Full-automatic multi-point sampling system and method for particulate matters in large flue

Abstract

The invention relates to the technical field of particulate matter sampling analysis, and discloses a full-automatic multi-point sampling system and method for particulate matters in a large flue. The system comprises an analysis module for acquiring airflow data through an airflow parameter sensor; the arrangement module is used for optimizing sampling point positions according to the airflow data; the monitoring module monitors particulate matter distribution in real time; the acquisition module is used for acquiring samples by adopting a gradient constant-speed strategy; the generation module is used for analyzing the sample to generate a component characteristic spectrum; and the fusion module comprehensively analyzes and generates a detection report. According to the application, full-automatic multi-point sampling and comprehensive analysis of the particulate matters in the large flue are realized, the defects of strong dependence on manual operation, limited sampling point positions, incapability of adapting to a complex flue environment and the like in a traditional sampling method are overcome, and the accuracy, representativeness and working efficiency of particulate matter sampling are improved.

Description

Technical Field

[0001] This application relates to the technical field of particulate matter sampling and analysis, and particularly to a fully automatic multi-point sampling system and method for particulate matter in a large flue. Background Art

[0002] During industrial production processes, particulate matter emitted from flues is one of the important sources of environmental pollution. To monitor and control the concentration of particulate matter emitted from flues, currently, mainly two types of technologies are adopted: manual sampling and on-line monitoring. Manual sampling methods mainly follow national standard specifications such as "Determination of Particulate Matter in Exhaust Gas from Stationary Pollution Sources and Sampling Methods for Gaseous Pollutants" (GB / T 16157-1996), obtain representative particulate matter samples through the isokinetic sampling principle, and then conduct weighing analysis and component detection; on-line monitoring methods mainly adopt technologies such as light scattering, β-ray attenuation, or vibrating balances to achieve continuous measurement of particulate matter concentration. These technologies have been widely applied in the fields of environmental monitoring and pollution control, providing important data support for air pollution prevention and control.

[0003] However, the existing flue particulate matter sampling and monitoring technologies have obvious deficiencies. Manual sampling methods rely on the professional level of operators, have a large workload and low efficiency, and it is difficult to achieve long-term continuous and multi-point sampling and monitoring; although on-line monitoring equipment can achieve continuous monitoring, it usually can only monitor the particulate matter concentration at fixed points, cannot comprehensively reflect the concentration distribution of the entire flue cross-section, and it is difficult to accurately analyze key characteristics such as the particle size distribution and chemical composition of particulate matter. In addition, large industrial flues generally have complex environments such as high temperature, high humidity, and uneven air flow, resulting in poor representativeness of sampling points and inaccurate measurement data. Especially in areas where the particulate matter concentration gradient changes greatly, traditional fixed-point sampling methods are difficult to capture complex spatial distribution characteristics, affecting the representativeness and accuracy of monitoring results. Summary of the Invention

[0004] This application provides a fully automatic multi-point sampling system and method for particulate matter in a large flue, which is used to realize the fully automatic multi-point sampling and comprehensive analysis of particulate matter in a large flue, overcome the defects of strong dependence on manual operation, limited sampling points, and inability to adapt to complex flue environments in traditional sampling methods, and improve the accuracy, representativeness, and working efficiency of particulate matter sampling.

[0005] In the first aspect, this application provides a fully automatic multi-point sampling system for particulate matter in a large flue, and the fully automatic multi-point sampling system for particulate matter in a large flue includes: An analysis module, which is used to analyze the characteristics of the flue gas flow through a multi-dimensional air flow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; Arrangement module, configured to optimize the arrangement of sampling points according to the three-dimensional parameter data of the flue gas flow by using a particle size distribution prediction algorithm, so as to obtain a sampling point matrix; Monitoring module, configured to monitor the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix, so as to obtain a particulate matter concentration distribution map; Collection module, configured to synchronously collect multiple parameters of the particulate matter by adopting a gradient isokinetic sampling strategy according to the particulate matter concentration distribution map, so as to form a raw particulate matter sample; Generation module, configured to perform in-situ characteristic analysis on the raw particulate matter sample through a micro-spectral analyzer to generate a particulate matter component characteristic spectrum; Fusion module, configured to comprehensively analyze the particulate matter characteristics through a multi-dimensional data fusion algorithm according to the particulate matter component characteristic spectrum, so as to generate a comprehensive detection report on the particulate matter discharged from the flue.

[0006] In a second aspect, the present application provides a fully automatic multi-point sampling method for particulate matter in a large flue. The fully automatic multi-point sampling method for particulate matter in a large flue includes: Performing characteristic analysis on the flue gas flow through a multi-dimensional gas flow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; Optimizing the arrangement of sampling points according to the three-dimensional parameter data of the flue gas flow by using a particle size distribution prediction algorithm to obtain a sampling point matrix; Monitoring the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix to obtain a particulate matter concentration distribution map; Synchronously collecting multiple parameters of the particulate matter by adopting a gradient isokinetic sampling strategy according to the particulate matter concentration distribution map to form a raw particulate matter sample; Performing in-situ characteristic analysis on the raw particulate matter sample through a micro-spectral analyzer to generate a particulate matter component characteristic spectrum; Comprehensively analyzing the particulate matter characteristics through a multi-dimensional data fusion algorithm according to the particulate matter component characteristic spectrum to generate a comprehensive detection report on the particulate matter discharged from the flue.

[0007] In the technical solution provided by this application, by using a multi-dimensional airflow parameter sensor array to comprehensively analyze the characteristics of the flue gas flow, three-dimensional parameter data including flow velocity, temperature, pressure, and turbulence intensity are obtained, comprehensively and accurately grasping the airflow state in the flue, laying a foundation for optimizing the sampling points. Based on these airflow parameter data, the system uses a particle size distribution prediction algorithm to scientifically and reasonably optimize the layout of sampling points, generating a sampling point matrix, greatly improving the scientific nature and representativeness of the sampling process. The multi-functional particulate matter sampling head monitors in real time according to the sampling point matrix, and obtains a particulate matter concentration distribution map through dual-wavelength laser scattering technology, realizing an intuitive visualization expression of the spatial distribution of particulate matter in the flue. The system innovatively adopts a gradient isokinetic sampling strategy, dynamically adjusts the sampling parameters according to the distribution characteristics of the concentration gradient, solves the limitations of traditional sampling methods when facing unevenly distributed particulate matter, and the formed original particulate matter sample is more representative. The micro-spectral analyzer performs in-situ characteristic analysis of the full ultraviolet-visible-near-infrared band on the sample, generating a characteristic spectrum of the particulate matter components, providing a scientific basis for the accurate identification of the particulate matter components. The multi-dimensional data fusion algorithm integrates multi-faceted information such as component characteristics, particle size distribution, and spatial distribution, generating a comprehensive and detailed comprehensive detection report. The system applies artificial intelligence algorithms, especially in the particle size distribution prediction and multi-dimensional data fusion links, effectively solving the problem of complex non-linear relationships that are difficult to handle by traditional methods. The particle size distribution prediction algorithm accurately predicts the particulate matter characteristics in different regions by learning the complex relationship between airflow parameters and particulate matter behavior, optimizing the sampling strategy; the multi-dimensional data fusion algorithm integrates multi-source heterogeneous data, excavates the internal connections between the data, and generates more scientific analysis results. The application of these algorithms greatly improves the adaptability of the system to the complex environment in the flue and the intelligent level of data processing, transforming the particulate matter sampling and analysis process from traditional manual experience judgment to data-driven intelligent decision-making, improving the sampling accuracy and efficiency. Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0009] Figure 1 It is a schematic diagram of an embodiment of a full-automatic multi-point sampling system for particulate matter in a large flue in an embodiment of this application; Figure 2 It is a schematic diagram of an embodiment of a full-automatic multi-point sampling method for particulate matter in a large flue in an embodiment of this application. Detailed Embodiments

