and convert the flame image intensity with the radiation intensity, and obtain the flame intensity image under the corresponding channel through the flame image. Figure 4 and Figure 5 The flame images under loads of 285MW and 85MW are given respectively. Image intensity cannot quantitatively reflect the radiation intensity, so it is necessary to convert the image intensity into radiation intensity. The flame radiation images taken by cameras 1 and 2 under the two working conditions of 285MW and 85MW are selected, and the results are shown as follows: Figure 6 shown.

[0108] Step 3: Calculate the flame radiation intensity, flame temperature, flame emissivity, flame radiation intensity standard deviation and flame emissivity standard deviation. Calculate these parameters using the above formulas and get the following results: Figures 7 to 15 .

[0109] Step 4. Construct a load-flame characteristic parameter model: According to formulas (6) to (12), respectively calculate and analyze the intensity parameters such as average radiation intensity, temperature, emissivity, and spatial parameters such as standard deviation and flame area, as well as the response characteristics of these parameters under different load conditions, and establish a mapping relationship between load and each parameter to provide thermodynamic and data support for the construction of the combustion instability index, thereby solving the limitations of single parameter monitoring. Among them, a polynomial model is used to construct the response relationship between flame characteristic parameters and load. According to the flame characteristic parameter-load response model obtained by fitting, the flame characteristic parameters under different load requirements are calculated. The results are as follows: Figures 16 to 21 Then the error between the measured value and the calculated value is obtained and normalized. Finally, the normalized flame characteristic parameters are weighted and summed according to formula (13) to obtain the flame combustion instability index CII.

[0110] like Figures 7 to 13 , average radiation intensity and average flame temperature are positively correlated with actual power, increasing with increasing actual power and decreasing with decreasing actual power. Average radiation intensity and average flame temperature exhibit more pronounced fluctuations under low-load conditions. Considering that flames are prone to combustion instability under low-load conditions, average temperature can be used as an effective flame characteristic parameter to reflect combustion instability. In contrast to these two characteristic parameters, average emissivity exhibits a negative correlation with actual power and tends to fluctuate dramatically under low-load conditions. Therefore, changes in these three flame characteristic parameters can effectively reflect the stability of the combustion process under low-load conditions.

[0111] Figure 14 The flame emissivity standard deviation shown in Figure 2 shows a more pronounced negative linear correlation with load. The flame emissivity standard deviation decreases with increasing load, and correspondingly increases with decreasing load. Under high load conditions, the flame temperature is higher, resulting in more complete combustion, less incompletely burned pulverized coal, and a more uniform distribution. While emissivity reflects the relative concentration of pulverized coal to some extent, the emissivity standard deviation describes the uniformity of the flame emissivity, and therefore the uniformity of the pulverized coal concentration. Therefore, the emissivity standard deviation is also a high-quality flame characteristic parameter that can effectively measure the flame combustion process.

[0112] Figure 15There is no linear relationship between flame area and actual power, but at low loads, the flame area exhibits significant fluctuations. This indicates that flame combustion is unstable under low load conditions, resulting in significant and dramatic fluctuations in the flame area. However, at high loads, the flame area is very stable. Therefore, flame area can be used as an important parameter to reflect flame combustion instability.

[0113] like Figures 16 to 21 The response characteristics of flame characteristic parameters under different operating conditions are studied, and then the quantitative standards for flame characteristic parameters to measure flame combustion stability and instability are given. Polynomials are used to fit the relationship between load and these parameters.

[0114] Step 5: Use the load-flame characteristic parameter prediction model to predict the flame characteristic parameters based on the target load. In this embodiment, the load of a 330MW coal-fired power generation unit for 48 hours over two days from May 22 to 23, 2025 is selected.

[0115] Step 6: Calculate the flame instability index based on the deviation between the measured and predicted values: Calculate the deviation between the measured and predicted values ​​based on the real-time monitored flame radiation intensity, temperature, emissivity, radiation intensity standard deviation, and emissivity standard deviation. The deviations of these five flame characteristic parameters are normalized and weighted together, with the sum of the weight coefficients being 1. This is the flame instability index (CII).

