A flame element on-line measurement system and method

Abstract

This invention belongs to the field of laser spectral analysis and combustion process diagnosis technology, and relates to an online flame element measurement system and method. The system includes a high-energy pulsed laser, an optical focusing and collection subsystem, a measurement cavity, a fiber optic probe, a grating spectrometer, an ICCD camera, a digital delay generator, and an information processing unit. The measurement cavity contains a combustion flame. The high-energy pulsed laser emits a laser beam, which is guided into the measurement cavity through the optical focusing and collection subsystem to generate high-temperature plasma. One end of the fiber optic probe collects the plasma signal, and the other end is connected to the grating spectrometer. The grating spectrometer is connected to the ICCD camera to convert the collected optical signal into a spectral signal to be analyzed. The digital delay generator is electrically connected to both the high-energy pulsed laser and the ICCD camera. The information processing unit is electrically connected to the ICCD camera, the grating spectrometer, and the digital delay generator to receive the spectral signals and determine the flame elements and their content. This invention improves the stability and accuracy of online flame element measurement.

Description

Technical Field

[0001] This invention belongs to the field of laser spectroscopy analysis and combustion process diagnosis technology, specifically relating to an online flame element measurement system and method. Background Technology

[0002] Fuels contain alkali metals, alkaline earth metals, and heavy metals. Although their content is low, their impact on combustion systems is extremely profound and complex, directly affecting equipment operational safety, service life, and environmental pollution. Alkali metals are mainly potassium (K) and sodium (Na), commonly found in biomass, some lignite, and waste. Their presence leads to slagging, ash accumulation, corrosion, and aerosol and particulate emissions, causing reduced thermal efficiency, pipe rupture, shortened equipment life, and air pollution. Alkaline earth elements mainly include calcium (Ca) and magnesium (Mg), commonly found in coal, biomass, and waste. During combustion, alkaline earth elements affect ash fusion characteristics and participate in equipment corrosion. Heavy metals mainly include lead (Pb), mercury (Hg), arsenic (As), and chromium (Cr), primarily found in coal, waste, and some biomass. During combustion, they cause severe environmental pollution (atmosphere, fly ash, bottom ash), harming human health and affecting catalytic systems such as SCR.

[0003] Currently, most methods for measuring elements in combustion flames employ non-contact measurement techniques, including tunable diode laser absorption spectroscopy, laser-induced fragment fluorescence, laser-induced planar fluorescence, and laser-induced breakdown spectroscopy. Non-contact laser spectroscopy and optical methods have become the mainstream approaches for studying and diagnosing combustion processes due to their advantages such as high spatiotemporal resolution and non-interference with the flow field.

[0004] Laser-induced breakdown spectroscopy (LIBS) is an atomic emission spectroscopic analysis technique that uses a high-energy laser to focus and break down a sample to generate plasma. The characteristic elemental spectra emitted during plasma cooling are then analyzed to achieve qualitative and quantitative analysis of the sample's composition. LIBS offers advantages such as requiring no complex sample preparation, remote online detection, and simultaneous multi-element analysis. Applying LIBS to combustion flame diagnostics allows for direct analysis of the concentrations of alkali metals, alkaline earth metals, and heavy metals in the flame, which is of great value for studying combustion mechanisms, optimizing combustion efficiency, early warning of boiler slagging and corrosion, and monitoring pollutant emissions.

[0005] However, directly applying traditional LIBS technology to online flame monitoring environments faces significant challenges: Because flames have strong velocity fluctuations and compositional pulsations, laser-induced plasmas experience severe spatial and temporal jitters, resulting in poor repeatability of spectral signals. This leads to low stability and accuracy in online measurement of flame elements, making it difficult to meet the needs of online monitoring in industrial settings. Summary of the Invention

[0006] In view of this, the present invention provides an online flame element measurement system and method to solve the existing technical problems.

[0007] The technical solution of this invention is: An online method for measuring elements in a flame includes a high-energy pulsed laser, an optical focusing and collection subsystem, a measurement cavity, a fiber optic probe, a grating spectrometer, an ICCD camera, a digital delay generator, and an information processing unit, wherein: The measuring chamber is used to contain the combustion flame, and the measuring chamber has an adjustable air inlet window and an exhaust window; The high-energy pulsed laser is used to emit a laser beam; The optical focusing and collecting subsystem is used to guide the laser beam emitted by the high-energy pulsed laser into the measurement cavity, focus and break down the burning flame in the measurement cavity to generate high-temperature plasma, and simultaneously collect the flame background radiation light signal and the plasma light emission signal generated by laser-induced breakdown. One end of the fiber optic probe is used to collect plasma signals, and the other end is connected to a grating spectrometer. The grating spectrometer is connected to an ICCD camera to perform photoelectric conversion and spectral resolution on the collected optical signals to obtain the spectral signals for analysis. The digital delay generator is electrically connected to the high-energy pulsed laser and the ICCD camera respectively, and is used to achieve high-precision timing synchronization between laser pulse emission and spectral acquisition gate width; The information processing unit is electrically connected to the ICCD camera, grating spectrometer and digital delay generator respectively, and is used to control the system and receive spectral signals and convert them into the concentration of the target element.