[0010] The embodiments of the present application provide a full-automatic multi-point sampling system and method for particulate matter in a large flue. The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0011] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the full-automatic multi-point sampling system for particulate matter in a large flue in the embodiments of the present application includes: An analysis module, configured to analyze the characteristics of the flue gas flow through a multi-dimensional airflow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; An arrangement module, configured to optimize the arrangement of sampling points according to the three-dimensional parameter data of the flue gas flow by using a particle size distribution prediction algorithm to obtain a sampling point matrix; A monitoring module, configured to monitor the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix to obtain a particulate matter concentration distribution map; A collection module, configured to synchronously collect multi-parameters of the particulate matter by using a gradient isokinetic sampling strategy according to the particulate matter concentration distribution map to form a raw particulate matter sample; A generation module, configured to perform in-situ characteristic analysis on the raw particulate matter sample through a micro-spectral analyzer to generate a particulate matter component characteristic spectrum; A fusion module, configured to comprehensively analyze the characteristics of the particulate matter by using a multi-dimensional data fusion algorithm according to the particulate matter component characteristic spectrum to generate a comprehensive detection report on the particulate matter discharged from the flue.

[0012] It can be understood that the execution subject of the present application can be a full-automatic multi-point sampling system for particulate matter in a large flue, or a terminal or a server. Specifically, it is not limited here. The embodiments of the present application will be described by taking the server as the execution subject as an example.

[0013] In an embodiment of the present application, the analysis module uses a multi-dimensional airflow parameter sensor array such as a pitot tube, a K-type thermocouple temperature sensor, and a silicon piezoresistive pressure sensor to perform characteristic analysis on the flue airflow. When the sampling tube is inserted into the flue, the pitot tube measures the flow velocity at each point, the temperature sensor captures the temperature distribution, and the pressure sensor records the pressure change. After these data are converted into digital signals by the differential pressure sensor, an airflow density distribution map is generated through airflow tomography processing. Finally, multi-parameter data is fused to form three-dimensional parameter data of flue airflow including flow velocity, temperature, pressure and turbulence intensity. In an actual application, the analysis module collected data from 128 measurement points in a flue of a thermal power plant with a diameter of 4 meters, accurately capturing the uneven flow field distribution with a flow velocity of up to 18m / s in the center area of ​​the flue and only 4m / s near the wall. This flow field characteristic is crucial for the subsequent arrangement of sampling points.

[0014] The layout module uses the particle size distribution prediction algorithm to optimize the layout of sampling points based on the three-dimensional parameter data of the flue airflow. First, the velocity distribution gradient is analyzed, and the initial sampling area is divided according to the principle of equal area. Then, the basic sampling grid is generated by applying geometric adaptation processing according to the shape of the flue. Then, the airflow parameter data is substituted into the particle size distribution prediction algorithm to estimate the particle distribution, and the density of the basic sampling grid is adjusted accordingly to increase the density of sampling points in the high concentration gradient area. After coordinate transformation processing, three-dimensional space coordinates are generated, and finally the sampling parameters are allocated according to the preset standard requirements data to form a sampling point matrix containing spatial position, residence time and sampling flow. The preset standard requirements data is "Determination of Particles in Exhaust Gas from Fixed Pollution Sources and Sampling Methods for Gaseous Pollutants" (GB / T 16157-1996). The monitoring module monitors the particles in the flue in real time through the multifunctional particle sampling head based on the sampling point matrix. The sampling tube motion module positions the sampling head to the starting point of the preset path, and the dual-wavelength laser scatterer built into the sampling head continuously scans to obtain the particle scattering signal sequence. These signals are filtered for noise and adjusted for gain to form a scattering intensity data set, and a multi-point particle concentration data array is calculated through light scattering theory. After electrostatic interference compensation is performed on these data, spatial interpolation is used to form particle concentration spatial distribution data, and finally a particle concentration distribution map is generated through pseudo-color mapping technology.

[0015] The acquisition module uses a gradient isokinetic sampling strategy to perform multi-parameter synchronous acquisition based on the particle concentration distribution map. First, a gradient analysis is performed to identify high and low concentration gradient areas, and the isokinetic sampling flow rate is calculated in combination with the local flue gas flow rate to form a flow control parameter table. The sampling pump achieves precise flow control through a stepper motor, and dynamically allocates sampling time according to the particle concentration. The system continuously monitors changes in filter membrane resistance and compensates the sampling pump pressure in advance to keep the flow stable. Finally, the collected particle samples are stored in association with environmental parameters such as temperature, pressure, and humidity to form the original particle samples.

[0016] The generation module performs in-situ characteristic analysis on the original particulate matter sample through a micro-spectral analyzer. The sample is placed in an inert environment replaced by nitrogen, and irradiated with an ultraviolet light source (200 - 400 nm), a visible light source (400 - 700 nm), and a near-infrared light source (700 - 2500 nm) respectively to obtain full-band spectral data. The organic carbon, elemental carbon, and metal oxide components are distinguished through a characteristic spectral recognition algorithm, and quantitative calculations are carried out in combination with the results of micro X-ray fluorescence analysis to generate a particulate matter component characteristic spectrum containing the percentage content of each component.

[0017] The fusion module performs comprehensive analysis through a multi-dimensional data fusion algorithm based on the particulate matter component characteristic spectrum. The component characteristic spectrum is associated with the mass data to calculate the mass concentration of each component, and spatial distribution data is generated in combination with the sampling point location information. The particle size distribution information is extracted through high-resolution microscopic image processing to form a particle size-component association table associated with the components. Weighted calculations are carried out in combination with the airflow parameters to obtain the estimated result of the total emissions, and compliance assessment data is formed by comparing with the national standards. Finally, a comprehensive detection report of flue gas emissions particulate matter containing data tables, graphs, and analysis conclusions is generated.

