An online identification system for pollutant generation region of oxy-combustion based on field distribution perception

Abstract

The present application relates to the technical field of oxygen-enriched combustion diagnosis and pollution control, and particularly relates to an online identification system for pollution generation area of oxygen-enriched combustion based on field distribution sensing, which comprises a sensor array module, a signal acquisition and preprocessing module, a field distribution image reconstruction module, a region segmentation and feature extraction module, a pollution identification module and a result output and model updating module. The sensor array module applies a multi-frequency excitation signal to a furnace cross-section and collects boundary responses; the field distribution image reconstruction module reconstructs the cross-section complex permittivity distribution by solving a severely ill-conditioned inverse problem with total variation regularization and time sequence constraints; the region segmentation and feature extraction module extracts volatile release area and coke combustion area respectively; the pollution identification module establishes a NO x generation intensity, CO generation tendency and oxidation-reduction game index multi-dimensional judgment system to identify the pollution generation area and intensity trend online.

Description

Technical Field

[0001] This invention relates to the field of oxygen-enriched combustion diagnostics and pollutant control technology, specifically to an online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing. Background Technology

[0002] With the development of carbon capture and storage technology, O2 / CO2 oxy-fuel combustion technology has become one of the important pathways for the low-carbon transformation of coal-fired power plants. When low-calorific-value coal (such as coal gangue, middlings coal, lignite, etc.) is burned in an O2 / CO2 atmosphere, due to its low volatile matter, high ash content, and low calorific value, its combustion reaction zone frequently migrates within the furnace and overlaps spatially, making it difficult to predict the pollutant generation area. This creates an information blind spot regarding where to treat pollutants, which hinders proactive control measures such as staged air supply and zoned ammonia injection.

[0003] Electrical impedance tomography (EIT) technology has been applied in cold-state conditions such as gas-solid two-phase flow pattern identification and bubble distribution detection. Its basic principle is to reconstruct the conductivity distribution within the field by applying electrical excitation through an electrode array placed at the boundary of the measured field and measuring the boundary response voltage. However, applying EIT technology to identify pollutant generation regions in high-temperature, oxygen-enriched combustion environments faces the following challenges: First, the conductivity of the medium changes drastically with temperature at high temperatures, rendering the cold-calibrated sensitivity field model ineffective; second, CO2 in the combustion medium partially ionizes at high temperatures, generating additional ionic conductivity background that drowns out the effective signal; and third, there is a lack of a quantitative mapping model from electrical parameter distribution to pollutant generation regions. Currently, there is no solution in the technology that uses field distribution sensing methods for online identification of combustion pollutant generation regions.

[0004] Current research on the generation of pollutants from oxygen-enriched combustion mainly relies on the following methods: (1) Flue gas sampling and analysis method: Flue gas samples are collected at the furnace outlet or tail flue, and NO is analyzed by gas chromatograph, mass spectrometer or Fourier transform infrared spectrometer (FTIR). x This method only obtains the average concentration value at the furnace outlet and cannot reflect the spatial distribution of pollutants generated in different areas inside the furnace, let alone distinguish NO concentrations. x Whether it originates from the volatile nitrogen pathway (HCN / NH3 oxidation) or the coke nitrogen pathway (coke-N direct oxidation).

[0005] (2) Numerical simulation (CFD): By establishing a combustion reaction kinetic model and combining computational fluid dynamics methods, the velocity field, temperature field, and concentration field in the furnace are simulated and calculated. Although this method can provide full-field information, the calculation cycle is long (single simulation lasts for several hours to several days), which cannot meet the requirements for online real-time identification, and the simulation results are highly dependent on the selection of reaction kinetic parameters.

[0006] (3) Thermocouple temperature measurement or optical flame monitoring: It can only sense the spatial distribution of temperature or flame light intensity, and cannot directly reflect the spatial location and evolution trend of the active zone of combustion chemical reaction, let alone distinguish the contribution of different types of reaction zones to the generation of pollutants.

[0007] In addition, some technologies use neural network models to predict pollutant generation. However, this approach only focuses on predicting the total amount of pollutants and lacks the ability to spatially identify pollutant generation areas inside the furnace, thus failing to provide guidance for proactive spatial control measures.