[0008] Furthermore, the optical focusing and collection subsystem includes an optical mirror, a filter, a first plano-convex lens, and a second plano-convex lens. The optical mirror, filter, and first plano-convex lens are arranged on one optical axis. The second plano-convex lens is arranged at a 90-degree angle to the first plano-convex lens. It is used to simultaneously collect the background radiation light signal of the flame and the plasma light emission signal generated by the laser beam-induced breakdown. The measurement cavity is located on the side of the first plano-convex lens away from the filter, and the fiber optic probe is located on the side of the second plano-convex lens away from the filter.

[0009] Furthermore, the measuring cavity includes an observation box, which is rectangular, and observation windows made of quartz glass are arranged sequentially around its perimeter as follows: observation window one, observation window two, observation window three, and observation window four.

[0010] Furthermore, the observation box is also provided with an air inlet window for connecting to the air supply assembly and an exhaust window for exhausting air. The exhaust window is connected to a vacuum pump. The air supply assembly includes an air tank and a valve and a mass flow controller connected in sequence to the outlet of the air tank. The mass flow controller is connected to the air inlet window. The mass flow controller and the vacuum pump are electrically connected to the information processing unit.

[0011] A method for online measurement of flame elements, based on the aforementioned online flame element measurement system, includes the following steps: The air intake flow and extraction speed of the measuring cavity are controlled by the information processing unit, forming a local low-speed stable combustion zone in the laser focal area within the measuring cavity; The ICCD camera is controlled by a digital delay generator to acquire the flame background spectrum signal at the current measurement point in the first time period before the laser pulse is emitted. A digital delay generator controls a high-energy pulsed laser to emit a laser beam, which is focused and penetrates the flame in the local low-speed stable combustion zone to generate high-temperature plasma. During the second period of plasma radiation duration, the digital delay generator triggers the ICCD camera again to acquire the composite spectrum of the flame background spectrum and the laser-induced plasma spectrum; The flame background spectral signal is subtracted from the composite spectrum in real time to obtain the net LIBS spectrum; The obtained net LIBS spectrum was preprocessed and online deconvolution analysis was performed to separate overlapping spectral peaks and identify the characteristic spectral line intensities of the target element. Based on the characteristic spectral line intensity of the target element, combined with a machine learning calibration model, the concentration of the target element at the measurement point is determined.

[0012] Further, the net LIBS spectrum is obtained, including the following steps: The timing of the digital delay generator is controlled by the information processing unit to ensure that the acquired composite spectrum and the flame background spectrum signal correspond to the same measurement point. After normalizing the two acquired spectra, the net LIBS spectrum is obtained by performing differential operations on the pixels.

[0013] Furthermore, the net LIBS spectrum is preprocessed, including the following steps: the baseline is fitted to the net LIBS spectrum using an adaptive weighted least squares method to eliminate continuous background, resulting in a baseline-free spectrum. I c ; Select the average intensity of a band unaffected by characteristic spectral lines. I n As an internal standard, for the baseline-free spectrum I c Normalization is performed to obtain the spectrum with average intensity after intensity normalization.

[0014] Furthermore, the online deconvolution analysis includes the following steps: The preprocessed net LIBS spectrum is smoothed and differentiated to identify potential peak positions; Match the theoretical center wavelength in the database with the identified potential spectral peak positions, and initialize a Gaussian-Lorentz mixed line shape function for each successfully matched candidate spectral line; The measured spectra are modeled as a superposition of all candidate Gaussian-Lorentz mixture line functions and a baseline; A nonlinear least squares optimization algorithm is used to simultaneously optimize the parameters of all candidate spectral lines and the coefficients of the baseline in order to minimize the sum of squared residuals between the measured spectrum and the model spectrum. Remove spurious spectral lines with intensities close to zero or with excessively low signal-to-noise ratios, and finally output the characteristic spectral line intensities of each confirmed target element.

[0015] Furthermore, the machine learning calibration model is a support vector regression model, where the input of the support vector regression model is the intensity of the feature spectral lines of the target element, and the output is the concentration of the target element.

[0016] Furthermore, the decision function of the support vector regression model is: , in, , e i , e i * It is a Lagrange multiplier. K ( x i ，x ) is the kernel function. x i The input feature vector, x For support vectors, i It is a bias term. h It is a constant of the kernel function.