[0018] In a specific embodiment, the analysis module is used for: Measure the airflow velocity at multiple positions in the flue gas duct cross-section through a Pitot tube to obtain airflow velocity distribution data; Convert the signal of the airflow velocity distribution data through a differential pressure sensor to obtain a digital airflow velocity signal; Substitute the digital airflow velocity signal into the Reynolds number calculation formula for flow regime analysis to obtain the flue gas duct airflow turbulence characteristic index; Scan and measure the gas temperature in the flue gas duct based on a K-type thermocouple temperature sensor to form temperature distribution gradient data; Comprehensively process the temperature distribution gradient data and the flue gas duct airflow turbulence characteristic index through an airflow tomography algorithm to generate an airflow density distribution map; Measure the pressure distribution in the flue gas duct according to a piezoresistive pressure sensor to obtain pressure field distribution data; Perform data fusion processing on the airflow velocity distribution data, temperature distribution gradient data, and pressure field distribution data to generate three-dimensional parameter data of the flue gas duct airflow containing flow velocity, temperature, pressure, and turbulence intensity.

[0019] Specifically, the Pitot tube is installed at the front end of the sampling tube as a flow velocity measuring device. Through the positioning of the sampling tube movement module, a preset measurement grid is formed on the flue gas cross-section. The Pitot tube utilizes the principle of fluid dynamics to measure the difference between the dynamic pressure and the static pressure generated during the flow of flue gas, which essentially reflects the gas flow velocity from a physical perspective. When the sampling tube moves in the flue, the Pitot tube stays at each preset point to measure the local flow velocity. The sampling points are usually distributed according to the equal-area principle. For example, for a circular flue with a diameter of 4 meters, 5 points are set along the radial direction and 8 points are set in the circumferential direction, forming a total of 40 measurement points to ensure an accurate characterization of the flow velocity distribution across the entire flue cross-section. The measured pressure difference data is processed by a differential pressure sensor for signal conversion. The differential pressure sensor converts the physical pressure difference signal into an electrical signal, and then an analog-to-digital converter converts the analog electrical signal into a digital signal. 16-bit analog-to-digital conversion technology is adopted in this process to ensure that the measurement accuracy reaches within ±0.5%. The converted digital gas flow velocity signal forms a data stream, records the flow velocity changes at each measurement point at a sampling frequency of 10 times per second, and forms a preliminary velocity field data matrix. The digital gas flow velocity signal is substituted into the Reynolds number calculation formula for flow regime analysis. The Reynolds number is a dimensionless parameter characterizing the fluid flow state, which is calculated through the gas flow velocity, flue characteristics size, flue gas density, and viscosity. The calculation formula is that the Reynolds number is equal to the gas flow velocity multiplied by the flue diameter multiplied by the flue gas density divided by the flue gas dynamic viscosity. The system judges whether the local flow field is laminar (Reynolds number < 2300), transitional flow (2300 < Reynolds number < 4000), or turbulent flow (Reynolds number > 4000) by calculating the Reynolds number values at each point. For large industrial flues, the Reynolds number is usually in the order of 10^5 to 10^6, indicating that the flow field is in a strong turbulent state. The system divides and colors the calculated Reynolds numbers to form a turbulent intensity distribution map, revealing the spatial distribution characteristics of the turbulent degree inside the flue. In parallel with the flow velocity measurement, a K-type thermocouple temperature sensor performs a scanning measurement of the gas temperature inside the flue. The K-type thermocouple has the characteristic of a wide temperature measurement range (-200°C to 1300°C), which is suitable for high-temperature flue environments. The temperature sensor moves along the same path as the flow velocity measurement and stays at each preset point to collect temperature data. After cold-junction compensation and linear correction processing of the temperature measurement data, temperature spatial distribution data is formed. The system smoothly connects the temperature values at these discrete points through a cubic spline interpolation algorithm to generate a continuous temperature field distribution, and calculates the temperature gradients in each direction to form temperature distribution gradient data, accurately depicting the change trend and distribution characteristics of the temperature inside the flue.

[0020] The temperature distribution gradient data and the turbulence characteristic index of the flue airflow are comprehensively processed by the airflow tomography algorithm to generate an airflow density distribution map. The airflow tomography algorithm is a technology similar to CT scanning, which reconstructs the parameter distribution of the entire cross section through the inversion calculation of multi-angle measurement data. In the specific implementation, the system uses the temperature distribution gradient data and turbulence characteristic index as the input of the inversion calculation, establishes the airflow state equation model, and solves the spatial distribution of the airflow density through the iterative calculation method. The algorithm first divides the flue cross section into grid units, and then uses the algebraic reconstruction technology to iteratively calculate the density value. After 10 to 20 iterations, it converges to a stable solution to form an airflow density distribution map. At the same time, the silicon piezoresistive pressure sensor measures the pressure distribution in the flue. The silicon piezoresistive pressure sensor is based on the principle of piezoresistive effect. When the sensor is subjected to pressure, the resistance value of the silicon material changes and is converted into a measurable electrical signal. The sensor moves along the preset path at the front end of the sampling tube to capture the static pressure data of each point. The measured electrical signal is amplified, filtered and temperature compensated to form a pressure value. The system processes these discrete point pressure data through the Kriging interpolation method to generate pressure field distribution data to characterize the spatial distribution characteristics of the pressure in the flue.

[0021] The airflow velocity distribution data, temperature distribution gradient data and pressure field distribution data are integrated to generate the three-dimensional parameter data of the flue airflow. Data fusion adopts a method combining weighted average and Kalman filtering. First, the data sources are synchronized in time and space to ensure the temporal and spatial consistency of the data. Then, a physical correlation model between the parameters is established, and the complete airflow parameter set including flow velocity, temperature, pressure, turbulence intensity, gas density and viscosity is calculated by combining the ideal gas state equation and the basic laws of fluid mechanics. The fused data is stored in the form of a three-dimensional tensor, and each spatial point contains complete airflow state parameters.

[0022] In a specific embodiment, a module is arranged to: Perform dynamic analysis of the flow field on the three-dimensional parameter data of the flue gas flow and calculate the velocity distribution gradient of the flue cross section; The velocity distribution gradient is divided into regions according to the equal area principle to obtain the initial sampling region division map; According to the flue shape, the initial sampling area division map is optimized by applying geometric adaptation processing to generate a basic sampling grid; Substituting the three-dimensional parameter data of flue gas flow into the particle size distribution prediction algorithm to estimate the particle distribution and obtain a particle size distribution prediction graph; The density of the basic sampling grid is adjusted based on the particle size distribution prediction map, and the sampling point density is increased in areas with high concentration gradients to form an optimized sampling point distribution; The optimized sampling point distribution is converted into three-dimensional space coordinates through coordinate transformation processing to generate three-dimensional positioning data of the sampling points; Sampling parameters are assigned to each sampling point according to the three-dimensional positioning data of the sampling point and the preset standard requirement data, generating a sampling point matrix including spatial position, residence time and sampling flow.

[0023] Specifically, after obtaining the three-dimensional parameter data of the flue airflow, the layout module performs a dynamic analysis of the flow field and calculates the velocity distribution gradient of the flue cross section. The flow field dynamic analysis is the process of performing gradient calculations on the velocity components in the three-dimensional parameter data of the airflow. The specific operation is to calculate the partial derivatives of the velocity vector field in all directions. The layout module selects grid points on the flue cross section, calculates the velocity change rate between adjacent grid points, and obtains the velocity gradient vector field. Taking the circular flue as an example, the layout module divides the radial distance into several equal parts, and calculates the tangential velocity gradient at each radial position. At the same time, it calculates the velocity difference between different radial positions to form complete velocity distribution gradient data. These gradient data reflect the spatial characteristics of the airflow velocity changes in the flue, which are of great significance for identifying areas with drastic airflow changes.

