A method and system for multi-component flue gas monitoring based on spectral analysis

Abstract

This invention relates to the field of multi-component flue gas monitoring technology, specifically to a multi-component flue gas monitoring method and system based on spectral analysis. It includes the following steps: S1, sampling the flue gas at multiple points using a spectral device to obtain spectral signals from multiple angles; simultaneously illuminating the flue gas at multiple points using a composite light source device to obtain light source signals from multiple angles; by employing reference wavelength dust separation technology and a local occlusion recognition algorithm based on spatial differences, the industry problem of spectral technology being unable to distinguish between gas absorption and dust scattering is completely solved. Simultaneously, it can accurately separate light intensity attenuation caused solely by dust, identify local high-concentration dust clusters by attenuation differences between adjacent optical paths, and adopt a segmented effectiveness judgment, only eliminating invalid distance segments that are blocked, maximizing the retention of effective spectral information. This allows for stable and accurate measurements even under harsh dust concentration conditions, significantly improving anti-interference capability and environmental adaptability.

Description

Technical Field

[0001] This invention relates to the field of multi-component flue gas monitoring technology, and more specifically, to a multi-component flue gas monitoring method and system based on spectral analysis. Background Technology

[0002] Industrial flue gas emission monitoring is a core component of air pollution prevention and control, and also an important technical support for implementing environmental protection regulations and promoting green production in enterprises.

[0003] Existing spectral flue gas monitoring technologies mainly employ two working modes: single-path in-situ measurement or extraction-type preprocessing measurement. They are widely used in monitoring flue gas emission outlets of stationary pollution sources such as industrial boilers and kilns. In these scenarios, flue gas typically has the characteristics of complex composition and drastic fluctuations in operating conditions. It not only contains a variety of gaseous pollutants but also carries a large amount of solid dust particles and water vapor. Furthermore, the temperature, pressure, and flow rate of the flue gas fluctuate frequently with changes in production load. Existing operating modes cannot effectively cope with the interference of local high-concentration dust clouds. Single-path measurement cannot distinguish between gas absorption and dust scattering. When local dust clouds appear in the flue, it will cause the overall spectral baseline to drift, resulting in completely distorted measurement results. Existing baseline correction algorithms can only handle overall uniform dust attenuation and cannot identify and eliminate the influence of local shading. Secondly, single-point measurement is not representative enough. The concentration and velocity distribution of flue gas in the flue are usually uneven, especially in areas with turbulent flow fields such as bends and diameter changes, resulting in a large error in the total emission estimation. Therefore, a multi-component flue gas monitoring method and system based on spectral analysis is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-component flue gas monitoring method and system based on spectral analysis to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, one objective of this invention is to provide a method for monitoring multi-component flue gas based on spectral analysis, comprising the following steps: S1. The flue is sampled at multiple points using a spectral device to obtain spectral signals from multiple angles. Simultaneously, the flue is illuminated at multiple points using a composite light source device to obtain light source signals from multiple angles. S2. Divide the spectral signal and the light source signal into optical paths according to the coverage position. Based on the division results, divide the flue into multiple optical paths that cross and cover the entire flue. Based on the comparison between the light source signal and the corresponding spectral signal of the optical path, calculate the light intensity attenuation data of each optical path. S3. Separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path, identify local dust-blocked areas based on the light intensity attenuation data, set the dust-blocked areas as invalid distance segments, and filter out the valid distance segments accordingly. S4. Based on the effective and ineffective distance segments, establish the target component dataset of the flue gas based on the spectral signals of the effective distance segments and extract the target component parameters. S5. Based on the length of the effective distance segment and the distance to the ineffective distance segment, assign weights to each effective distance segment, and analyze the dust occlusion data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. S6. Based on the dust obstruction data obtained from the analysis in S5, the parameter range is predicted by combining the target component parameters corresponding to adjacent effective distance segments. The predicted parameter range is then compared with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

[0006] As a further improvement to this technical solution, in S1, the spectrometer and the composite light source are respectively fixedly installed on the opposite side walls of the flue, and the spectrometer and the composite light source are evenly distributed along the circumference of the flue cross section. The composite light source device is a broadband composite light source; The spectroscopic equipment is a multi-band spectral detector; A unified synchronous trigger signal is sent to all devices to control all composite light sources to light up sequentially. When any composite light source lights up, all spectral devices on the other side of the flue start collecting data and recording the transmitted light signal after passing through the flue gas. This process continues until all composite light sources have completed one lighting cycle and the corresponding spectral acquisition, thus completing the acquisition of spectral and light source signals.

[0007] As a further improvement to this technical solution, in step S2, a unique spatial location identifier is assigned to each composite light source and each spectral device, and its precise coordinates on the flue wall are recorded. Establish the correspondence between light source signals, spectral signals and spatial positions. Define the path that the light emitted by any composite light source illuminates any spectral detector as an independent optical path. Spatially encode all optical paths. Each optical path has an identifier and a corresponding spatial coverage area. All optical paths are combined to form a cross-optical path network that uniformly covers the entire flue section. Standard measurements are performed under dust-free and smoke-free flue conditions. The measurement results are used as a reference benchmark. Then, during actual flue emission measurements, the reference benchmark for each optical path is compared with the actual transmitted light intensity after passing through the flue gas. The light intensity attenuation degree of each optical path at each wavelength is calculated, and the attenuation degree of different wavelengths is arranged in order to form the light intensity attenuation data of each optical path.