[0008] In summary, existing technologies cannot solve the problem of real-time online spatial identification of pollutant generation areas under oxygen-enriched combustion conditions. There is an urgent need for a system that can achieve online positioning of pollutant generation areas through physical field sensing without relying on complex chemical models. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention aims to provide an online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution perception. Existing technologies cannot identify the spatial location of pollutant generation areas online, nor can they distinguish the contribution of different reaction zones to pollutants.

[0010] The basic solution provided by this invention is: an online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing, comprising: A sensor array module is arranged on at least one cross section of the furnace of an oxygen-enriched combustion boiler for applying an excitation signal and acquiring boundary response data of the combustion medium to the excitation signal. The signal acquisition and preprocessing module is used to amplify, filter, perform analog-to-digital conversion and temperature drift correction on the boundary response data to generate standardized boundary measurement data. The field distribution image reconstruction module is used to reconstruct the internal field distribution image of the cross section by solving an ill-conditioned inverse problem based on the boundary measurement data and a pre-established sensitivity field model. The region segmentation and feature extraction module is used to segment the field distribution image into regions and extract a first type of reaction region and a second type of reaction region, respectively. The first type of reaction region is a connected region in the field distribution image where the dielectric loss tangent is higher than a preset background noise threshold, representing the volatile matter release and pyrolysis region. The second type of reaction region is a connected region in the field distribution image where the real part of the complex conductivity is higher than a preset background noise threshold, representing the coke combustion and gasification region. The module also calculates the geometric feature parameters and their dynamic change rates for the two types of reaction regions. The pollutant identification module is used to establish a system containing NO based on the geometric feature parameters and spatial relative positional relationship of the two types of reaction zones. xA multi-dimensional determination system of generation intensity index, CO generation tendency index and redox game index can be used to identify the main generation areas and generation intensity trends of pollutants online. The results output and model update module is used to visualize the identification results and periodically adjust the mapping weights of the multidimensional judgment system based on offline or semi-online calibration data.

[0011] Furthermore, the sensor array module includes N electrodes arranged circumferentially at equal intervals on the cross-section of the furnace, where N≥16; the excitation signal is a multi-band AC signal with a frequency range of 100kHz~10MHz; the boundary response data is the voltage difference vector between each electrode pair; each electrode is flush with the inner wall surface of the furnace and is isolated from the furnace wall by a corundum insulation layer; the electrodes are made of platinum-iridium alloy material protected by yttrium-stabilized zirconium oxide.

[0012] Furthermore, the field distribution image reconstruction module divides the cross section into... Each grid cell is used to establish boundary measurement changes. Changes in internal field distribution Linearized approximation relationship between ,in This is the sensitivity matrix. Let be the vector of complex conductivity changes of each grid cell within the cross section; the physical meaning of the internal field distribution image is the change of complex conductivity of each grid cell within the cross section, including the real part. and the virtual part The field distribution image reconstruction module reconstructs the internal field distribution image by solving the following objective functional:

[0013] in, This represents the boundary voltage vector at the current moment. This is the total variational regularization term, used to preserve the edge characteristics of the combustion front; This is a priori for temporal extrapolation prediction based on historical frame data; These are the regularization weight coefficients. These are the weighting coefficients for time-series constraints.

[0014] Furthermore, after the region segmentation and feature extraction module extracts the first type of reaction region, it calculates the area of ​​that region. Centroid coordinates and its area change rate The area change rate It is obtained by dividing the difference between two consecutive frames by the frame interval.

[0015] Furthermore, after the region segmentation and feature extraction module extracts the second type of reaction region, it calculates the area of ​​that region. Centroid coordinates and its equivalent diameter shrinkage rate The equivalent diameter is calculated based on the area of ​​a circle. .

[0016] Furthermore, the NO x Generation Intensity Index The calculation method is as follows:

[0017] in, This represents the real part of the average conductivity of the second-type reaction region. , These are the weighting coefficients determined through a pre-calibration experiment; An increase in the value indicates that NO exports x Concentrations are rising.