[0024] The layout module divides the velocity distribution gradient into regions according to the equal area principle to obtain the initial sampling area division map. The equal area principle refers to dividing the flue cross section into several sub-regions of equal area to ensure the representativeness of the sampling. For circular flues, the layout module divides the cross section into concentric ring regions, each of which has an equal area. The calculation method is to start from the center of the flue and determine the radius of each ring according to the square root ratio of the area. For rectangular flues, the equal area grid method is used for division. After the division is completed, the layout module generates an initial sampling area division map, which clearly marks the boundaries and center points of each sub-region. These center points serve as candidates for preliminary sampling points. The layout module applies geometric adaptation processing to the initial sampling area division map according to the shape of the flue to optimize it and generate a basic sampling grid. Geometric adaptation processing is a special processing step for irregular flues. The sampling area is adjusted through boundary fitting and grid deformation technology to better match the actual geometric shape of the flue. For circular flues, the concentric ring division remains unchanged; for rectangular flues, an equidistant grid is used; for elliptical or other special-shaped flues, the layout module uses conformal mapping technology to map the standard shape grid to the actual flue shape while maintaining the topological structure of the grid. The basic sampling grid generated after processing is more in line with the actual shape of the flue, ensuring the rationality of the distribution of sampling points.

[0025] The layout module substitutes the three-dimensional parameter data of the flue gas flow into the particle size distribution prediction algorithm to estimate the particle distribution, and obtains the particle size distribution prediction map. The particle size distribution prediction algorithm is a calculation method based on fluid mechanics theory and empirical models, which predicts the particle size distribution characteristics at different positions through the gas flow parameters in the flue. The algorithm first analyzes the relationships between parameters such as gas flow velocity, temperature, and turbulence intensity and the movement behavior of particles, and then calculates the movement trajectories and sedimentation characteristics of particles with different particle sizes in the gas flow according to the Stokes number. The layout module evaluates the possible particle concentration and particle size distribution in different regions of the flue through this algorithm, generates a particle size distribution prediction map, and uses different colors in the map to represent the estimated concentrations of particles in different particle size ranges. Based on the particle size distribution prediction map, the layout module adjusts the density of the basic sampling grid, increases the sampling point density in the high-concentration gradient area, and forms an optimized sampling point distribution. The density adjustment is an adaptive sampling point optimization process. The layout module analyzes the magnitude of the particle concentration gradient, increases the sampling points in the areas with larger gradients, and appropriately reduces the sampling points in the gently sloping areas. The specific operation is to calculate the spatial second derivative of the particle concentration, determine the areas where the concentration changes violently, and insert additional sampling points in these areas. At the same time, the layout module also considers the particle size distribution characteristics and conducts targeted sampling point layout for the areas dominated by different particle sizes. After the adjustment, an optimized sampling point distribution is formed, which not only ensures the sampling efficiency but also ensures the sampling accuracy in the high-gradient areas. The layout module converts the optimized sampling point distribution into three-dimensional space coordinates through coordinate transformation processing and generates the three-dimensional positioning data of the sampling points. The coordinate transformation processing is the process of converting the sampling point positions on the two-dimensional section into three-dimensional space coordinates in the flue. The layout module first establishes a three-dimensional coordinate system of the flue, determines the exact position of the sampling section in the flue, and then extends the two-dimensional coordinates of the sampling points to three-dimensional space. For a vertical flue, the Z coordinate is fixed at the height of the sampling section; for a horizontal flue, the Y coordinate is fixed at the height of the sampling section. For an inclined flue, the layout module applies a rotation matrix for coordinate transformation to ensure that the sampling point positions are perpendicular to the central axis of the flue. The generated three-dimensional positioning data of the sampling points after conversion contains the exact spatial coordinates of each sampling point, providing precise instructions for the positioning movement of the sampling tube. The layout module assigns sampling parameters to each sampling point according to the three-dimensional positioning data of the sampling points and the preset standard requirement data, and generates a sampling point matrix containing spatial position, residence time, and sampling flow rate. The preset standard requirement data is the national standard "Methods for Determination of Particulates and Sampling of Gaseous Pollutants from Exhaust Gas of Stationary Pollution Sources" (GB / T 16157-1996), which stipulates the technical requirements for particulate sampling from stationary pollution sources. The layout module calculates the sampling parameters for each sampling point according to this standard: determines the isokinetic sampling flow rate according to the local flow velocity of the sampling point; determines the sampling time according to the estimated particle concentration, shortening the time appropriately in the high-concentration area and prolonging the time in the low-concentration area; assigns a sampling point weight factor according to the sampling point density.All parameter combinations form a sampling point matrix, which is a complete instruction set for subsequent sampling execution.

[0026] In a specific embodiment, the monitoring module is used for: Convert the sampling point matrix into sampling control instructions, and drive the sampling tube movement module to position the multi-functional particulate matter sampling head to the starting point of the preset sampling path; Continuously scan the flue gas through the dual-wavelength laser scatterer built in the sampling head to obtain a particulate matter scattering signal sequence; Perform noise filtering and gain adjustment on the particulate matter scattering signal sequence through a signal processing algorithm to obtain a scattering intensity data set; Calculate the particulate matter concentration according to the scattering intensity data set using the light scattering theory to generate a multi-point particulate matter concentration data array. The formula for concentration calculation is as follows: ; Where, is the particulate matter mass concentration, representing the mass of particulate matter in unit volume of flue gas; is the transmitted light intensity, the light intensity measured after passing through the flue gas containing particulate matter; is the incident light intensity, the light intensity emitted by the initial light source, is the extinction coefficient of the i-th type of particulate matter, related to the wavelength , the complex refractive index and the particle size ; is the optical path length, the distance that the light beam passes through the flue gas; is the total number of particulate matter classifications, dimensionless; is the particulate matter radius, which is an integration variable; is the minimum particulate matter radius within the measurement range; is the maximum particulate matter radius within the measurement range; is the density function of particulate matter with particle size r; is the particulate matter size distribution function, representing the normalized number distribution of particulate matter with different particle sizes; Perform electrostatic interference compensation processing based on the multi-point particulate matter concentration data array to eliminate the influence of charged particulate matter on the measurement and form corrected concentration data; Perform spatial interpolation processing on the corrected concentration data to form particulate matter concentration spatial distribution data; Generate a particulate matter concentration distribution map according to the particulate matter concentration spatial distribution data through pseudo-color mapping technology.