[0008] As a further improvement to this technical solution, in step S3, a reference wavelength is obtained in which all target gas components to be measured will not be absorbed. Then, in the light intensity attenuation data of each optical path, the light intensity attenuation data corresponding to the reference wavelength is extracted. This data is the dust light intensity attenuation data of the optical path. Set a deviation threshold and simultaneously calculate the difference in dust light intensity attenuation data between any two adjacent optical paths; When the difference exceeds the deviation threshold, there is a local high-concentration dust cloud in the area between these two adjacent optical paths. The spatial area where the local high-concentration dust cloud is located is marked as a local dust-blocking area. The portion of each optical path that passes through a localized dust-blocked area is designated as an invalid distance segment; The segment of each optical path that does not pass through any dust-obstructed area is defined as the effective distance segment; If the difference does not exceed the deviation threshold, then no area is classified as an occluded area.

[0009] As a further improvement to this technical solution, in S4, the entire flue cross section is uniformly divided into multiple square grid units of the same size. Then, the grid units covered by the effective distance segment are marked as effective grids, and the grid units covered by the invalid distance segment are marked as invalid grids. All effective grids and invalid grids together form a complete flue grid model. For each target gas component that needs to be measured, determine its unique characteristic absorption wavelength; From the spectral signal corresponding to the effective distance segment, extract the total light intensity attenuation data under the characteristic absorption wavelength, and subtract the dust light intensity attenuation data at the corresponding position from the total light intensity attenuation data to obtain the light intensity attenuation data caused only by the absorption of the target gas component. Use the gas absorption light intensity attenuation data as the parameter of the target component at the corresponding position. Arrange the target component parameters of all valid grids according to the spatial order of the grid cells to form a dataset containing the location of each valid grid and all corresponding target component parameters, which is the target component dataset of the flue.

[0010] As a further improvement to this technical solution, in step S5, weights are assigned to each effective distance segment; The longer the effective distance segment, the higher the weight; The closer the effective distance segment is to the ineffective distance segment, the lower the weight. Taking any invalid distance segment as the center, select all surrounding valid distance segments, multiply the dust light intensity attenuation data of these valid distance segments by their respective weights, and sum and average all the weighted dust light intensity attenuation data to obtain the dust light intensity attenuation data of the invalid distance segment.

[0011] As a further improvement to this technical solution, in step S6, taking any invalid distance segment as the center, all surrounding valid distance segments are selected, and the target component parameters corresponding to these valid distance segments are multiplied by their respective weights. All weighted target component parameters are summed and averaged to obtain the predicted value of the target component parameters of the invalid distance segment. Based on the fluctuation range of historical measurement data for each grid cell, the parameter prediction range corresponding to the predicted value is determined.

[0012] As a further improvement to this technical solution, in step S6, the predicted range of target component parameters for each invalid distance segment is compared with the parameter range at the corresponding position in the target component dataset; If the prediction range falls within the parameter range of the target component dataset, the prediction result is considered valid. If the prediction range exceeds the parameter range of the target component dataset, the prediction result is deemed invalid. The spectral equipment and composite light source equipment are adjusted, and the invalid distance segment is re-acquired and recalculated. The target component parameters corresponding to all valid grids and the target component parameters corresponding to all predicted valid invalid grids are summarized to generate a distribution result of target component parameters covering all locations in the entire flue section. This distribution result is then output as the final flue gas monitoring result.

[0013] The second objective of this invention is to provide a multi-component flue gas monitoring system based on spectral analysis, including any one of the above-mentioned methods for multi-component flue gas monitoring based on spectral analysis, comprising a data calculation module, a dataset modeling module, and a monitoring output module; The data calculation module is used to sample the flue at multiple points using a spectral device to obtain spectral signals from multiple angles of the flue, and simultaneously illuminate the flue at multiple points using a composite light source device to obtain light source signals from multiple angles of the flue. The spectral signals and light source signals are divided into optical paths according to the coverage position. Based on the division result, the flue is divided into multiple optical paths that intersect and cover the entire flue. Based on the comparison between the light source signals and the corresponding spectral signals of the optical paths, the light intensity attenuation data of each optical path is calculated. The dataset modeling module is used to separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path, identify local dust-covered areas based on the light intensity attenuation data, set the dust-covered areas as invalid distance segments, and filter out the effective distance segments accordingly. Based on the effective distance segments and invalid distance segments, a target component dataset of the flue is established according to the flue model. The spectral signals of the effective distance segments are selected to extract the target component parameters, and the target component dataset of the flue is established based on the target component parameters. The monitoring output module is used to assign weights to each effective distance segment based on its length and distance from the ineffective distance segment. It analyzes the dust obstruction data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. Based on the dust obstruction data obtained from the analysis, it predicts the parameter range by combining the target component parameters corresponding to adjacent effective distance segments. It then compares the predicted parameter range with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. A multi-component flue gas monitoring method and system based on spectral analysis, which completely solves the industry problem that spectral technology cannot distinguish between gas absorption and dust scattering by using reference wavelength dust separation technology and a local occlusion recognition algorithm based on spatial differences. At the same time, it can accurately separate light intensity attenuation caused only by dust, identify local high-concentration dust clusters by the attenuation difference of adjacent optical paths, and adopt a segmented effectiveness judgment to remove only the invalid distance segments that are blocked, so as to retain the effective spectral information to the maximum extent. This enables stable and accurate measurement even under harsh working conditions with high dust concentration, and greatly improves anti-interference ability and environmental adaptability.