[0017] Compared with existing technologies, this invention provides an online flame element measurement system and method. The system utilizes a measurement cavity to house a combustion flame. This cavity has adjustable inlet and outlet windows, and the airflow state can be controlled through a sealed combustion design and the inlet / outlet windows. This creates a localized low-speed stable combustion zone in the laser focal region, suppressing velocity fluctuations and component pulsations, thus improving the spatial stability of the plasma signal. Furthermore, a digital delay generator is electrically connected to a high-energy pulsed laser and an ICCD camera to achieve high-precision timing synchronization between laser pulse emission and spectral acquisition gate width. This reduces laser-induced plasma temporal jitter, enhances the anti-interference capability of the acquired signal, improves repeatability, and significantly improves measurement stability and accuracy. In addition, the method combines temporal background subtraction and ICCD gating control to effectively eliminate continuous flame background radiation, highlighting trace element characteristic spectral lines, reducing background radiation interference, and increasing the detection limit. This improves the stability and accuracy of target element identification in complex flame spectra, making it highly practical and worthy of widespread application. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the system structure of the present invention.

[0019] Figure 2 This is a schematic diagram of the windows on the measuring cavity 3 of the present invention. Figure 1 .

[0020] Figure 3 This is a schematic diagram of the windows on the measuring cavity 3 of the present invention. Figure 2 .

[0021] Figure 4 This invention describes the online measurement process of flame elements.

[0022] Figure 5 This is a flowchart of the measurement process of the present invention.

[0023] Figure label: 1. High-energy pulsed laser; 2. Optical focusing and collection subsystem; 2-1. Optical mirror; 2-2. Filter; 2-3. First plano-convex lens; 2-4. Second plano-convex lens; 3. Measurement cavity; 3-1. Observation window one; 3-2. Observation window two; 3-3. Observation window three; 3-4. Observation window four; 3-5. Exhaust window; 3-6. Inlet window; 4. Fiber optic probe; 5. Grating spectrometer; 6. ICCD camera; 7. Digital delay generator; 8. Information processing unit; In this diagram, the connecting wires for each device are black, the laser beam is blue, and the optical signal is red. Detailed Implementation

[0024] This invention provides an online flame element measurement system and method to solve the above-mentioned problems. In order to enable those skilled in the art to better understand the technical solution of this invention and implement it, the technical solution of this invention will be clearly and thoroughly described below with reference to the accompanying drawings.

[0025] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0026] Furthermore, it should be further explained that in the description of the embodiments of the present invention, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of the present invention, "multiple" means two or more.

[0027] The terms "first," "second," "third," and "fourth" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first," "second," "third," or "fourth" may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0028] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0029] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0030] It should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on the other element or there may be an intervening element. When an element is considered to be "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used in this invention are for illustrative purposes only and do not represent the only possible implementation.

[0031] Example 1 An online flame element measurement system, such as Figure 1 As shown, its structure includes: a high-energy pulsed laser 1, an optical focusing and collection subsystem 2, a measurement cavity 3, an optical fiber probe 4, a grating spectrometer 5, an ICCD camera 6, a digital delay generator 7, and an information processing unit 8.

[0032] in: The optical focusing and collection subsystem 2 includes an optical mirror 2-1, a filter 2-2, a first plano-convex lens 2-3, and a second plano-convex lens 2-4. The optical mirror 2-1, the filter 2-2, and the first plano-convex lens 2-3 are arranged on an optical axis. The measurement cavity 3, as an active flame stabilization device, is located on the side of the first plano-convex lens 2-3 away from the filter 2-2. Through a sealed combustion design and air intake / exhaust windows, the airflow state is controlled to form a local low-speed stable combustion zone in the laser focal area, which is used to suppress turbulence interference and improve the stability of plasma signals.

[0033] The second plano-convex lens 2-4 is positioned at a 90-degree angle to the first plano-convex lens 2-3, and is used to synchronously acquire the background radiation light signal of the flame and the plasma emission signal generated by laser-induced breakdown. The fiber optic probe 4 is located on the side of the second plano-convex lens 2-4 away from the filter 2-2, with one end used to acquire the plasma signal and the other end connected to the grating spectrometer 5. The grating spectrometer 5 is connected to the ICCD camera 6, which performs photoelectric conversion and spectral resolution on the collected light signal to obtain an analyzable spectral signal. The digital delay generator 7 is electrically connected to the high-energy pulsed laser 1 and the ICCD camera 6, respectively, to achieve high-precision timing synchronization between laser pulse emission and the spectral acquisition gate width.

[0034] The information processing unit 8 is electrically connected to the ICCD camera 6, the grating spectrometer 5, and the digital delay generator 7, respectively, and is used to receive spectral signals and perform subsequent processing and analysis, while also controlling the system. The ICCD camera 6 contains a background light suppression module, the switching or gating timing of which is controlled by the digital delay generator 7.

[0035] It should be noted that the information processing unit 8 is implemented using a computer.