[0027] Specifically, the monitoring module converts the sampling point matrix into sampling control instructions. The sampling point matrix contains the three-dimensional spatial coordinates, sampling order and path planning information of multiple sampling points. The monitoring module first parses this information, converts the spatial coordinates into the displacement of the servo motor, and generates precise control instructions for the movement of the sampling tube. The control instructions adopt a standardized motion control protocol format, including three basic parameters: position, speed and acceleration, and are transmitted to the sampling tube motion module through a high-speed data bus. After receiving the instructions, the sampling tube motion module controls the servo motor to position the multifunctional particle sampling head to the starting point of the preset sampling path according to the set trajectory, with a positioning accuracy of ±1mm, ensuring the spatial accuracy of subsequent sampling. When the sampling head reaches the specified position, the dual-wavelength laser scatterer built into the sampling head begins to continuously scan the flue gas. The dual-wavelength laser scatterer is an online particle monitoring device based on the Mie scattering theory, which includes two laser emitters with different wavelengths (532nm and 650nm) and a high-sensitivity photodetector. The two wavelengths of lasers can produce different scattering characteristics for particles of different particle size ranges. The 532nm laser is mainly used to detect small particle size particles (PM2.5 and below), while the 650nm laser is more sensitive to large particle size particles (PM2.5-PM10). The laser beam forms a parallel beam through a collimator and irradiates the flue gas sample. The light scattered by the particles is captured by the scattered light receiver and converted into an electrical signal to form a particle scattering signal sequence. The intensity of the scattered signal is directly related to the particle concentration, particle size distribution and optical properties, providing raw data for subsequent analysis.

[0028] The original particulate matter scattering signal sequence contains a large amount of noise and interference. The monitoring module performs noise filtering and gain adjustment through signal processing algorithms. The signal processing algorithm first uses wavelet transform denoising technology to decompose the time-domain signal into different frequency components, identify and filter out high-frequency noise. Then, an adaptive filter is applied to further process the signal to eliminate random fluctuations and background interference. For areas with too low signal intensity, gain adjustment is performed to ensure that weak signals can also be effectively detected. The gain adjustment adopts a piecewise linear amplification method, applying different amplification factors to signals in different intensity ranges to avoid signal distortion. The processed scattering intensity data set has a higher signal-to-noise ratio and dynamic range, laying a foundation for accurately calculating the particulate matter concentration. After obtaining the scattering intensity data set, the monitoring module calculates the particulate matter concentration according to the light scattering theory. The light scattering theory describes the interaction relationship between particulate matter and light, and realizes the concentration inversion calculation by establishing a quantitative relationship between the scattered light intensity and the particulate matter concentration. The specific calculation process is based on the proportional coefficient method and the dual-wavelength scattering difference analysis. First, the scattering ratio is calculated using the scattering data of two wavelengths, and the particle size distribution characteristics are deduced from the scattering ratio; then, combined with the calibration curve, the scattering intensity is converted into the mass concentration. The calibration curve is the corresponding relationship between the scattering intensity and the mass concentration established through laboratory standard particulate matter samples. Applying the above calculation method to the data of each sampling point generates a multi-point particulate matter concentration data array containing spatial position information. Since the particulate matter in the flue is usually a multi-component mixture, containing particles with different chemical compositions, different particle size distributions, and different optical properties, and there is obvious spatial inhomogeneity. A simple linear relationship model or a single-wavelength scattering model cannot accurately characterize the complex system. In addition, the high temperature, turbulence, and electrostatic environment in the flue further exacerbate the measurement difficulty. In this application, by combining the Beer-Lambert law with the Mie scattering theory and introducing an integral term of the particle size distribution, considering the extinction characteristic differences of N different types of particulate matter, the scattering data of two wavelengths, 532 nm and 650 nm, are fused and processed to achieve the accurate characterization of the complex mixture.

[0029] Particles in the flue gas often carry static charges, which can interfere with light scattering measurements. The monitoring module performs electrostatic interference compensation processing based on the particulate matter concentration data array at multiple points. The electrostatic interference compensation processing first uses the electrostatic field intensity sensor in the sampling head to measure the local electrostatic field intensity and establish a correlation model between the electrostatic field intensity and the measurement deviation. Then, according to the model, the concentration values at each measurement point are corrected to eliminate the influence of charged particles on the measurement. The correction process takes into account factors such as the charge amount of the particles, the electrostatic field distribution, and the particle size, and uses an iterative calculation method to gradually approach the true value. After the electrostatic interference compensation processing, more accurate corrected concentration data are formed, reducing the measurement error caused by the electrostatic effect. The corrected concentration data are still discrete point data, and the monitoring module performs spatial interpolation processing on them to form the particulate matter concentration spatial distribution data. The spatial interpolation processing uses the Kriging interpolation method, which is an optimal interpolation method based on distance weights and spatial autocorrelation. Kriging interpolation first analyzes the spatial correlation between the sampling points and establishes a variogram model, and then calculates the particulate matter concentration value at any point on the flue gas cross-section according to the variogram and the position relationship of each sampling point. Compared with simple linear interpolation, Kriging interpolation takes into account the spatial structure characteristics of the data and produces a more realistic continuous distribution result. The particulate matter concentration spatial distribution data formed after interpolation cover the entire flue gas cross-section, and the resolution is much higher than the number of original sampling points. The monitoring module generates a particulate matter concentration distribution map through pseudo-color mapping technology based on the particulate matter concentration spatial distribution data. Pseudo-color mapping is a visualization technology that converts scalar data into color representation. By mapping different concentration values to different colors, it intuitively displays the particulate matter distribution characteristics. The monitoring module uses a rainbow color spectrum (blue represents low concentration and red represents high concentration) as the basic color mapping scheme and applies smooth gradient processing to enhance the visual effect. Contour lines are superimposed on the map to mark the boundaries of areas with the same concentration, facilitating the identification of concentration gradients. The particulate matter concentration distribution map generated by pseudo-color mapping intuitively displays the spatial distribution characteristics of the particulate matter in the flue gas, providing a decision-making basis for subsequent sampling strategy adjustment.

[0030] In a specific embodiment, the acquisition module is used for: Perform gradient analysis on the particulate matter concentration distribution map, identify high-concentration gradient regions and low-concentration gradient regions, and generate a gradient partition map; Calculate the isokinetic sampling flow rate required for each sampling point according to the gradient partition map and the local flue gas flow rate measured by the Pitot tube, and form a flow control parameter table; Transmit the flow control parameter table to the sampling pump control system, and drive the sampling pump through a stepper motor to adjust the sampling flow rate to reach the isokinetic sampling state; Dynamically allocate the sampling time based on the particulate matter concentration values at the sampling points, reduce the sampling time for high-concentration areas, and extend the sampling time for low-concentration areas to obtain a sampling time control strategy; Real-time monitor the change in filter membrane resistance through a differential pressure sensor before and after the filter membrane to form resistance change trend data; Conduct predictive analysis on the resistance change trend data, perform early compensation on the output pressure of the sampling pump, and maintain a stable sampling flow rate to obtain flow compensation control data; Associate and store the collected particulate matter samples with the sampling process parameters to form particulate matter original samples with complete environmental information, where the sampling process parameters include temperature, pressure, humidity, oxygen content, and carbon dioxide content.