[0018] Furthermore, the result output and model update module utilizes the NO measured by the outlet flue gas analyzer at historical time points. x CO concentration data, adjusted at preset time intervals for the weighting coefficients. , Perform least squares recursive correction.

[0019] Furthermore, the CO generation tendency index is determined by: based on the centroidal movement rate of the second type of reaction zone. Combined with equivalent diameter shrinkage rate Make a judgment; when Greater than the preset rate threshold and When the shrinkage rate is less than the preset shrinkage threshold, it is determined that the coke burnout rate has decreased, and a high CO emission warning signal is output.

[0020] Furthermore, the redox game index The calculation method is as follows:

[0021] in, The area representing the overlap between the first type of reaction zone and the second type of reaction zone. Preset positive numbers; A higher value indicates a greater degree of spatial overlap between the volatile matter leaching zone and the coke combustion zone, and a higher probability of homogeneous NO reduction reaction.

[0022] Furthermore, in the sensor array module, each electrode is embedded with a thermocouple for synchronously measuring the wall temperature at each electrode mounting position; the signal acquisition and preprocessing module performs online temperature drift correction on the boundary response data based on the wall temperature.

[0023] The principle of this invention is as follows: 1. Acquisition and Processing of Field Distribution Sensing Signals. The electrode array in the sensor array module applies multi-frequency AC excitation signals to the combustion medium in the furnace cross-section. The frequency response characteristics of the medium in different combustion zones to the excitation signals are fundamentally different: the graphitized microcrystalline carbon in coke has a delocalized π-electron system, exhibiting significant electronic conductivity in the low-frequency range (≤500kHz), manifested as the real part of complex conductivity. Local high values; polar molecules in the volatiles (hydrocarbons containing CH, OH, and NH bonds) exhibit dielectric relaxation loss in the high-frequency range (≥1MHz), manifesting as local high values ​​of the dielectric loss tangent. Boundary measurement data, after temperature drift correction and CO2 ionization background subtraction, generate a standardized boundary voltage vector. This vector contains spatial distribution information of electrical properties in different regions within the cross-section.

[0024] 2. Regularized solution to the ill-conditioned inverse problem. Divide the cross section into... Establish a sensitivity matrix using grid cells. ( dimension, (For independent measurements). Because much smaller This inverse problem is severely ill-conditioned. This invention introduces a dual constraint for a stable solution: a total variational regularization term... By utilizing the uniform and sharply defined prior spatial structure within the combustion reaction zone, the frontal edge remains clear; temporal extrapolation predicts the prior. By leveraging the gradual continuity of the combustion state between adjacent frames, the smoothness of the solution in the time dimension is constrained. Solving the objective functional using Split-Bregman iteration yields the complex conductivity distribution image of the current cross-section within ≤25ms. .

[0025] 3. Extraction of electrical features of the reaction region. Two types of connected regions were extracted from the reconstructed image: (1) The first type of reaction region (connected region with high dielectric loss tangent): This region has dielectric loss due to dipole relaxation of polar molecules in volatiles, and the dielectric loss tangent is significantly increased, corresponding to HCN / NH3 and other NO x (2) Precursor release zone; (3) Second type reaction zone ( High-value connected regions): This region is due to the electronic conductivity of the coke carbon skeleton. The value increases significantly, corresponding to the combustion region where coke-N is converted to NO. Calculate the area of ​​both types of regions. , Centroid coordinates , Area change rate and equivalent diameter shrinkage rate This transforms the abstract field distribution image into quantifiable geometric feature parameters of the reaction zone.

[0026] 4. Multidimensional determination of pollutant formation trends. Three independent and complementary identification indices are constructed based on the characteristic parameters of the reaction zone: NO... x Generation Intensity Index The intensity of volatile matter release ( Expansion term) and coke oxidation intensity ( A weighted combination was used to comprehensively characterize NO through two pathways: volatile nitrogen and coke nitrogen. x The potential for CO generation; the CO generation tendency index identifies a surge in CO generation caused by incomplete combustion by determining whether the coke combustion zone is in a state of drastic centroid displacement and slowed equivalent diameter contraction; the oxidation-reduction game index. By calculating the spatial overlap between the two types of reaction zones ( The three indices quantify the mixing state between the volatile matter release zone (producing reducing agents HCN / NH3) and the coke combustion zone (producing NO). The higher the degree of overlap, the greater the probability of homogeneous NO reduction reaction. The three indices describe the pollutant formation pattern from three dimensions: formation intensity, incomplete combustion, and self-reduction capability, forming a complete identification logic.