[0036] The laser beam emitted from the high-energy pulsed laser 1 is guided into the measurement cavity 3 through an optical system consisting of an optical mirror 2-1, a filter 2-2, and a first plano-convex lens 2-3 placed outside the measurement cavity. The laser beam is then focused and penetrates the combustion flame within the measurement cavity 3 to generate high-temperature plasma. In practical use, by adjusting the parameters of the first plano-convex lens 2-3, its depth of focus is precisely located at the core of the "local low-speed stable combustion zone" formed under the control of the information processing unit 8. This ensures that the laser penetrates the combustion flame within the measurement cavity 3, avoiding the influence of turbulence.

[0037] It should be noted that the structure of measuring cavity 3 is as follows: Figure 2 and Figure 3 As shown, the device includes an observation box, which is rectangular in shape. Observation windows 3-1, 3-2, 3-3, and 3-4 are arranged sequentially around its four sides. Observation windows 3-1, 3-2, 3-3, and 3-4 are all made of high-temperature resistant quartz glass. In addition, the observation box is equipped with an air inlet window 3-6 for connecting to an external air supply assembly and an exhaust window 3-5 for exhausting air. The exhaust window 3-5 is connected to a vacuum pump. The air supply assembly includes an air tank and a valve and a mass flow controller connected sequentially to the outlet of the air tank. The mass flow controller is connected to the air inlet window 3-6. The mass flow controller and the vacuum pump are electrically connected to the information processing unit 8.

[0038] Digital delay generator 7 is configured to perform the following timing operations: Within a preset time window before triggering the high-energy pulsed laser 1, the ICCD camera 6 is turned on, the pure flame background spectrum is collected, and then the ICCD camera 6 is turned off. The high-energy pulsed laser 1 is triggered, and after the high-temperature plasma is formed, the digital delay generator 7 is controlled by the information processing unit 8 to open the detector gate and collect the plasma spectrum. The collected high-temperature plasma spectrum is transmitted to the information processing unit 8, where the pure flame background spectrum is subtracted in real time to obtain the net LIBS spectrum.

[0039] Information processing unit 8 has a built-in flame element analysis algorithm module, which includes: The spectral pre-information processing unit is used for background subtraction, baseline correction, wavelength correction, and intensity normalization. The feature spectral line deconvolution unit uses a chemometric algorithm, combined with the pre-stored NIST atomic spectrum database and flame environment feature spectral line database, to separate overlapping spectral peaks by fitting a Gaussian-Lorentz mixture model. The quantitative analysis unit calculates the concentration of the target element based on the intensity of the characteristic spectral lines after deconvolution using a support vector regression (SVR) machine learning model.

[0040] A method for online measurement of elements in a flame, such as Figure 5 As shown, it includes the following steps: S1: System Initialization and Establishment of Stable Combustion Zone: Start the high-energy pulsed laser 1, grating spectrometer 5, ICCD camera 6, and digital delay generator 7, and preheat to a stable operating state. Perform wavelength calibration on the grating spectrometer 5 using a standard mercury lamp light source to ensure the accuracy of the wavelength axis of the spectral data. Control the mass flow controller connected to the inlet window 3-6 and the vacuum pump connected to the exhaust window 3-5 through the information processing unit 8, dynamically adjust the inlet flow rate of the inlet window 3-6 and the pumping speed of the vacuum pump connected to the exhaust window 3-5, so that the flow rate and pressure in the laser focal area within the measurement cavity 3 are stabilized within the required range, forming a local low-speed stable combustion zone in the laser focal area;

[0041] S2: Flame background spectrum acquisition: The information processing unit 8 controls the digital delay generator 7 to output a TTL pulse to the ICCD camera 6, such as... Figure 4 As shown, during the T0-T1 period before the laser pulse is emitted, the ICCD camera 6 is turned on to acquire the flame background spectrum signal at the current measurement point, i.e., the flame background spectrum signal at time T0-T1 is acquired. I background After the data acquisition is complete, ICCD camera 6 is turned off. S3: Plasma excitation: In step S2, the digital delay generator 7 then controls the high-energy pulsed laser 1 to emit a laser beam. The laser beam passes through the optical mirror 2-1, the filter 2-2, and the first plano-convex lens 2-3 and is focused into the measurement cavity 3. The focused beam breaks through the flame in the local stable combustion zone and generates high-temperature plasma. S4: Plasma Spectral Acquisition: During the T2-T3 period of plasma radiation, the digital delay generator 7 triggers the ICCD camera 6 again to acquire signals. The high-temperature plasma signal passes through the first plano-convex lens 2-3, filter 2-2, and the second plano-convex lens 2-4 before entering the fiber optic probe 4, thus acquiring the spectral signal at the T2-T3 time. This signal is a composite spectrum of the flame background spectrum and the laser-induced plasma spectrum. I composite Subsequently, the grating spectrometer 5 and ICCD camera 6 convert the acquired optical signals into electrical signals and send the electrical signals to the information processing unit 8. After the acquisition is completed, the ICCD camera 6 is turned off. S5: Background Subtraction and Net Spectrum Extraction: The composite spectrum acquired in step S4 is used to subtract the flame background spectrum acquired in step S2 from the two signals obtained by the information processing unit 8 in real time to obtain the net LIBS spectrum. I net ; S6: Spectral Deconvolution and Element Recognition: Net LIBS Spectrum I net Preprocessing is performed, and the feature spectral line deconvolution unit is invoked to identify the intensity of the feature spectral lines of the target element; S7: Concentration Quantitative Analysis: Based on the identified characteristic spectral line intensities, combined with a machine learning calibration model using support vector machine regression, the concentration of the target element at the measurement point is calculated for subsequent combustion state diagnosis.