[0116] Step 7: Determine flame instability: Analyze and evaluate the constructed combustion instability index. The closer the CII is to 0, the lower the flame instability index and the more stable the flame. Conversely, the closer the CII is to 1, the more unstable the flame. Additional criteria are added to determine combustion instability. When the measured flame area is 0 or the average flame temperature is below 900°C, the CII is directly 1, indicating extreme instability.

[0117] In this embodiment, the CII changes during the 48-hour operation period from May 22 to May 23, 2025 are calculated. Figure 22 ,Under low load conditions, the CII value is relatively higher, indicating that the flame combustion process is more unstable under low load conditions. In contrast, the CII under high load conditions is lower, indicating that the overall combustion process is stable.

[0118] This embodiment integrates the intensity parameters (average temperature, radiation intensity, and emissivity) of the flame radiation image with spatial distribution parameters (temperature standard deviation, emissivity standard deviation, and flame area), covering both energy level and spatial uniformity. Combined with unit operating parameters (load, coal flow, air volume, etc.), this approach overcomes the limitations of single image features and forms a collaborative analysis framework for "load-flame characteristics." A quantitative relationship is established between radiation intensity and temperature and emissivity, converting the optical signal (image intensity) into thermodynamic parameters (temperature and pulverized coal concentration) to achieve physical characterization of the combustion state.

[0119] In actual operation, it breaks through the limitations of single characteristics and combines intensity and spatial distribution parameters to more comprehensively reflect the combustion status; considers the time series characteristics of parameter change rate to adapt to dynamic working conditions such as deep peak regulation; based on digital imaging technology, it is low-cost, easy to deploy, and realizes real-time monitoring.

[0120] Furthermore, combustion control can be performed after calculating the CII index:

[0121] If CII is ≥0.8 or CII=1, the sound and light alarm is triggered and the adjacent layer oil guns (≤2) are put into operation. At the same time, the primary air volume is increased to stabilize the combustion. After 60 seconds, the CII is rechecked. If it is still ≥0.8, MFT is performed.

[0122] If 0.5≤CII<0.8, adjust the coal mill output (within ±5%) to balance the pulverized coal concentration, and adjust the secondary air damper opening by +5%;

[0123] Store CII and corresponding working condition parameters in the database in real time.

[0124] Example 2

[0125] This embodiment provides a coal-fired boiler combustion instability monitoring system based on multi-parameter fusion, including:

[0126] The imaging module, consisting of an image detector, lens, stainless steel cavity, and cooling air duct, is used to capture real-time images of the furnace flame. The image detector is connected to the lens, which is then connected to the stainless steel cavity, which in turn is connected to the cooling air duct. The image detector receives the flame image signal, while the lens captures the flame light. The stainless steel cavity provides protection for internal components, and the cooling air duct cools the system by supplying cooling air to adapt to the high temperature environment within the furnace and ensure stable operation of the imaging system.

[0127] Calibration Module: Based on Planck's law and blackbody furnace calibration data, a nonlinear mapping relationship between image intensity and radiation intensity is established. During the calibration process, multiple parameters are calibrated, including flame image intensity, exposure time, and radiation intensity. Calibration of the image detector using a blackbody furnace allows for the conversion of flame image intensity and radiation intensity, thereby obtaining a flame intensity image for the corresponding channel from the flame image. Based on the results of the blackbody furnace radiation calibration, the monochrome image intensity of the industrial camera's R, G, and B channels in the image detector can be converted to radiation intensity, laying the foundation for the subsequent calculation of parameters such as flame temperature and emissivity based on Planck's law.

[0128] Feature Extraction Module: Calculates flame characteristic parameters, including average flame radiation intensity, average flame temperature, average emissivity, radiation intensity standard deviation, flame temperature standard deviation, emissivity standard deviation, and flame area. These parameters are calculated based on flame radiation images calibrated with a blackbody furnace. By processing the radiation intensity, temperature, and emissivity of each pixel in the image, these characteristic parameters are ultimately extracted, providing basic data for the subsequent construction of a load-flame characteristic parameter model and calculation of the combustion instability index.