[0027] 5. Adaptive Update of the Mapping Model. Using measured data from the outlet flue gas analyzer, the mapping model is updated via least-squares recursion. Weighting coefficients , Periodic corrections are performed to adaptively adjust the mapping relationship between electrical parameters and pollutant generation as the coal type or operating conditions change.

[0028] In summary, the core principle of this invention is to transform the differences in material composition in the combustion reaction zone into differences in electrical parameters, transform the differences in electrical parameters into spatial features of the field distribution image, and then transform the spatial features into quantitative indicators of pollutant generation trends, thereby achieving cross-level identification from physical field perception to chemical pollutant generation diagnosis.

[0029] The beneficial effects of this invention are as follows: 1. In terms of diagnostic methods, this invention introduces the concept of physical field distribution perception into the diagnosis of pollutant generation during oxygen-enriched combustion. Unlike existing technologies that rely on flue gas sampling (only obtaining the outlet average) or CFD simulation (with calculation cycles of several hours to several days), this invention acquires boundary response data and reconstructs cross-sectional field distribution images through a sensor array. This enables spatial localization of the pollutant generation area inside the furnace (spatial resolution better than 0.1m) with a response cycle of ≤2 seconds, without relying on complex chemical reaction kinetics. This overcomes the technical problem of existing methods being unable to identify the distribution of the combustion reaction zone online, in real time, and with spatial resolution.

[0030] 2. At the region identification level, this invention extracts a first type of reaction region (a region with a high dielectric loss tangent, characterizing the release of volatiles) and a second type of reaction region (with different electrical characteristics) The high-value zone (characterizing coke combustion) enables online differentiation within the furnace between the volatile nitrogen → HCN / NH3 release zone and the coke nitrogen → NO oxidation zone. Based on this, an oxidation-reduction game index was constructed. By calculating the spatial overlap ratio of the two types of reaction zones, the spatial probability of homogeneous NO reduction reactions (HCN / NH3 and NO→N2) was quantitatively characterized, providing a basis for NO reduction in the furnace. x The online evaluation of the generation-reduction competition relationship provides quantitative criteria. Compared to existing technologies that cannot distinguish NO... x Source pathways have enabled spatial decoupling of pollutant generation mechanisms.

[0031] 3. At the algorithm level, a dual constraint strategy of total variational (TV) regularization and temporal extrapolation prediction prior is introduced to solve the ill-conditioned inverse problem. The TV regularization term utilizes the uniform and sharp-boundary prior space characteristics within the combustion reaction zone to effectively suppress the soft field effect and boundary artifacts commonly found in field distribution reconstruction under high temperature and high ash content environments. The temporal extrapolation prediction prior utilizes the continuity of combustion state changes between adjacent frames to constrain the solution space, further improving reconstruction stability. With a 16-electrode, 1024-pixel configuration, compared to traditional linear back-projection algorithms (resolution > 0.3m) and Landweber iterations (slow convergence, susceptible to noise interference), the reconstruction algorithm of this invention improves the spatial resolution to better than 0.1m, while controlling the single-frame reconstruction time to within 25ms, achieving both accuracy and real-time performance.

[0032] 4. At the data link level, this invention establishes a complete data mapping chain from boundary voltage measurement to field distribution reconstruction, then to geometric characteristic parameters of the reaction zone, and finally to a multi-dimensional index of pollutant generation trends. Three identification indices characterize the pollutant generation status from three independent dimensions: generation intensity, degree of incomplete combustion, and self-reduction capability, forming a cross-validated multi-dimensional judgment system that avoids the risk of misjudgment by a single indicator. Simultaneously, the mapping weights are recursively corrected using actual outlet data, enabling the identification model to adaptively track changes in coal type and load adjustments.