[0042] It should be noted that the machine learning calibration model is used for wavelength calibration of the grating spectrum.

[0043] Specifically, to better understand this invention, the following is a description of wavelength calibration of the grating spectrometer 5 using a standard mercury lamp light source: Calibration models typically employ polynomial fitting: , in, l The known wavelengths of the standard mercury lamp spectrum. p This represents the pixel position on the detector corresponding to the spectral line. a 0 , a 1 , a 2 ∙∙∙ represents the polynomial fitting coefficients.

[0044] Specifically, to better understand this invention, the following is a description of the timing control of the digital delay generator 7: The digital delay generator 7 performs high-precision timing control, such as... Figure 3 As shown, within each laser pulse cycle, the digital delay generator 7 first outputs a TTL pulse to the ICCD camera 6 at time T0 to acquire the pure flame background spectrum, and then turns off the ICCD camera 6. At time T1, the laser is triggered to emit a pulse, and at a time after plasma formation (delayed by 1 μs), the ICCD camera 6 is turned on again to acquire the plasma spectrum.

[0045] Specifically, to better understand this invention, the following is a description of the data information processing unit: The method for obtaining the net spectrum in step S5 is as follows: First, the timing of the digital delay generator 7 is controlled by the information processing unit 8 to ensure that the composite spectrum acquired in step S4 and the flame background spectrum signal acquired in step S2 correspond to the same measurement point. Second, the spectra acquired in the two steps are normalized to eliminate the differences in laser energy fluctuations and detector response. Finally, differential calculations are performed on the pixels. , in, k This is the background intensity correction factor, which is generally selected based on experiments. It can be determined by comparing the stability of the background spectrum at different times under conditions without plasma signals. I composite This is a composite spectrum of the flame background spectrum and the laser-induced plasma spectrum, with the flame background spectral signal... I background The background spectral signal of the flame is used as the output, which is the net LIBS spectrum.

[0046] Step S6, net spectral preprocessing, includes the following steps: 1) Baseline correction: Adaptive weighted least squares (AWLS) is used to fit the baseline of the net LIBS spectrum to eliminate continuous background and obtain the baseline-free spectrum. I c .

[0047] 2) Intensity normalization: Select the average intensity of a band unaffected by characteristic spectral lines. I n As an internal standard, for the baseline-free spectrum I c The entire spectrum is normalized to compensate for laser energy fluctuations and plasma state differences, resulting in a normalized spectrum with average intensity: , in,I n0 The average intensity after intensity normalization. I c The spectrum after baseline correction. I n The average intensity of the band unaffected by characteristic spectral lines.

[0048] The feature spectral line deconvolution unit in step S6 is implemented through the following process: database establishment, online deconvolution analysis, and result extraction and verification. The specific process is as follows: 1) Database Establishment: Standard samples of the target element at known concentrations were prepared experimentally. LIBS measurements of these standard samples were repeatedly performed in a simulated flame environment to collect LIBS spectra, forming the initial spectral library. Atomic and ionic spectral lines appearing within the flame temperature range were extracted from the NIST atomic spectral database. The experimentally acquired standard spectra were correlated and labeled with NIST theoretical spectral line information to generate a structured database. Each record in the database includes: element, spectral line center wavelength, transition energy level, theoretical relative intensity, and typical broadening type.

[0049] 2) Online deconvolution analysis: The input is the preprocessed net LIBS spectrum. The spectrum is smoothed and differentiated to initially identify all local maxima as potential spectral peak locations. l peak Based on the measured wavelength range, retrieve spectral information of all target elements that might appear within that range from the database. Compare the theoretical center wavelength λ from the database with the identified wavelengths in the spectrum. l peak For each successfully matched candidate spectral line, a Gaussian-Lorentz mixture linetype function is initialized. V ( l; l i ，A i ,s i ,c i The entire observed spectrum is modeled as a mixture of all candidate Gaussian-Lorentz linear functions and a low-order polynomial baseline. B(λ) Superposition:

[0050] , in, l i It was matched. l peak , A i It is the first i Peak intensity of the spectral lines s iIt is the Gaussian component's half-width at half-height. c i λ is the half-width at half maximum (FWHM) of the Lorentz component, and λ is the theoretical center wavelength.