[0031] Specifically, after the acquisition module obtains the particulate matter concentration distribution map generated by the monitoring module, it first performs gradient analysis processing to identify high-concentration gradient areas and low-concentration gradient areas. Gradient analysis is a process of calculating the first derivative of the concentration spatial distribution data. By calculating the concentration change rate between adjacent grid points, the spatial gradient value of the particulate matter concentration is obtained. The acquisition module uses the finite difference method to calculate the gradient, performs differential operations on the concentration distribution data in the horizontal and vertical directions to obtain a two-dimensional gradient vector field. Then calculate the modulus of the gradient vector, which represents the severity of the concentration change at that point. According to the magnitude of the gradient modulus, the acquisition module divides the flue gas cross-section into different areas: the area where the gradient value is greater than the preset threshold is marked as the high-concentration gradient area, and the area where the gradient value is less than the threshold is marked as the low-concentration gradient area. This division forms a gradient partition map, which intuitively shows the areas where the particulate matter concentration changes violently. These areas are usually the locations where pollutant transmission or chemical reactions are active and require more refined sampling. Based on the generated gradient partition map, the acquisition module combines the local flue gas flow velocity measured by the pitot tube to calculate the isokinetic sampling flow rate required for each sampling point. Isokinetic sampling means that the velocity of the sampling air flow is exactly the same as the velocity of the sampled air flow in terms of magnitude and direction, which is a key technical requirement to ensure the representativeness of particulate matter sampling. The acquisition module calculates the flow rate required for isokinetic sampling according to the flue gas flow velocity value measured by the pitot tube at each sampling point and combines the diameter of the sampling nozzle. The calculation formula is: the sampling flow rate is equal to the flue gas flow velocity multiplied by the cross-sectional area of the sampling nozzle. For the sampling points in different areas, the acquisition module adjusts the sampling strategy according to the gradient partition map: in the high-concentration gradient area, appropriately increase the sampling flow rate to improve the sampling accuracy; in the low-concentration gradient area, maintain the standard flow rate. The flow rate calculation results of all sampling points are summarized to form a flow control parameter table, which includes the number, location, flue gas flow velocity, and corresponding sampling flow rate value of each sampling point.

[0032] The acquisition module transmits the generated flow control parameter table to the sampling pump control system via the data bus. The control system analyzes the content of the parameter table and converts the flow value into a control signal for the stepper motor. The sampling pump adopts a double-piston structure driven by a stepper motor, and realizes flow regulation by controlling the rotation speed and stroke of the stepper motor. The control signal is transmitted in the form of pulse width modulation (PWM). The motor driver controls the motor speed according to the PWM signal, and the motor speed directly determines the pumping flow of the sampling pump. The sampling pump control system also includes a closed-loop feedback mechanism. The actual sampling flow is measured in real time through a flow sensor, compared with the set value, the deviation is calculated, and the motor output is automatically adjusted through the PID (Proportional-Integral-Differential) control algorithm to make the actual sampling flow consistent with the theoretical isokinetic sampling flow, achieving the isokinetic sampling state. The acquisition module dynamically allocates the sampling time based on the particulate matter concentration value at the sampling point. The sampling time allocation follows the following principle: the sampling time is shorter in the high-concentration area to avoid filter membrane overload; the sampling time is longer in the low-concentration area to ensure sufficient sample volume is collected. In specific calculations, the acquisition module sets a reference total sampling volume, and then allocates the sampling time according to the reciprocal ratio of the concentration values at each point. For example, if the concentration at a certain point is twice that of another point, the sampling time at this point is half of the latter. In addition, the acquisition module also considers the particle size distribution characteristics of the particulate matter and conducts targeted time allocation for areas dominated by different particle sizes. The time calculation results of all sampling points are summarized to form a sampling time control strategy, which cooperates with the flow control parameter table to guide the entire sampling process.

[0033] During the sampling process, the acquisition module monitors the change in filter membrane resistance in real time through differential pressure sensors installed before and after the filter membrane. The differential pressure sensor measures the pressure difference across the filter membrane, and this pressure difference reflects the resistance state of the filter membrane. As particulate matter accumulates on the filter membrane, the filter membrane resistance gradually increases, and the pressure difference increases accordingly. The acquisition module records the pressure difference values at fixed time intervals (usually 1 second) to form time series data. Regression analysis is applied to these data to fit the resistance change trend curve and form resistance change trend data. These data intuitively reflect the dynamic change process of the filter membrane blockage degree and provide an important reference for subsequent flow control.

[0034] The acquisition module performs predictive analysis on the obtained resistance change trend data and compensates the output pressure of the sampling pump in advance. The predictive analysis uses time series prediction technology. Based on the observed resistance change trend, it predicts the change of the filter membrane resistance in the next period of time. The prediction algorithm combines linear regression and exponential smoothing methods to construct a resistance change prediction model and calculates the expected resistance values at future time points. According to the prediction results, the acquisition module adjusts the output pressure of the sampling pump in advance to compensate for the upcoming increase in the filter membrane resistance and ensures that the sampling flow rate remains stable under the condition of resistance change. This feedforward control strategy has more advantages than simple feedback control, can reduce the flow rate fluctuation caused by the change of the filter membrane resistance, and improve the stability of the sampling flow rate. The adjusted pressure control parameters form flow compensation control data, which directly guides the pressure output of the sampling pump. The acquisition module associates and stores the collected particulate matter samples with the sampling process parameters to form particulate matter original samples with complete environmental information. The sampling process parameters include temperature, pressure, humidity, oxygen content, and carbon dioxide content, which are obtained by real-time measurement through the corresponding sensors integrated in the sampling system. Each particulate matter sample is associated one-to-one with the environmental parameters at the time of its collection to form a complete data set. The associated storage adopts a structured database format, and each sample record includes sample ID, sampling point location, sampling time, sampling flow rate, particulate matter mass, and a complete set of environmental parameters. This associated storage method ensures that the sample analysis results can be accurately interpreted under specific environmental conditions and improves the scientificity and reliability of the analysis results.

[0035] In a specific embodiment, a generation module is used for: Place the particulate matter original sample in the sample chamber of the micro-spectrometer, remove the surrounding air by nitrogen replacement to form an inert analysis environment; Irradiate the particulate matter original sample with an ultraviolet light source to obtain ultraviolet band absorption spectrum data; Irradiate the particulate matter original sample with a visible light source to obtain visible light band absorption spectrum data; Irradiate the particulate matter original sample with a near-infrared light source to obtain near-infrared band absorption spectrum data; Merge and process the ultraviolet band absorption spectrum data, visible light band absorption spectrum data, and near-infrared band absorption spectrum data to form full-band spectrum data; Classify the components of the particulate matter according to the full-band spectrum data through a characteristic spectrum recognition algorithm, distinguish organic carbon, elemental carbon, and metal oxides, and generate a component classification table; Based on the component classification table and the results of micro X-ray fluorescence analysis, quantitatively calculate the content of each component to generate a particulate matter component characteristic spectrum containing the percentage content of each component.