[0015] 2. A multi-component flue gas monitoring method and system based on spectral analysis, which forms a cross-optical network covering the entire cross section of the flue by arranging multiple sets of light sources and detectors on both sides of the flue. The flue is discretized into multiple grid units, and the pollutant concentration distribution of the entire cross section is reconstructed. This can truly reflect the spatial distribution characteristics of flue gas in the flue. The total emissions calculated based on the full cross section distribution data can provide accurate and reliable measurement basis for environmental law enforcement, carbon emission accounting and pollution rights trading.

[0016] 3. A multi-component flue gas monitoring method and system based on spectral analysis, which is installed in situ on the flue wall, eliminating the need for complex sampling and pretreatment systems, completely avoiding the adsorption loss of soluble gases and gaseous mercury, and significantly improving the detection accuracy of trace components. At the same time, through a two-factor adaptive weighting algorithm and a prediction result self-verification mechanism, it can scientifically calculate the parameters of the ineffective shading area and perform closed-loop re-acquisition for unreliable prediction results to ensure that the final output is complete and accurate full-section monitoring results. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of a multi-component flue gas monitoring method based on spectral analysis according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, one of the objectives of this invention is to provide a method for monitoring multi-component flue gas based on spectral analysis, comprising the following steps: S1. The flue is sampled at multiple points using a spectral device to obtain spectral signals from multiple angles. Simultaneously, the flue is illuminated at multiple points using a composite light source device to obtain light source signals from multiple angles. In S1, the spectrometer and the composite light source are fixedly installed on opposite side walls of the flue, and the spectrometer and the composite light source are evenly distributed along the circumference of the flue cross section. Multiple broadband composite light sources are uniformly arranged along the circumference of the flue section. The center-to-center distance between two adjacent light sources is controlled between 0.3 and 0.5 meters. For rectangular flues, the light sources are arranged at equal intervals in the horizontal or vertical direction. For circular flues, the light sources are arranged at equal angles in the circumferential direction. The resulting optical network can uniformly cover the entire flue section without any detection blind spots greater than 0.5 meters. The same number of multi-band spectral detectors are installed at positions corresponding to the light sources to ensure that the collimated beam emitted by each light source can illuminate all detectors on the opposite side without obstruction. The optical windows of all light sources and detectors are installed flush with the inner wall of the flue. The composite light source device is a broadband composite light source; It integrates one ultraviolet deuterium lamp, two near-infrared DFB lasers and three mid-infrared QCL lasers, and can simultaneously emit continuous light signals covering three bands: 200~400nm ultraviolet, 1.5~2.0μm near-infrared and 3~12μm mid-infrared, covering the characteristic absorption peaks of all conventional flue gas pollutants such as sulfur dioxide, nitrogen oxides, carbon monoxide, carbon dioxide, ammonia and hydrogen chloride in one go; The spectroscopic equipment is a multi-band spectral detector; It integrates an ultraviolet photodiode, a near-infrared InGaAs detector, and a mid-infrared MCT detector, each corresponding to one of the three light source bands. It can simultaneously receive and convert the transmitted light signal across the entire band into an electrical signal, with a spectral resolution better than 0.1 nm, ensuring that the narrowband absorption peaks of different gases can be clearly distinguished. A unified synchronous trigger signal is sent to all devices to control all composite light sources to light up sequentially. When any composite light source lights up, all spectral devices on the other side of the flue start collecting data and recording the transmitted light signal after passing through the flue gas. This process continues until all composite light sources have completed one lighting cycle and the corresponding spectral acquisition, thus completing the acquisition of spectral and light source signals.