[0033] 5. In terms of engineering applications, the sensor array of this system is non-invasively installed (electrodes flush with the inner wall), requiring no large-scale modification of the furnace structure. The yttrium-stabilized zirconium oxide-protected platinum-iridium alloy electrodes can operate stably for extended periods in an oxygen-rich, sulfur-containing atmosphere at 900~1100℃. The output NO... x / CO generation area spatial location information (e.g., "The coke combustion zone in the upper right side of the furnace is active, and NO is expected to increase significantly"). x"It will rise in 5 to 8 seconds", which can provide clear guidance on where and when to treat for active control measures in the spatial dimension, such as graded air supply, zoned ammonia injection, and optimization of the location of circulating flue gas injection, and upgrade pollutant control from passive treatment at the outlet to active intervention in the furnace zone. Attached Figure Description

[0034] Figure 1 This is a system block diagram of an embodiment of an online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to the present invention. Detailed Implementation

[0035] The following detailed description illustrates the specific implementation method: The basic implementation examples are as follows: Figure 1 As shown: An online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution perception includes a sensor array module, a signal acquisition and preprocessing module, a field distribution image reconstruction module, a region segmentation and feature extraction module, a pollutant identification module, and a result output and model update module.

[0036] This embodiment uses a 50kW circulating fluidized bed O2 / CO2 oxygen-enriched combustion test device as an application scenario for illustration. The fuel is low-calorific-value coal (low calorific value of about 14MJ / kg on received basis, ash content of 35%~40%), and the operating temperature range is 900~1000℃.

[0037] I. Sensor Array Module and Signal Acquisition and Preprocessing Module Sixteen sensor electrodes (N=16) are circumferentially arranged at equal intervals along the inner wall of the furnace at a cross-section 1.8m from the air distribution plate on the circulating fluidized bed riser (located above the dense phase zone and in the lower middle part of the main combustion zone, where both volatile matter release and coke combustion occur simultaneously, resulting in the most significant spatial distribution differences). The electrodes are made of platinum-iridium alloy (Pt90%-Ir10%) protected by yttrium-stabilized zirconium oxide (YSZ), with a 6mm diameter tip. A corundum insulation layer (Al2O2 content ≥99%, wall thickness 3mm, measured insulation resistance >200MΩ at 1000℃) is placed between the electrode and the furnace wall. The electrode tip is flush with the inner wall surface of the furnace. Each electrode is led out of the furnace via a high-temperature shielded cable (temperature resistance ≥400℃, characteristic impedance 50Ω) and connected to a multi-frequency excitation source and signal acquisition device. Actual measurements show that, under the above insulation conditions, the bypass leakage current is less than 0.05% of the excitation current, and the signal-to-noise ratio of the boundary response signal is not less than 45dB across the entire operating temperature range.

[0038] A type K thermocouple (1 mm in diameter, located at the rear end of the electrode's central axis) is embedded in the electrode to simultaneously measure the wall temperature at each electrode mounting position. The signal acquisition and preprocessing module performs temperature drift compensation on the raw voltage response data based on the wall temperature. Specifically, it obtains the linear mapping relationship between the wall temperature Tw and the corresponding temperature Tg inside the furnace, as well as the temperature correction coefficient for complex conductivity, through offline calibration. The response data at different temperatures are then uniformly corrected to the equivalent value at the reference temperature (900℃) to eliminate the influence of temperature changes on the quantitative accuracy of the reconstructed image.

[0039] A multi-frequency excitation source outputs a constant 2mA (RMS) AC excitation signal, sequentially injecting sinusoidal currents at five characteristic frequencies (200kHz, 500kHz, 1MHz, 3MHz, and 8MHz) into each pair of adjacent electrodes. An adjacent excitation-adjacent measurement mode is employed: when a pair of adjacent electrodes is used as the excitation terminal and current is injected, the response voltage values ​​are measured sequentially on the remaining adjacent electrode pairs. A single full-frequency scan (16 excitation positions × 5 frequencies = 80 independent excitations) takes 20ms. The acquired voltage signals are pre-amplified (40dB gain), bandpass filtered (passband 100kHz~10MHz), and converted to digital signal by 16-bit analog-to-digital conversion. The signal acquisition and preprocessing module then generates a standardized boundary voltage vector. .