[0036] Specifically, after receiving the raw particulate matter sample from the collection module, the generation module first places it in the sample chamber of the micro-spectral analyzer for pretreatment. The micro-spectral analyzer is a compact analytical device integrating a multi-band light source and a high-sensitivity spectral detector. Its sample chamber is designed as a closed structure to facilitate the creation of a controlled analysis environment. When the particulate matter sample is placed in the sample chamber, the generation module starts the nitrogen replacement program. By continuously introducing high-purity nitrogen (purity ≥ 99.999%) into the sample chamber while discharging the original air, complete gas replacement is achieved. The nitrogen replacement process usually lasts for 3 - 5 minutes until the oxygen concentration in the sample chamber drops below 0.1% and the carbon dioxide concentration drops below 0.01%, forming a chemically inert analysis environment. This inert environment effectively prevents the oxidation reaction of easily oxidizable components (such as polycyclic aromatic hydrocarbons, elemental carbon, etc.) in the particulate matter during the analysis process, and at the same time avoids the secondary reaction of volatile components in the sample with active molecules in the air, ensuring the accuracy and repeatability of the spectral analysis results. After the sample environment preparation is completed, the generation module first irradiates the raw particulate matter sample with an ultraviolet light source. The ultraviolet light source uses deuterium lamp technology and outputs a continuous spectrum with a wavelength range of 200 - 400 nm, covering the near-ultraviolet to deep-ultraviolet region. The ultraviolet light forms a parallel beam through the collimation optical system and irradiates on the particulate matter sample. Different components in the particulate matter have different absorption characteristics for ultraviolet light. In particular, the transitions of π bonds and non-bonding electrons in organic compounds show obvious absorption peaks in the ultraviolet region. The transmitted or reflected light is decomposed into different wavelengths by the grating spectroscopy system, received by the photomultiplier tube array detector and converted into an electrical signal, and then formed into digital ultraviolet band absorption spectral data after analog-to-digital conversion. These data contain wavelengths and corresponding absorbance values, recording the absorption characteristics of the sample in the ultraviolet region, which is particularly suitable for identifying organic pollutants such as polycyclic aromatic hydrocarbons and benzo[a]pyrene.

[0037] After the ultraviolet analysis is completed, the generating module switches the light source and irradiates the original particulate sample with a visible light source. The visible light source adopts tungsten halogen lamp technology and outputs a continuous spectrum with a wavelength range of 400 - 700 nm, covering the entire spectral region visible to the human eye. The method of irradiating the sample with visible light is similar to that of ultraviolet light, but the detection mechanism is different. The visible light transmitted or reflected by the sample is captured by a linear array detector of CCD (Charge Coupled Device) and converted into high-resolution spectral data. Metal oxides and pigment compounds in the particulate matter have characteristic absorption in the visible light region, showing specific colors. For example, iron oxide has strong absorption at 450 - 480 nm and appears reddish-brown; chromium oxide has an absorption peak at 520 - 540 nm and appears green. The generating module records the complete visible light absorption spectrum of the sample to form absorption spectral data in the visible light band, and these data are of great value for the identification of metal oxides and colored organic substances. The generating module irradiates the original particulate sample with a near-infrared light source. The near-infrared light source adopts quartz halogen lamp technology and outputs a continuous spectrum with a wavelength range of 700 - 2500 nm, covering the near-infrared region. Near-infrared light mainly excites the overtone and combination frequencies of molecular vibrations and has a specific response to the functional groups of molecular structures. The near-infrared light irradiating the sample is received by an InGaAs (Indium Gallium Arsenide) detector array and converted into an electrical signal to form absorption spectral data in the near-infrared band. These data are particularly suitable for identifying chemical bonds such as C-H, N-H, and O-H in organic substances, as well as the characteristics of crystal water and hydroxyl groups in inorganic salts, providing an important basis for distinguishing components such as organic carbon, nitrates, and sulfates.

[0038] The generation module merges the acquired ultraviolet band absorption spectrum data, visible light band absorption spectrum data and near-infrared band absorption spectrum data to form full-band spectrum data covering 200-2500nm. The merging process is not a simple data splicing, but a complex data fusion process. First, the wavelength of each band data is calibrated to ensure the accuracy and consistency of the wavelength scale; then the baseline correction is performed to eliminate the baseline offset between different bands; then the spectral response is normalized to compensate for the difference in sensitivity of detectors in different bands; finally, the spectral curve of the transition area is smoothed by the splicing algorithm of the overlapping area. The full-band spectrum data formed after processing is a continuous and smooth spectrum curve, which fully reflects the spectral characteristics of the particulate matter sample in the ultraviolet, visible and near-infrared regions, and provides complete spectral information for subsequent component identification. Based on the formed full-band spectrum data, the generation module classifies the particles into components through the characteristic spectrum recognition algorithm. The characteristic spectrum recognition algorithm is a data processing method based on pattern recognition and chemometrics. It realizes the qualitative identification of unknown sample components through comparison with the standard material spectrum library and mathematical processing. The algorithm first preprocesses the sample spectrum, including smoothing, derivative transformation and standardization; then extracts key spectral features, such as the position, shape and intensity of characteristic absorption peaks; then calculates similarity with standard spectra in the spectral library, and uses correlation coefficients, Euclidean distance and other indicators to evaluate the degree of matching; finally, the matching results are combined to determine the main component categories in the sample. Through this process, the generation module can distinguish organic carbon (such as aliphatic hydrocarbons, aromatic compounds), elemental carbon (such as carbon black, graphite) and metal oxides (such as iron oxide, calcium oxide, silicon dioxide), and form a detailed component classification table. The table contains the category name, spectral feature description and matching confidence score of each component.

[0039] The generation module quantitatively calculates the content of each component based on the component classification table and the results of micro X-ray fluorescence analysis. Micro X-ray fluorescence analysis is an elemental analysis technique that excites the inner electrons of the element by irradiating the sample with X-rays to generate characteristic fluorescent X-rays, thereby identifying and quantifying the element content. The generation module integrates the results of spectral analysis and X-ray fluorescence analysis to establish a quantitative calculation model for component content. The calculation process uses multivariate correction technology. First, a quantitative relationship model between spectral characteristics and component content is established. Then, the elemental composition information provided by X-ray fluorescence is used for constraints and corrections. Finally, the content percentage of each component is solved through an iterative optimization algorithm. The calculation results form a characteristic spectrum of particulate matter components, which lists in detail the specific content percentages of each major component such as organic carbon, elemental carbon, metal oxides, and the content information of important trace components.

[0040] In a specific embodiment, the fusion module is used to: Associate the characteristic spectra of particulate matter components with the mass data measured by an ultra-microelectronic balance to calculate the mass concentration values of each component; Perform spatial distribution processing on the mass concentration values combined with the sampling point location information to generate particulate matter component spatial distribution data; Perform morphological scanning on the particulate matter samples, extract the particle size distribution information of the particulate matter to form particle size statistical data, and conduct correlation analysis on the particle size statistical data and the particulate matter component spatial distribution data to determine the component composition within different particle size ranges and generate a particle size-component association table; Perform weighted calculation on the particle size-component association table combined with the three-dimensional parameter data of the flue gas flow to form an estimated result of the total emission considering the gas flow characteristics; Compare and analyze the estimated result of the total emission with the preset particulate matter emission concentration limit to generate compliance assessment data; Based on the compliance assessment data, the particulate matter component spatial distribution data, and the particle size statistical data, generate a comprehensive detection report of flue gas emissions of particulate matter including data tables, graphs, and conclusion analysis.