[0129] Modeling module: Construct a load-flame characteristic parameter model through multiple linear regression. With the unit load as the independent variable and the various flame characteristic parameters as the dependent variables, a multiple linear regression relationship model between the load and the various parameters is established through polynomial fitting. This process aims to analyze the response characteristics of different characteristic parameters under different load conditions, establish a mapping relationship between the load and the various parameters, and provide data support for the subsequent calculation of the combustion instability index. Based on the constructed regression relationship model, when the target load value is input, the baseline value of each flame characteristic parameter under the load can be predicted. This baseline value serves as a reference standard for measuring whether the real-time flame characteristic parameters are normal. It is an important basis for calculating parameter deviations and then generating the combustion instability index, thereby solving the limitations of traditional single parameter monitoring and improving the accuracy of combustion stability assessment.

[0130] The CII calculation module integrates real-time characteristic parameters with model predictions to output a weighted CII index. The CII calculation module presets weight coefficients summing to 1 and includes logic to force a CII = 1 when the flame area is zero or the average temperature is below 900°C. This module receives the real-time flame characteristic parameters calculated by the feature extraction module and the parameter baseline values ​​at the corresponding load, predicted by the modeling module using the load-flame characteristic parameter model. Based on the real-time characteristic parameters and the baseline values, it calculates the deviations of each parameter and normalizes these deviations to eliminate the influence of dimensional differences between the parameters. The normalized parameter deviations are weighted and summed using preset weight coefficients to generate the CII index, where the sum of the weight coefficients is 1. The weights are set to reflect the impact of different characteristic parameters on combustion stability. This module includes a forced judgment rule that directly determines the combustion instability index CII = 1 when the real-time flame area is zero or the average flame temperature is below 900°C, indicating extreme combustion instability.

[0131] The judgment module, which includes a comparator unit and a threshold storage unit, compares the CII index with a preset threshold (0.5) to determine flame stability. This module uses the CII index output by the CII calculation module as the core judgment indicator and evaluates the combustion state based on the CII index's numerical range: CII values ​​closer to 0 indicate more stable flame combustion; CII values ​​closer to 1 indicate less stable combustion.

[0132] Control execution module: Displays the CII index and combustion status assessment results in real time; triggers different control strategies based on the CII index's numerical range: If CII ≥ 0.8 or CII = 1, an audible and visual alarm is triggered, and no more than two adjacent layer oil guns are activated, while the primary air volume is increased to stabilize combustion; rechecks CII after 60 seconds; if it is still ≥ 0.8, MFT (main fuel trip) is executed. If 0.5 ≤ CII < 0.8, adjusts the mill output (within ±5%) to balance the pulverized coal concentration, and optimizes the secondary air distribution (burnout damper opening + 5%). The CII index and corresponding operating parameters are stored in a database in real time, providing historical data support for subsequent combustion status analysis and model optimization.

[0133] According to industrial tests, the system's early warning accuracy rate reached 92% under 85MW low-load conditions, a 37% improvement over traditional single-parameter monitoring methods, and the false alarm rate was less than 5%, meeting the real-time monitoring requirements of power plants.

[0134] It should be noted that the above description is only an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the scope of protection of the claims of the present invention.

Claims

1. A method for monitoring combustion instability of coal-fired boilers based on multi-parameter fusion, characterized in that: The following steps are involved: (1) Construct a furnace flame imaging system, including an image detector, a lens, a stainless steel cavity, and a cooling air channel; (2) The flame image intensity, exposure time, and radiation intensity are established through blackbody furnace calibration, and the R, G, and B channel monochrome image intensities of the image detector are converted into radiation intensity based on Planck's law; (3) Calculate flame characteristic parameters, including average flame radiation intensity, average flame temperature, average emissivity, radiation intensity standard deviation, flame temperature standard deviation, emissivity standard deviation, and flame area; (4) A load-flame characteristic parameter dynamic prediction model is constructed based on multiple linear regression. By analyzing the response characteristics of each flame characteristic parameter under different loads, a mapping relationship between the load and each flame characteristic parameter is established to predict the baseline value of the flame characteristic parameter under the target load; (5) Based on the real-time flame characteristic parameters calculated in step (3) and the reference values ​​predicted in step (4), the deviations of the parameters are calculated and normalized, and weighted fusion is performed according to preset weight coefficients to generate the CII index, where the sum of the weight coefficients is 1; (6) Analyze, evaluate and judge the flame stability based on the combustion instability index.