[0040] Regarding the correction of high-temperature CO2 ionization background interference, this embodiment adopts a multi-frequency differential measurement strategy: the average response of two high-frequency points, 3MHz and 8MHz, is used as the benchmark estimate of ion conductivity background. The background contribution is subtracted from the low-frequency measurement data of 200kHz and 500kHz to obtain the effective complex conductivity change after deducting CO2 ionization interference.

[0041] II. Field Distribution Image Reconstruction Module The field distribution image reconstruction module divides the furnace cross-section into 32×32=1024 grid elements. Based on a pre-established sensitivity field model (obtained through finite element simulation, where the response coefficients of each electrode to the measured value are set for a unit change in the conductivity of each grid element), the boundary measurement change is established. Changes in internal field distribution Linearized approximation relationship between ,in It is a sensitivity matrix of 120×1024.

[0042] The field distribution image reconstruction module constructs the following objective functional to solve the ill-conditioned inverse problem:

[0043] in, This represents the boundary voltage vector at the current moment. This is the total variational regularization term, used to preserve the edge characteristics of the combustion front; This is a priori for temporal extrapolation prediction based on historical frame data; These are the regularization weight coefficients. These are the weighting coefficients for time-series constraints.

[0044] Specifically, the total variational regularization term Adopting an isotropic TV format, It is 0.02. It is 0.05. Based on the historical reconstruction data from the first two frames and By linear extrapolation get.

[0045] The objective functional was solved using the Split-Bregman iterative algorithm, with 50 iterations. On an industrial PC equipped with an Intel Core i7-12700 CPU and 32GB of memory, the single-frame reconstruction time was ≤25ms, and the overall system response frequency was ≥30Hz, far exceeding the characteristic frequency of combustion changes (typically <10Hz), meeting the requirements for "online" identification. The output is a field distribution image at the current moment (a 32×32 grid map of complex conductivity changes), with a resolution capable of effectively distinguishing different reaction regions with intervals greater than 0.1m.

[0046] III. Region Segmentation and Feature Extraction Module The region segmentation and feature extraction module processes the reconstructed field distribution image as follows: (1) Extracting the first type of reaction region (Region A - Volatile matter release region): Extracting connected regions where the dielectric loss tangent corresponding to the imaginary part of the complex conductivity is higher than the background noise threshold. In this embodiment, the background noise threshold is set to 1.5 times the average dielectric loss tangent of the entire cross-section. For each connected region, after removing noise patches with an area smaller than 4 grid cells, the largest connected region is retained as Region A. Calculate the area of ​​this region. (in terms of grid cell count), centroid coordinates and area change rate (Obtained by dividing the difference between two consecutive frames by the frame interval).

[0047] In the typical operating condition of this embodiment, when the fuel feed rate increases sharply, the area of ​​region A... It expands from approximately 80 grid cells to 150 grid cells within 2 seconds. Reaching approximately 35 grid cells / second corresponds to the rapid precipitation stage of volatiles.

[0048] (2) Extraction of the second type of reaction zone (region B - coke combustion zone): Extraction of the real part of complex conductivity Connected components whose noise level exceeds the background noise threshold. In this embodiment, the background noise threshold is set to the entire cross-section. 2.0 times the mean. After removing noise patches with an area smaller than 4 grid cells, the largest connected component is retained as region B. Calculate the area of ​​this region. Centroid coordinates and equivalent diameter And obtain the equivalent diameter shrinkage rate. .

[0049] In the typical operating conditions of this embodiment, the equivalent diameter of region B gradually shrinks from 0.35m to 0.28m, corresponding to the coke burnout process; when the fluidizing velocity suddenly increases, the centroid of region B shifts by approximately 0.15m within 1 second, and the centroid movement rate is... It reaches 0.15 m / s.