[0051] The Levenberg-Marquardt algorithm, a nonlinear least squares optimization algorithm, is used to minimize the measured spectrum. I (l) With model spectrum I m (l) The objective is to optimize the sum of squared residuals between all candidate spectral lines while simultaneously optimizing the parameters of all candidate spectral lines. l, l i ，A i ， s i ,c i ) and baseline B(λ) The coefficient.

[0052] 3) Result Extraction and Validation: After optimization, check the fitting results for each spectral line and remove intensity values. A i False spectral lines that are close to zero or have too low a signal-to-noise ratio are identified. Finally, the characteristic spectral line intensity of each confirmed target element is output as the input to the quantitative analysis unit in step S7.

[0053] The quantitative analysis unit in step S7 uses a support vector regression (SVR) model, which consists of the following aspects: model structure, feature vector construction, model training, and online application of the model. Specifically:

[0054] 1) Model Structure: The basic idea of ​​Support Vector Regression (SVR) is to map the data into a high-dimensional feature space, and then perform linear regression on the data. The decision function is: , in, e i , e i * It is a Lagrange multiplier. K ( x i ，x ) is the kernel function. x i The input feature vector, x For support vectors, i It is a bias term.

[0055] The kernel function described above is a radial basis function (RBF): , in, h It is a constant of the kernel function. K ( x i ，x ) is the kernel function. x i The input feature vector, x For support vectors, the influence range of a single training sample is defined. There are parameters. D This is the penalty factor, which controls the severity of the penalty for samples that exceed the error tolerance.

[0056] 2) Feature Vector Construction: The core feature is the net intensity of the feature spectral lines of each target element output from the "feature spectral line deconvolution unit". A i The plasma temperature was calculated using the Boltzmann diagram method. T The electron density was calculated using the Stark broadening method. n e Together, they constitute the feature vector.

[0057] 3) Model Training: Standard sample spectral data acquired using the "Feature Spectral Line Deconvolution Unit" in step S6. For each standard spectrum, perform the complete steps S1-S6 to obtain its corresponding feature vector. x Each x True value of known concentration of the standard sample corresponding to this spectrum y To form a training sample ( x,y All samples are randomly divided into training, validation, and test sets. A grid search method is used to optimize the penalty factor and kernel function constant on the validation set, maximizing the model's root mean square error (RMSE) on the validation set. RMSE Minimize, coefficient of determination ( R² Maximize the optimal penalty factor and kernel function constant. Retrain the model on the entire training and validation sets to obtain the final SVR quantitative model. Evaluate the performance of the final model using independent test sets and calculate the correlation coefficient between predicted and actual concentrations. R² ), root mean square error ( RMSE ) indicator. Correlation coefficient ( R² ) and root mean square error ( RMSE The calculation formula for ) is as follows:

[0058] , , in, R² The correlation coefficient is... RMSE The root mean square error, For the sample size, , , The first i The actual concentration value of the first sample, the model for the first... i The predicted mean and predicted value for each sample.

[0059] 4) Online application of results: The characteristic spectral line intensities of the target element obtained through online measurement and processing are input into the trained model, and the model outputs... f(x test ） This is the predicted element concentration value.

[0060] Specifically, the calculation process of the present invention is concentrated in the information processing unit 8, and the information processing unit 8 can be used to uniformly control and coordinate the various components in the system, thereby realizing the remote online measurement function of the target element.

[0061] Specifically, the target elements include at least one of potassium, sodium, calcium, magnesium, lead, and zinc, which are used to achieve early warning of combustion slagging tendency, real-time monitoring of pollutant emissions, or online analysis of fuel quality.