[0020] A unified synchronous trigger signal is generated by a high-precision FPGA chip, and the time synchronization accuracy of the signal is controlled within 1 microsecond. All light sources and detectors are connected to the FPGA chip through a dedicated synchronous signal line to receive the same trigger signal, ensuring that the actions of all devices are completely synchronized in time. S2. Divide the spectral signal and the light source signal into optical paths according to the coverage position. Based on the division results, divide the flue into multiple optical paths that cross and cover the entire flue. Based on the comparison between the light source signal and the corresponding spectral signal of the optical path, calculate the light intensity attenuation data of each optical path. In S2, a unique spatial location identifier is assigned to each composite light source and each spectral device, and its precise coordinates on the flue wall are recorded. First, establish a unified two-dimensional rectangular coordinate system on the selected flue measurement section. For a rectangular flue, set the origin of the coordinate system at the lower left corner of the section, with the X-axis extending to the right horizontally and the Y-axis extending upward vertically. For a circular flue, set the origin of the coordinate system at the center of the section, with the X-axis extending to the right horizontally and the Y-axis extending upward vertically. Using a laser rangefinder or total station with an accuracy of no less than ±1 mm, measure the precise coordinates of the optical center of each composite light source and each multi-band spectral detector in the above coordinate system. Assign a unique device identifier to each composite light source, using letters and numbers, starting with L to represent the light source followed by a two-digit serial number. Assign a unique device identifier to each spectral detector, starting with D to represent the detector followed by a two-digit serial number. Bind the unique identifiers of all devices to their corresponding coordinate values ​​to generate a device coordinate mapping table. Establish the correspondence between light source signals, spectral signals and spatial positions. Define the path that the light emitted by any composite light source illuminates any spectral detector as an independent optical path. Spatially encode all optical paths. Each optical path has an identifier and a corresponding spatial coverage area. All optical paths are combined to form a cross-optical path network that uniformly covers the entire flue section. The straight path along which a collimated beam emitted from any composite light source illuminates any spectral detector is defined as an independent optical path. The spatial coverage of each optical path is determined by the diameter of the beam. At the same time, according to the rule of matching the light source identifier with the detector identifier, a unique spatial code is assigned to each independent optical path. For example, the optical path from the 1st light source to the 5th detector is coded as L01-D05. A mapping table between the optical path code and the corresponding light source and detector coordinates is established. The entire flue section is divided into square grid cells with a side length of 0.2 meters. The number of optical paths passing through each grid cell is calculated. The deviation of the number of optical paths in all grid cells is required to be no more than 20%. If the number of optical paths in a grid cell is less than 80% of the average, the installation angle of the adjacent light source or detector is finely adjusted to increase the number of optical paths passing through that grid cell until the uniformity of optical path coverage of the entire flue section meets the requirements.

[0021] Standard measurements are performed under dust-free and smoke-free flue conditions. The measurement results are used as a reference benchmark. Then, during actual flue emission measurements, the reference benchmark for each optical path is compared with the actual transmitted light intensity after passing through the flue gas. The light intensity attenuation degree of each optical path at each wavelength is calculated, and the attenuation degree of different wavelengths is arranged in order to form the light intensity attenuation data of each optical path.

[0022] When the flue is in a clean air state free of dust and smoke, the transmitted light intensity of each optical path at all measurement wavelengths is measured and recorded according to the acquisition process of S1. This data is used as the full-band reference light intensity data of the optical path. Once this data is calibrated, it will not be changed unless the equipment is under major repair or replacement. In the actual flue gas emission measurement process, for each optical path, the actual transmitted light intensity at each wavelength is compared with the corresponding reference light intensity one by one, and the light intensity attenuation degree of the optical path at each wavelength is calculated. The greater the light intensity attenuation degree, the stronger the absorption or scattering effect of the flue gas or dust on the path on that wavelength. The calculated wavelength-by-wavelength light intensity attenuation data is smoothed using a 5-point moving average filtering method to remove high-frequency interference from detector noise and stray light, while preserving the details of the gas characteristic absorption peaks. The light intensity attenuation of all wavelengths is arranged in ascending order of wavelength to form a continuous curve, which is the standardized light intensity attenuation data of the optical path.

[0023] S3. Separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path, identify local dust-blocked areas based on the light intensity attenuation data, set the dust-blocked areas as invalid distance segments, and filter out the valid distance segments accordingly. In S3, a reference wavelength is obtained in which all target gas components to be measured will not be absorbed. Then, the light intensity attenuation data corresponding to the reference wavelength is extracted from the light intensity attenuation data of each optical path. This data is the dust light intensity attenuation data of the optical path. All target gas components that need to be measured, including sulfur dioxide, nitrogen oxides, carbon monoxide, carbon dioxide, ammonia, hydrogen chloride, hydrogen fluoride, etc., should have their standard absorption spectra collected and summarized in the full band of ultraviolet (200-400 nm), near-infrared (1.5-2.0 μm), and mid-infrared (3-12 μm). Within the aforementioned full-band range, a wavelength range was identified where the absorption spectra of all target gases were blank. This range required that the absorption coefficients of all gases within it be below the detection limit; that is, at this wavelength, even if the gas concentration reaches the maximum permissible emission concentration, there would be no measurable attenuation of light intensity. A reference wavelength in the mid-infrared band was preferentially selected because the dust scattering characteristics in the mid-infrared band are more stable and less affected by changes in dust particle size and composition. Finally, a center wavelength and a bandwidth of ±10 nm were determined as the standard reference band. Introduce a standard gas mixture of all target gases at the highest concentration into the calibration chamber and measure the light intensity attenuation at that wavelength. If the attenuation is below the system detection limit, the reference wavelength is confirmed to be valid; otherwise, re-screen until a suitable reference wavelength is found.