[0050] IV. Pollutant Identification Module Based on the results of region segmentation and feature extraction, the pollutant identification module calculates three independent and complementary identification indices: (1) NO x Generation Intensity Index :

[0051] in, This represents the real part of the average conductivity of the second-type reaction region. , These are the weighting coefficients determined through a pre-calibration experiment; An increase in the value indicates that NO exports x The concentration increased. In this embodiment, a pre-calibration experiment was conducted (synchronous collection of outlet NO under stable operating conditions). x Concentration, collected from 60 groups under different combustion conditions Calculated value and export NO x The weighting coefficients are determined using measured data and least squares fitting to determine the optimal coefficients. 0.3 It is 0.7. After dimensionless conversion, the range is 0~1. When If the value remains above 0.7 for 3 consecutive seconds, the system will determine it as NO. x High generation rate. Experimental verification shows that... Rising Leading Export NO x The peak concentration lasted for 5 to 8 seconds.

[0052] (2) CO generation propensity index : Based on the centroidal movement rate of the second type of reaction zone (region B) Combined with equivalent diameter shrinkage rate Make a judgment; when Greater than the preset rate threshold and When the shrinkage rate is less than the preset shrinkage threshold, it is determined that the coke burnout rate has decreased, and a high CO emission warning signal is output.

[0053] In this embodiment, the preset rate threshold is 0.08 m / s, and the preset contraction threshold is -0.002 m / s (negative values ​​indicate contraction).

[0054] (3) Oxidation-reduction game index :

[0055] in, The area representing the overlap between the first type of reaction zone and the second type of reaction zone. The default positive number is set to 1 to prevent division by zero. A higher value indicates a greater degree of spatial overlap between the volatile matter emission zone and the coke combustion zone, and a higher probability of homogeneous NO reduction reaction. In this embodiment, when the centroid distance between region A and region B is less than 0.15m, Significantly increased, A value greater than 0.3 indicates the presence of homogeneous NO reduction reaction space conditions; when the two types of regions are separated (centroid distance greater than 0.4m), A value less than 0.05 indicates unfavorable reduction reaction conditions, and NO tends to escape.

[0056] V. Results Output and Model Update Module The results output and model update module will , Warning status and The spatial locations of Region A and Region B are packaged together to generate a visual identification result. On the monitoring interface, Region A is marked with an orange heatmap overlaid with color blocks (identified as volatile matter-NO). x High-risk area for precursors), Area B is marked with a red heat map overlaid with color blocks (identified as coke-NO). x / CO high-risk area), and mark the centroid displacement direction arrow.

[0057] Meanwhile, this module uses historical data from the outlet flue gas analyzer to measure NO every 5 minutes. x CO concentration data, weighting coefficients , Perform least-squares recursive correction. The updated coefficients are automatically applied to subsequent calculations. During the calculation, the identification model is adaptively updated as the coal type or operating conditions change.

[0058] It should be noted that, in this embodiment of the 50kW circulating fluidized bed test device, sensor arrays were arranged at three cross-sections at distances of 1.2m (inside the dense phase zone), 1.8m (lower part of the main combustion zone), and 2.5m (dilute phase zone) from the air distribution plate for comparative testing. The results show that at the 1.2m cross-section, the particle concentration is too high, resulting in severe electrode signal attenuation (signal-to-noise ratio <25dB), making it impossible to effectively reconstruct the field distribution image; at the 2.5m cross-section, the combustion reaction is basically complete, and the spatial distribution difference between region A and region B is small. (Constant > 0.6), making it difficult to identify pollutant generation trends; the spatial distribution difference between the two types of reaction zones is most significant at the 1.8m cross-section, with a signal-to-noise ratio > 40dB, resulting in the best identification sensitivity.