[0062] The following specific examples use a modified double-base propellant as an example for measurement: The sample is a modified double-base propellant containing KNO3 flame suppressant, with a nominal potassium concentration of 600 mg / kg. It is made into a Ф5 mm × 10 mm propellant column and fixed on a high-temperature resistant bracket in the measuring chamber 3. A nickel-chromium ignition wire is arranged in the center of the upper surface of the propellant column. A 24 V DC power supply is applied at t=0 through a DC regulated power supply. The ignition wire heats up to about 1100 ℃ and ignites the surface of the propellant column. After the power is removed, a self-sustaining combustion flame is formed. Step 1: Start the high-energy pulsed laser 1, grating spectrometer 5, ICCD camera 6 and digital delay generator 7, and preheat for 30 minutes; calibrate the wavelength of the spectrometer with a mercury lamp (fitting residual <0.03nm); adjust the mass flow controller to an inlet flow rate of 15L / min and an exhaust back pressure of 0.3MPa to form a steady-state flame in the laser focal region. At this time, the flame propagation speed is about 8mm / s, the gas velocity is <0.3m / s, the turbulence intensity is about 1 / 6 of that of free combustion, and the combustion temperature is 1900℃–2200℃. At this time, potassium is uniformly distributed in the flame region in the form of K atoms / K ions vapor. Step 2: The digital delay generator 7 outputs a TTL pulse to the ICCD camera 6 at time T0 before the laser pulse, opens the gating, sets the sampling gate width to 1μs (T0-T1) and collects the flame background spectrum signal. At this time, the flame background spectrum signal contains the continuous radiation of propellant combustion and CN and NO molecular bands. Step 3: At time T1, laser 1 is triggered. The laser is focused onto the flame in the stable combustion zone through optical mirror 2-1, filter 2-2, and first plano-convex lens 2-3, and breaks down to generate high-temperature plasma. Step 4: After a 1.2μs delay following plasma formation, the digital delay generator 7 triggers the ICCD camera 6 again, opens the gate, sets the sampling gate width to 10μs (T2-T3), and collects the composite spectrum. The signal is transmitted through the second plano-convex lens 2-4 and the fiber optic probe 4 to the grating spectrometer 5 and the ICCD camera 6, and is converted into an electrical signal and sent to the information processing unit 8. Step 5: Information processing unit 8 calculates the net LIBS spectrum, at which point k=1.01, determined by background stability experiments without laser. Step 6: The net spectrum was baseline corrected using adaptive weighted least squares (AWLS) and normalized with the intensity of the Ar I 751.5 nm spectral line as the internal standard. The characteristic spectral line deconvolution unit called the NIST database to decompose the overlapping peaks of the K I 766.5 / 769.9 nm and the CN molecular band using a Gaussian-Lorentz mixture model (Levenberg-Marquardt optimization, residual <2.5%), and extracted the net intensity of the K I 766.5 nm characteristic spectral line. Step 7: Construct a feature vector using the net intensity of the characteristic spectral lines, plasma temperature (Boltzmann diagram method, using four spectral lines: K I 766.5 / 769.9 nm and K I 404.4 / 404.7 nm), and electron density (Hα 656.3 nm Stark broadening). Input this feature vector into a pre-trained support vector regression (SVR) model (penalty factor of 12, kernel parameter of 0.08) to output potassium concentration.

[0063] The combustion process of the same propellant grain was measured 100 times (with a 1-second interval between each measurement) to obtain the following results: The average potassium concentration was 608 mg / kg, with a relative error of 1.3% and a relative standard deviation (RSD) of 4.7%.

[0064] This invention provides an online flame element measurement system and method. The system utilizes a measurement cavity to house a combustion flame. The cavity has adjustable inlet and outlet windows, and the airflow state can be controlled through a sealed combustion design and the inlet / outlet windows. This creates a localized low-speed stable combustion zone in the laser focal region, suppressing velocity fluctuations and component pulsations, thus improving the spatial stability of the plasma signal. Furthermore, a digital delay generator is electrically connected to a high-energy pulsed laser and an ICCD camera to achieve high-precision timing synchronization between laser pulse emission and spectral acquisition gate width. This reduces laser-induced plasma temporal jitter, enhances the anti-interference capability of the acquired signal, improves repeatability, and significantly improves measurement stability and accuracy. In addition, the method combines temporal background subtraction and ICCD gating control to effectively eliminate continuous flame background radiation, highlighting trace element characteristic spectral lines, reducing background radiation interference, and increasing the detection limit. This improves the stability and accuracy of target element identification in complex flame spectra, making it highly practical and worthy of widespread application.

[0065] The above-disclosed embodiment is only one specific embodiment of the present invention. However, the embodiments of the present invention are not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention.

Claims

1. An online flame element measurement system, characterized in that, It includes a high-energy pulsed laser, an optical focusing and collection subsystem, a measurement cavity, a fiber optic probe, a grating spectrometer, an ICCD camera, a digital delay generator, and an information processing unit, among which: The measuring chamber is used to contain the combustion flame, and the measuring chamber has an adjustable air inlet window and an exhaust window; The high-energy pulsed laser is used to emit a laser beam; The optical focusing and collecting subsystem is used to guide the laser beam emitted by the high-energy pulsed laser into the measurement cavity, focus and break down the burning flame in the measurement cavity to generate high-temperature plasma, and simultaneously collect the flame background radiation light signal and the plasma light emission signal generated by laser-induced breakdown. One end of the fiber optic probe is used to collect plasma signals, and the other end is connected to a grating spectrometer. The grating spectrometer is connected to an ICCD camera to perform photoelectric conversion and spectral resolution on the collected optical signals to obtain the spectral signals for analysis. The digital delay generator is electrically connected to the high-energy pulsed laser and the ICCD camera respectively, and is used to achieve high-precision timing synchronization between laser pulse emission and spectral acquisition gate width; The information processing unit is electrically connected to the ICCD camera, grating spectrometer and digital delay generator respectively, and is used to control the system and receive spectral signals and convert them into the concentration of the target element.