[0024] Set a deviation threshold and simultaneously calculate the difference in dust light intensity attenuation data between any two adjacent optical paths; If the minimum vertical distance between two optical paths within the flue section is less than 0.3 meters, or the distance between the intersection points of two optical paths is less than 0.5 meters, they are determined to be adjacent optical paths. According to this rule, a corresponding list of adjacent optical paths is generated for each optical path. Calculate the absolute value of the difference between the dust intensity attenuation data of each optical path and all its adjacent optical paths. When the dust in the flue is uniformly distributed, the difference in dust attenuation between adjacent optical paths will be very small. Only when there are local high-concentration dust clumps will the attenuation of the optical path passing through the dust clump increase significantly, while the attenuation of adjacent optical paths does not change much, thus forming a significant difference. Under normal production conditions and uniform dust distribution, dust attenuation data was continuously collected for 1 hour, and the average and standard deviation of the differences between all adjacent optical paths were calculated. The deviation threshold was set as the average value plus 3 times the standard deviation. For any pair of adjacent optical paths, if the difference in dust attenuation exceeds the set deviation threshold, it is determined that there is a local high-concentration dust cloud in the area between the two adjacent optical paths. The spatial area where the dust cloud is located is marked with the smallest bounding rectangle. The boundary of the rectangle is determined by the position and difference of the two adjacent optical paths and is marked as a local dust-blocked area. When the difference exceeds the deviation threshold, there is a local high-concentration dust cloud in the area between these two adjacent optical paths. The spatial area where the local high-concentration dust cloud is located is marked as a local dust-blocking area. The portion of each optical path that passes through a localized dust-blocked area is designated as an invalid distance segment; The segment of each optical path that does not pass through any dust-obstructed area is defined as the effective distance segment; If the difference does not exceed the deviation threshold, then no area is classified as an occluded area.

[0025] S4. Based on the effective and ineffective distance segments, establish the target component dataset of the flue gas based on the spectral signals of the effective distance segments and extract the target component parameters. In S4, the entire flue section is uniformly divided into multiple square grid cells of the same size. Then, the grid cells covered by the effective distance segment are marked as effective grids, and the grid cells covered by the ineffective distance segment are marked as ineffective grids. All effective grids and ineffective grids together form a complete flue grid model. For small and medium-sized flues with a side length of less than 3 meters, a grid with a side length of 0.2 meters is used; for large flues with a side length of more than 3 meters, a grid with a side length of 0.3 to 0.5 meters is used. The grid size must be less than half of the optical path spacing to ensure that each grid cell is traversed by at least two optical paths. For rectangular flues, starting from the origin of the coordinate system, the grid is divided sequentially along the X and Y axes according to the set grid side lengths to form a regular square grid array. For circular flues, the same square grid division is used. For incomplete grids located at the circular boundary, as long as their area exceeds 50% of the area of ​​the complete grid, they are considered as valid grid units to participate in the calculation. When a segment of an optical path passes through a grid cell and the length of the pass exceeds 30% of the diagonal length of that grid cell, it is determined that the optical path segment covers that grid cell. Traverse all valid distance segments of all valid optical paths, and mark all grid cells covered by valid distance segments as valid grids according to the coverage rule. Traverse all invalid distance segments of all optical paths, and mark all grid cells covered only by invalid distance segments and not covered by any valid distance segments as invalid grids.

[0026] For each target gas component that needs to be measured, determine its unique characteristic absorption wavelength; For each target gas component that needs to be measured, characteristic absorption peaks with high absorption intensity and clear spectral lines are screened from its standard absorption spectrum database. The fundamental absorption peak located in the mid-infrared band is preferred, as its absorption intensity is 1 to 2 orders of magnitude higher than that of the overtone absorption peak in the near-infrared band, which can achieve higher detection sensitivity. Analyze the spectral environment around the selected characteristic absorption peak to ensure that there are no strong absorption peaks of other target gases in this wavelength range, and minimize cross-interference. If there is slight overlap, select the wavelength point with the lowest overlap. From the spectral signal corresponding to the effective distance segment, extract the total light intensity attenuation data under the characteristic absorption wavelength, and subtract the dust light intensity attenuation data at the corresponding position from the total light intensity attenuation data to obtain the light intensity attenuation data caused only by the absorption of the target gas component. Use the gas absorption light intensity attenuation data as the parameter of the target component at the corresponding position. For each target gas component, its corresponding characteristic absorption wavelength is extracted. All effective optical paths are traversed, and the total light intensity attenuation data at the characteristic wavelength is extracted from the full-band light intensity attenuation data of each optical path. At the same time, according to the grid cells through which the optical path passes, the dust light intensity attenuation data of the optical path is allocated to each grid cell according to the length ratio. That is, the dust attenuation amount allocated to a certain grid cell is equal to the total dust attenuation amount of the optical path multiplied by the proportion of the length of the optical path passing through the grid to the total length of the optical path. For each effective grid cell, the total light intensity attenuation data of all optical paths passing through that grid at the characteristic wavelength is subtracted from the dust attenuation data allocated to that grid for the corresponding optical path, yielding the pure gas absorption attenuation data for each optical path at that grid. Then, for the same grid cell, a weighted average is calculated for the pure gas absorption attenuation data obtained from all optical paths passing through that grid, with the weight proportional to the effective length of the optical path. The final average value is the target component parameter for that grid cell, as shown in the following formula: ; in, Assign the dust attenuation amount to the k-th grid cell for the optical path Lij. This represents the total dust intensity attenuation in the optical path Lij. Let Lij be the length of the light path passing through the k-th grid cell. The total length of the optical path Lij; ; in, Let be the pure absorption attenuation of the i-th target gas in the k-th grid cell. For the optical path Lij at the characteristic wavelength of the i-th gas The total light intensity attenuation. This is the set of all valid optical paths passing through the k-th grid cell; Arrange the target component parameters of all valid grids according to the spatial order of the grid cells to form a dataset containing the location of each valid grid and all corresponding target component parameters, which is the target component dataset of the flue.