2. The method according to claim 1, characterized in that The calculation of flame characteristic parameters in step (3) includes: (a) Average flame radiation intensity I m : Among them, I j are the radiation intensity of the jth pixel, and n is the total number of pixels within the flame surface; (b) Average flame temperature T m and the average emissivity ε m : Among them, T j , ε j are the flame temperature and emissivity of the j-th pixel respectively; (c) Spatial parameters: Calculate the standard deviation of radiation intensity I s , flame temperature standard deviation T s , emissivity standard deviation ε s and the flame area F a , the formula is as follows: Where m is the total number of image pixels.

3. The method according to claim 2, characterized in that The flame temperature and emissivity of the pixel in step (3) are solved by the equation system: I(λ j ,T)=(a0+a1λ j )I b (λ j ,T),j=R,G,B (8) Among them, I(λ j ,T) is the flame at wavelength λ j , radiation intensity at temperature T, I b (λ j ,T) is the wavelength λ j , blackbody radiation intensity at temperature T; a0 and a1 are the coefficients to be solved for the linear emissivity model; the radiation intensity of the R, G, and B channels is obtained by calibration in step (2); when the radiation intensity and wavelength are known, the equation group is solved by the nonlinear least squares method to obtain the optimal solution, and the flame temperature T, a0 and a1 are obtained; the emissivity ε=a0+a1*λ j .

4. The method according to claim 1, wherein Step (4) specifically includes: (4a) Collect flame characteristic parameters under different load conditions; (4b) Analyze the response characteristics of each flame characteristic parameter and load; (4c) Taking load as the independent variable and each flame characteristic parameter as the dependent variable, a multivariate linear regression relationship model between load and each parameter is established through polynomial fitting; (4d) Based on the regression relationship model, the target load value is input and the baseline value of each flame characteristic parameter under the load is predicted.

5. The method according to claim 1, characterized in that In step (5), when CII is closer to 0, the flame instability index is lower and the flame is more stable; when CII is closer to 1, the flame is more unstable; when the measured flame area is 0 or the average flame temperature is lower than 900°C, CII is directly 1.

6. The method according to claim 1, characterized in that The method further includes step (7): executing a combustion control strategy after calculating the CII index: If CII is ≥0.8 or CII=1, an audible and visual alarm is triggered and no more than two oil guns in adjacent layers are put into operation. At the same time, the primary air volume is increased to stabilize combustion. After 60 seconds, recheck CII. If it is still ≥0.8, perform MFT. If 0.5≤CII<0.8, adjust the pulverizer output within the range of ±5% to balance the coal powder concentration, and at the same time adjust the secondary air damper opening by +5%.

7. A coal-fired boiler combustion instability monitoring system based on multi-parameter fusion based on the method according to any one of claims 1 to 6, characterized in that: include: Imaging module: Contains image detector, lens, stainless steel cavity and cooling air channel, used to collect furnace flame images in real time; Calibration module: converts flame image intensity into radiation intensity through a blackbody furnace; Feature extraction module: calculate flame characteristic parameters; Modeling module: Construct load-flame characteristic parameter model through multiple linear regression; CII calculation module: integrates real-time feature parameters and model prediction values ​​to output weighted CII index; Judgment module: compares the CII index with the CII threshold to judge the flame stability; Control execution module: displays the CII index and combustion status evaluation results in real time; triggers different control strategies according to the numerical range of the CII index.

Images