2. The online flame element measurement system according to claim 1, characterized in that, The optical focusing and collection subsystem includes an optical mirror, a filter, a first plano-convex lens, and a second plano-convex lens. The optical mirror, filter, and first plano-convex lens are arranged on an optical axis. The second plano-convex lens is set at a 90-degree angle to the first plano-convex lens. It is used to simultaneously collect the background radiation light signal of the flame and the plasma light emission signal generated by the laser beam-induced breakdown. The measurement cavity is located on the side of the first plano-convex lens away from the filter, and the fiber optic probe is located on the side of the second plano-convex lens away from the filter.

3. The online flame element measurement system according to claim 1, characterized in that, The measuring cavity includes an observation box, which is rectangular in shape, and has four observation windows made of quartz glass arranged sequentially around its perimeter.

4. The online flame element measurement system according to claim 3, characterized in that, The observation box is also equipped with an air inlet window for connecting to the air supply component and an exhaust window for exhausting air. The exhaust window is connected to a vacuum pump. The air supply component includes an air tank and a valve and a mass flow controller connected in sequence at the outlet of the air tank. The mass flow controller is connected to the air inlet window. The mass flow controller and the vacuum pump are electrically connected to the information processing unit.

5. A method for online measurement of elements in a flame, characterized in that, The online flame element measurement system according to any one of claims 1-4 includes the following steps: The air intake flow and extraction speed of the measuring cavity are controlled by the information processing unit, forming a local low-speed stable combustion zone in the laser focal area within the measuring cavity; The ICCD camera is controlled by a digital delay generator to acquire the flame background spectrum signal at the current measurement point in the first time period before the laser pulse is emitted. A digital delay generator controls a high-energy pulsed laser to emit a laser beam, which is focused and penetrates the flame in the local low-speed stable combustion zone to generate high-temperature plasma. During the second period of plasma radiation duration, the digital delay generator triggers the ICCD camera again to acquire the composite spectrum of the flame background spectrum and the laser-induced plasma spectrum; The flame background spectral signal is subtracted from the composite spectrum in real time to obtain the net LIBS spectrum; The obtained net LIBS spectrum was preprocessed and online deconvolution analysis was performed to separate overlapping spectral peaks and identify the characteristic spectral line intensities of the target element. Based on the characteristic spectral line intensity of the target element, combined with a machine learning calibration model, the concentration of the target element at the measurement point is determined.

6. The online flame element measurement method according to claim 5, characterized in that, To obtain the net LIBS spectrum, the following steps are included: The timing of the digital delay generator is controlled by the information processing unit to ensure that the acquired composite spectrum and the flame background spectrum signal correspond to the same measurement point. After normalizing the two acquired spectra, the net LIBS spectrum is obtained by performing differential operations on the pixels.

7. The online flame element measurement method according to claim 5, characterized in that, Preprocessing the net LIBS spectrum includes the following steps: Adaptive weighted least squares method is used to fit the baseline of the net LIBS spectrum to eliminate continuous background, resulting in a baseline-free spectrum. I c ; Select the average intensity of a band unaffected by characteristic spectral lines. I n As an internal standard, for the baseline-free spectrum I c Normalization is performed to obtain the spectrum with average intensity after intensity normalization.

8. The online flame element measurement method according to claim 5, characterized in that, The online deconvolution analysis includes the following steps: The preprocessed net LIBS spectrum is smoothed and differentiated to identify potential peak positions; Match the theoretical center wavelength in the database with the identified potential spectral peak positions, and initialize a Gaussian-Lorentz mixed line shape function for each successfully matched candidate spectral line; The measured spectra are modeled as a superposition of all candidate Gaussian-Lorentz mixture line functions and a baseline; A nonlinear least squares optimization algorithm is used to simultaneously optimize the parameters of all candidate spectral lines and the coefficients of the baseline in order to minimize the sum of squared residuals between the measured spectrum and the model spectrum. Remove spurious spectral lines with intensities close to zero or with excessively low signal-to-noise ratios, and finally output the characteristic spectral line intensities of each confirmed target element.

9. The online flame element measurement method according to claim 5, characterized in that, The machine learning calibration model is a support vector regression model. The input of the support vector regression model is the intensity of the feature spectral lines of the target element, and the output is the concentration of the target element.

10. The online flame element measurement method according to claim 9, characterized in that, The decision function of the support vector regression model is: ， in, , e i , e i * It is a Lagrange multiplier. K ( x i ，x ) is the kernel function. x i The input feature vector, x For support vectors, θ It is a bias term. h It is a constant of the kernel function.

Images