[0027] Following the order of grid numbers, starting from the first row and first column, all valid grid cells are traversed sequentially. For each valid grid cell, its grid number, center coordinates, and parameter values ​​of all target gas components are summarized. All parameter values ​​are validated for reasonableness, and outliers that clearly exceed the normal range are removed. For outliers, the average value of adjacent grid cells is used as the replacement, generating a structured two-dimensional data table. Each row corresponds to a grid cell, and each column corresponds to a target gas component. The table header includes the grid number, X coordinate, Y coordinate, and the names of all target gases.

[0028] S5. Based on the length of the effective distance segment and the distance to the ineffective distance segment, assign weights to each effective distance segment, and analyze the dust occlusion data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. Traverse all valid optical paths generated in step S3, and break down each optical path into all valid distance segments. For each valid distance segment, extract its complete basic parameters, including the starting point coordinates, ending point coordinates, actual geometric length, optical path number, and corresponding dust light intensity attenuation data. In S5, weights are assigned to each effective distance segment; The longer the effective distance segment, the higher the weight; The longer the effective range segment, the more effective optical information it contains, the higher the data reliability, and the higher the corresponding weight. The lengths of all effective range segments are normalized, and the longest effective range segment is used as the benchmark with its length weight set to 1. The length weights of the remaining effective range segments are calculated according to the ratio of their length to the longest length. The closer the effective distance segment is to the ineffective distance segment, the lower the weight. The closer the effective distance segment is to the ineffective distance segment, the greater the influence of local dust clusters, the lower the data reliability, and the lower the corresponding weight. The distance between all effective distance segments and the target ineffective distance segment is normalized. The farthest effective distance segment is used as the benchmark, and its distance weight is set to 1. The distance weight of the remaining effective distance segments is calculated according to the ratio of its distance to the farthest distance. Multiply the length weight of each effective distance segment by the distance weight to obtain the initial total weight of that effective distance segment. At the same time, sum the initial total weights of all effective distance segments involved in the calculation, and then divide each initial total weight by the sum of the weights to obtain the final normalized total weight. The sum of the final total weights of all effective distance segments equals 1. Taking any invalid distance segment as the center, select all surrounding valid distance segments, multiply the dust light intensity attenuation data of these valid distance segments by their respective weights, and sum and average all the weighted dust light intensity attenuation data to obtain the dust light intensity attenuation data of the invalid distance segment.

[0029] Using the midpoint of any invalid distance segment as the center, define a circular local neighborhood range with a radius of 3 times the side length of the grid cell. Traverse all valid distance segments, calculate the straight-line distance between the center point of each valid distance segment and the center point of the target invalid distance segment, and filter out all valid distance segments whose center points are within the local neighborhood range to form the set of estimated valid segments for the invalid distance segment. From the effective distance segment parameter database, dust intensity attenuation data corresponding to all neighboring effective distance segments are extracted. The dust intensity attenuation data for each effective distance segment is multiplied by its corresponding final normalized total weight. Simultaneously, all weighted dust intensity attenuation data are summed to obtain the estimated dust intensity attenuation value for that ineffective distance segment. The final dust intensity attenuation data for the ineffective distance segment is then bound to the corresponding ineffective distance segment identifier and spatial location to generate a complete full-section dust attenuation dataset. The formula is as follows: ; in, This is the estimated value of dust light intensity attenuation for the nth invalid distance segment. This represents the final normalized total weight of the m-th valid distance segment relative to the n-th invalid distance segment. This represents the measured value of dust light intensity attenuation in the m-th effective distance segment.

[0030] S6. Based on the dust obstruction data obtained from the analysis in S5, the parameter range is predicted by combining the target component parameters corresponding to adjacent effective distance segments. The predicted parameter range is then compared with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

[0031] In S6, taking any invalid distance segment as the center, all surrounding valid distance segments are selected, and the target component parameters corresponding to these valid distance segments are multiplied by their respective weights. The weighted target component parameters are summed and averaged to obtain the predicted value of the target component parameters of the invalid distance segment. Based on the fluctuation range of historical measurement data for each grid cell, the parameter prediction range corresponding to the predicted value is determined.

[0032] Extract all valid historical measurement data of the invalid grid cell under the same production conditions in the past 24 hours. The same production conditions are determined by parameters such as boiler load, fuel type, and desulfurization and denitrification system operation status. Calculate the average value and standard deviation of these historical data. Take the parameter prediction value as the center and fluctuate up and down by 3 times the standard deviation as the confidence range of the prediction value. If the grid cell does not have enough historical data, the fluctuation range of historical data from adjacent valid grid cells is used as a substitute; In S6, the predicted range of target component parameters for each invalid distance segment is compared with the parameter range at the corresponding location in the target component dataset; If the prediction range falls within the parameter range of the target component dataset, the prediction result is considered valid. If the prediction range exceeds the parameter range of the target component dataset, the prediction result is deemed invalid. The spectral equipment and composite light source equipment are adjusted, and the invalid distance segment is re-acquired and recalculated. Adjust the illumination angle of the composite light source in the corresponding area to allow more light paths to pass through the invalid grid cell. Perform a local rapid acquisition of the invalid area, only illuminate the light source that can illuminate the area, and only collect data from the detectors that can receive the signals from these light sources. Use the newly acquired data to recalculate the optical path validity and gas parameters of the area until the prediction result is determined to be valid, or the newly acquired data of the area updates it to the valid range segment. If the prediction result is still invalid after three consecutive re-collections, the system will issue an early warning signal, indicating that there is severe occlusion in the area and manual inspection is required. The target component parameters corresponding to all valid grids and the target component parameters corresponding to all predicted valid invalid grids are summarized to generate a distribution result of target component parameters covering all locations in the entire flue section. This distribution result is then output as the final flue gas monitoring result.