[0059] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. An online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing, characterized in that, include: A sensor array module is arranged on at least one cross section of the furnace of an oxygen-enriched combustion boiler for applying an excitation signal and acquiring boundary response data of the combustion medium to the excitation signal. The signal acquisition and preprocessing module is used to amplify, filter, perform analog-to-digital conversion and temperature drift correction on the boundary response data to generate standardized boundary measurement data. The field distribution image reconstruction module is used to reconstruct the internal field distribution image of the cross section by solving an ill-conditioned inverse problem based on the boundary measurement data and a pre-established sensitivity field model. The region segmentation and feature extraction module is used to segment the field distribution image into regions and extract a first type of reaction region and a second type of reaction region, respectively. The first type of reaction region is a connected region in the field distribution image where the dielectric loss tangent is higher than a preset background noise threshold, representing the volatile matter release and pyrolysis region. The second type of reaction region is a connected region in the field distribution image where the real part of the complex conductivity is higher than a preset background noise threshold, representing the coke combustion and gasification region. The module also calculates the geometric feature parameters and their dynamic change rates for the two types of reaction regions. The pollutant identification module is used to establish a pollutant identification system based on the geometric feature parameters and spatial relative positional relationship of the two types of reaction zones, including NO. x A multi-dimensional determination system of generation intensity index, CO generation tendency index and redox game index can be used to identify the main generation areas and generation intensity trends of pollutants online. The results output and model update module is used to visualize the identification results and periodically adjust the mapping weights of the multidimensional judgment system based on offline or semi-online calibration data.

2. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, The sensor array module includes N electrodes arranged circumferentially at equal intervals on the cross-section of the furnace, where N≥16; the excitation signal is a multi-band AC signal with a frequency range of 100kHz~10MHz; the boundary response data is the voltage difference vector between each electrode pair; each electrode is flush with the inner wall surface of the furnace and is isolated from the furnace wall by a corundum insulation layer; the electrodes are made of platinum-iridium alloy material protected by yttrium-stabilized zirconium oxide.

3. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, The field distribution image reconstruction module divides the cross section into... Each grid cell is used to establish boundary measurement changes. Changes in internal field distribution Linearized approximation relationship between ,in This is the sensitivity matrix. Let be the vector of complex conductivity changes of each grid cell within the cross section; the physical meaning of the internal field distribution image is the change of complex conductivity of each grid cell within the cross section, including the real part. and the virtual part The field distribution image reconstruction module reconstructs the internal field distribution image by solving the following objective functional: in, This represents the boundary voltage vector at the current moment. This is the total variational regularization term, used to preserve the edge characteristics of the combustion front; This is a priori for temporal extrapolation prediction based on historical frame data; These are the regularization weight coefficients. These are the weighting coefficients for time-series constraints.

4. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, After the region segmentation and feature extraction module extracts the first type of reaction region, it calculates the area of ​​the region. Centroid coordinates and its area change rate The area change rate It is obtained by dividing the difference between two consecutive frames by the frame interval.

5. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, After the region segmentation and feature extraction module extracts the second type of reaction region, it calculates the area of ​​the region. Centroid coordinates and its equivalent diameter shrinkage rate The equivalent diameter is calculated based on the area of ​​a circle. .

6. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, The NO x Generation Intensity Index The calculation method is as follows: in, This represents the real part of the average conductivity of the second-type reaction region. , These are the weighting coefficients determined through a pre-calibration experiment; An increase in the value indicates that NO exports x Concentrations are rising.

7. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 6, characterized in that, The result output and model update module utilizes historical NO levels measured by the outlet flue gas analyzer. x CO concentration data, adjusted at preset time intervals by the weighting coefficients. , Perform least squares recursive correction.

8. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, The CO generation tendency index The determination method is based on the centroid movement rate of the second type of reaction zone. Combined with equivalent diameter shrinkage rate Make a judgment; when Greater than the preset rate threshold and When the shrinkage rate is less than the preset shrinkage threshold, it is determined that the coke burnout rate has decreased, and a high CO emission warning signal is output.

9. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, The redox game index The calculation method is as follows: in, The area representing the overlap between the first type of reaction zone and the second type of reaction zone. Preset positive numbers; A higher value indicates a greater degree of spatial overlap between the volatile matter leaching zone and the coke combustion zone, and a higher probability of homogeneous NO reduction reaction.

10. The online identification system for pollutant generation areas in oxygen-enriched combustion based on field distribution sensing according to claim 1, characterized in that, In the sensor array module, each electrode is embedded with a thermocouple for synchronously measuring the wall temperature at each electrode mounting position; the signal acquisition and preprocessing module performs online temperature drift correction on the boundary response data based on the wall temperature.

Images