[0033] The target component parameters of all effective grid cells obtained from S4 are summarized with the target component parameters of all predicted effective invalid grid cells. According to the spatial position order of the grid cells, all parameter values ​​are arranged into a two-dimensional matrix corresponding to the flue grid model, generating a target component parameter distribution cloud map containing all positions of the entire flue cross section, which intuitively shows the spatial distribution of pollutants in the flue.

[0034] The second objective of this invention is to provide a multi-component flue gas monitoring system based on spectral analysis, including a multi-component flue gas monitoring method based on spectral analysis, comprising a data calculation module, a dataset modeling module, and a monitoring output module. The data calculation module is used to sample the flue at multiple points using a spectral device to obtain spectral signals from multiple angles of the flue. Simultaneously, it uses a composite light source device to illuminate the flue at multiple points to obtain light source signals from multiple angles of the flue. The spectral signals and light source signals are divided into optical paths according to the coverage position. Based on the division results, the flue is divided into multiple optical paths that intersect and cover the entire flue. Based on the comparison between the light source signals and the corresponding spectral signals of the optical paths, the light intensity attenuation data of each optical path is calculated. The dataset modeling module is used to separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path. Based on the light intensity attenuation data, it identifies local dust-covered areas, sets the dust-covered areas as invalid distance segments, and selects effective distance segments accordingly. Based on the effective and invalid distance segments, it establishes a target component dataset for the flue based on the flue model, selects the spectral signals of the effective distance segments to extract target component parameters, and establishes the target component dataset for the flue based on the target component parameters. The monitoring output module assigns weights to each effective distance segment based on its length and distance from the ineffective distance segment. It then analyzes the dust obstruction data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. Based on the dust obstruction data obtained from the analysis, it predicts the parameter range by combining the target component parameters corresponding to adjacent effective distance segments. Finally, it compares the predicted parameter range with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

[0035] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method of multi-component flue gas monitoring based on spectroscopic analysis, characterized in that: Includes the following steps: S1. The flue is sampled at multiple points using a spectral device to obtain spectral signals from multiple angles. Simultaneously, the flue is illuminated at multiple points using a composite light source device to obtain light source signals from multiple angles. S2. Divide the spectral signal and the light source signal into optical paths according to the coverage position. Based on the division results, divide the flue into multiple optical paths that cross and cover the entire flue. Based on the comparison between the light source signal and the corresponding spectral signal of the optical path, calculate the light intensity attenuation data of each optical path. S3. Separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path, identify local dust-blocked areas based on the light intensity attenuation data, set the dust-blocked areas as invalid distance segments, and filter out the valid distance segments accordingly. S4. Based on the effective and ineffective distance segments, establish the target component dataset of the flue gas based on the spectral signals of the effective distance segments and extract the target component parameters. S5. Based on the length of the effective distance segment and the distance to the ineffective distance segment, assign weights to each effective distance segment, and analyze the dust occlusion data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. S6. Based on the dust obstruction data obtained from the analysis in S5, the parameter range is predicted by combining the target component parameters corresponding to adjacent effective distance segments. The predicted parameter range is then compared with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

2. A multi-component flue gas monitoring method based on spectroscopic analysis according to claim 1, characterized in that: In S1, the spectrometer and the composite light source are fixedly installed on opposite side walls of the flue, and the spectrometer and the composite light source are evenly distributed along the circumference of the flue cross section. The composite light source device is a broadband composite light source; The spectroscopic equipment is a multi-band spectral detector; A unified synchronous trigger signal is sent to all devices to control all composite light sources to light up sequentially. When any composite light source lights up, all spectral devices on the other side of the flue start collecting data and recording the transmitted light signal after passing through the flue gas. This process continues until all composite light sources have completed one lighting cycle and the corresponding spectral acquisition, thus completing the acquisition of spectral and light source signals.

3. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In S2, a unique spatial location identifier is assigned to each composite light source and each spectral device, and its precise coordinates on the flue wall are recorded. Establish the correspondence between light source signals, spectral signals and spatial positions. Define the path that the light emitted by any composite light source illuminates any spectral detector as an independent optical path. Spatially encode all optical paths. Each optical path has an identifier and a corresponding spatial coverage area. All optical paths are combined to form a cross-optical path network that uniformly covers the entire flue section. Standard measurements are performed under dust-free and smoke-free flue conditions. The measurement results are used as a reference benchmark. Then, during actual flue emission measurements, the reference benchmark for each optical path is compared with the actual transmitted light intensity after passing through the flue gas. The light intensity attenuation degree of each optical path at each wavelength is calculated, and the attenuation degree of different wavelengths is arranged in order to form the light intensity attenuation data of each optical path.

4. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In step S3, a reference wavelength is obtained in which all target gas components to be measured will not be absorbed. Then, the light intensity attenuation data corresponding to the reference wavelength is extracted from the light intensity attenuation data of each optical path. This data is the dust light intensity attenuation data of the optical path. Set a deviation threshold and simultaneously calculate the difference in dust light intensity attenuation data between any two adjacent optical paths; When the difference exceeds the deviation threshold, there is a local high-concentration dust cloud in the area between these two adjacent optical paths. The spatial area where the local high-concentration dust cloud is located is marked as a local dust-blocking area. The portion of each optical path that passes through a localized dust-blocked area is designated as an invalid distance segment; The segment of each optical path that does not pass through any dust-obstructed area is defined as the effective distance segment; If the difference does not exceed the deviation threshold, then no area is classified as an occluded area.

5. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In S4, the entire flue section is uniformly divided into multiple square grid cells of the same size. Then, the grid cells covered by the effective distance segment are marked as effective grids, and the grid cells covered by the ineffective distance segment are marked as ineffective grids. All effective grids and ineffective grids together form a complete flue grid model. For each target gas component that needs to be measured, determine its unique characteristic absorption wavelength; From the spectral signal corresponding to the effective distance segment, extract the total light intensity attenuation data under the characteristic absorption wavelength, and subtract the dust light intensity attenuation data at the corresponding position from the total light intensity attenuation data to obtain the light intensity attenuation data caused only by the absorption of the target gas component. Use the gas absorption light intensity attenuation data as the parameter of the target component at the corresponding position. Arrange the target component parameters of all valid grids according to the spatial order of the grid cells to form a dataset containing the location of each valid grid and all corresponding target component parameters, which is the target component dataset of the flue.

6. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In S5, weights are assigned to each effective distance segment; The longer the effective distance segment, the higher the weight; The closer the effective distance segment is to the ineffective distance segment, the lower the weight. Taking any invalid distance segment as the center, select all surrounding valid distance segments, multiply the dust light intensity attenuation data of these valid distance segments by their respective weights, and sum and average all the weighted dust light intensity attenuation data to obtain the dust light intensity attenuation data of the invalid distance segment.

7. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In S6, taking any invalid distance segment as the center, all surrounding valid distance segments are selected, and the target component parameters corresponding to these valid distance segments are multiplied by their respective weights. All weighted target component parameters are summed and averaged to obtain the predicted value of the target component parameters of the invalid distance segment. Based on the fluctuation range of historical measurement data for each grid cell, the parameter prediction range corresponding to the predicted value is determined.

8. The method for monitoring multi-component flue gas based on spectral analysis according to claim 1, characterized in that: In step S6, the predicted range of target component parameters for each invalid distance segment is compared with the parameter range at the corresponding position in the target component dataset; If the prediction range falls within the parameter range of the target component dataset, the prediction result is considered valid. If the prediction range exceeds the parameter range of the target component dataset, the prediction result is deemed invalid. The spectral equipment and composite light source equipment are adjusted, and the invalid distance segment is re-acquired and recalculated. The target component parameters corresponding to all valid grids and the target component parameters corresponding to all predicted valid invalid grids are summarized to generate a distribution result of target component parameters covering all locations in the entire flue section. This distribution result is then output as the final flue gas monitoring result.

9. A multi-component flue gas monitoring system based on spectral analysis, used to implement the multi-component flue gas monitoring method based on spectral analysis as described in any one of claims 1-8, characterized in that: It includes a data calculation module, a dataset modeling module, and a monitoring output module; The data calculation module is used to sample the flue at multiple points using a spectral device to obtain spectral signals from multiple angles of the flue, and simultaneously illuminate the flue at multiple points using a composite light source device to obtain light source signals from multiple angles of the flue. The spectral signals and light source signals are divided into optical paths according to the coverage position. Based on the division result, the flue is divided into multiple optical paths that intersect and cover the entire flue. Based on the comparison between the light source signals and the corresponding spectral signals of the optical paths, the light intensity attenuation data of each optical path is calculated. The dataset modeling module is used to separate the light intensity attenuation data caused only by dust scattering from the total light intensity attenuation curve of each optical path, identify local dust-covered areas based on the light intensity attenuation data, set the dust-covered areas as invalid distance segments, and filter out the effective distance segments accordingly. Based on the effective distance segments and invalid distance segments, a target component dataset of the flue is established according to the flue model. The spectral signals of the effective distance segments are selected to extract the target component parameters, and the target component dataset of the flue is established based on the target component parameters. The monitoring output module is used to assign weights to each effective distance segment based on its length and distance from the ineffective distance segment. It analyzes the dust obstruction data of the ineffective distance segment based on the weights of the effective distance segments and the corresponding light intensity attenuation data. Based on the dust obstruction data obtained from the analysis, it predicts the parameter range by combining the target component parameters corresponding to adjacent effective distance segments. It then compares the predicted parameter range with the target component dataset. If the predicted parameter range does not exceed the target component dataset, the parameter range of the target component dataset is used as the monitoring result.

Images