[0041] Specifically, an ultra-microelectronic balance precisely weighs the filter membrane before and after sampling, with a measurement accuracy reaching the 0.1 microgram level to obtain the total mass of particulate matter. The ultra-microelectronic balance adopts electrostatic elimination and temperature and humidity control technologies to ensure the stability of the weighing environment and exclude the interference of electrostatic attraction and environmental humidity on mass measurement. Then, the fusion module correlates the measured total mass data with the particulate matter component characteristic spectrum provided by the generation module and calculates the absolute mass of each component according to the percentage content of each component. After that, combined with the sampling volume data, the mass concentration value of each component is calculated, with the unit of milligram per cubic meter. This calculation process takes into account parameters such as the temperature, pressure, and humidity of the sampling gas, and converts the volume under actual sampling conditions into the volume under standard conditions (273.15K, 101.325kPa) to ensure the standardization and comparability of the data. After obtaining the mass concentration values of each component, the fusion module performs spatial distribution processing on these data in combination with the sampling point location information to generate particulate matter component spatial distribution data. The spatial distribution processing uses geostatistical methods to expand the discrete sampling point data into a continuous spatial distribution model. In specific implementation, the fusion module first determines the spatial coordinate system, correlates the three-dimensional coordinates of each sampling point with the corresponding component concentration data, and constructs an initial spatial point data set. Then, through spatial variogram analysis, the spatial correlation of the data is evaluated to determine appropriate interpolation methods and parameters. Commonly used interpolation methods include inverse distance weighting method, radial basis function method, and Kriging method, among which the Kriging method can consider the spatial structure characteristics of the data and produce the optimal unbiased estimate. Through the selected interpolation method, the fusion module expands the discrete point data into continuous distribution data covering the entire flue cross-section, forms the spatial distribution map of each component, and intuitively displays the distribution characteristics and change trends of different components in the flue.

[0042] Meanwhile, the fusion module performs morphological scanning on the particulate matter samples to extract the particle size distribution information. The morphological scanning uses high-resolution digital microscopy technology to microscopically image the particulate matter captured on the filter membrane. The microscopic images are processed through edge detection and image segmentation to identify individual particulate matters and measure their geometric size parameters, such as equivalent diameter, perimeter, area, etc. Based on the measurement data of a large number of particulate matters, the fusion module calculates the statistical distribution parameters of the particle size, including count median diameter, mass median diameter, geometric standard deviation, etc., to form complete particle size statistical data. These data are usually presented in the form of a particle size distribution histogram or a cumulative distribution curve to show the particle size distribution characteristics of the particulate matter. Next, the fusion module conducts a correlation analysis on the particle size statistical data and the particulate matter component spatial distribution data generated previously, using multivariate statistical methods to explore the variation rules of the component composition within different particle size ranges. The analysis results form a particle size-component correlation table, which details the content distribution of different components within each particle size interval (such as PM1, PM2.5, PM10, etc.), revealing the internal relationship between the particle size and the chemical composition. The fusion module performs weighted calculation on the particle size-component correlation table combined with the three-dimensional parameter data of the flue gas flow to form an estimation result of the total emissions considering the gas flow characteristics. The weighted calculation is a method for calculating emissions that takes into account the gas flow non-uniformity and is different from the simple average calculation. In specific operations, the fusion module first divides the flue gas cross-section into several sub-regions, and each sub-region corresponds to the data of one or more sampling points. Then, based on the three-dimensional parameter data of the gas flow, it determines the gas flow velocity and volume flow rate of each sub-region, and calculates the weight coefficient of each sub-region in the total emissions. Finally, it multiplies the particulate matter concentration of each sub-region by the corresponding weight coefficient and sums them up to obtain the weighted average concentration of the entire flue gas, and then multiplies it by the total volume flow rate of the flue gas to calculate the total emissions per unit time. This weighted calculation method takes into account the non-uniform distribution characteristics of the gas flow in the flue gas and can more accurately reflect the actual emission situation than the simple average. The fusion module compares and analyzes the calculated total emissions estimation result with the preset particulate matter emission concentration limit to generate compliance assessment data. The preset particulate matter emission concentration limit is determined by specifying the applicable emission standards, such as the "Emission Standards for Air Pollutants from Thermal Power Plants" (GB 13223-2011), the "Emission Standards for Air Pollutants from Cement Industry" (GB13223-2003), etc., to determine the corresponding particulate matter emission limits. Then, it directly compares the estimated emission concentration with the standard limit, calculates the percentage of the actual emission value to the standard limit, and evaluates the compliance degree of the emission. For particulate matters with different components or different particle size ranges, if there are specific limit requirements, they are compared and evaluated separately. The comparison results form compliance assessment data, including compliance status determination, over-standard multiple or compliance margin and other information, providing a direct basis for emission management and pollution control.

[0043] Based on all the above analysis results, the fusion module generates a comprehensive detection report on the particulate matter emissions from the flue. The comprehensive detection report is a structured document, which includes three major parts: data tables, graphs and charts, and conclusion analysis. The data table part lists all the original measurement and calculation results, including sampling point information, concentration data, particle size distribution parameters, component composition, etc.; the graph and chart part visually displays the key data, including contour maps of concentration distribution, pie charts of component composition, histograms of particle size distribution, etc.; the conclusion analysis part, based on the data results, conducts professional interpretation and evaluation on the emission characteristics, compliance status and potential problems, and puts forward targeted improvement suggestions. The report adopts a standardized format, meets the requirements of the environmental protection department, and is convenient for archiving and comparative analysis.

[0044] The above describes the full-automatic multi-point sampling system for particulate matter in a large flue in the embodiments of the present application. Next, the full-automatic multi-point sampling method for particulate matter in a large flue in the embodiments of the present application will be described. Please refer to Figure 2 , an embodiment of the full-automatic multi-point sampling method for particulate matter in a large flue in the embodiments of the present application includes: S101. Analyze the characteristics of the flue gas flow through a multi-dimensional gas flow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; S102. Optimize the layout of the sampling points according to the three-dimensional parameter data of the flue gas flow by using a particle size distribution prediction algorithm to obtain a sampling point matrix; S103. Monitor the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix to obtain a particulate matter concentration distribution map; S104. Synchronously collect multiple parameters of the particulate matter by adopting a gradient isokinetic sampling strategy according to the particulate matter concentration distribution map to form a raw particulate matter sample; S105. Conduct in-situ characteristic analysis on the raw particulate matter sample through a micro-spectral analyzer to generate a particulate matter component characteristic spectrum; S106. Comprehensively analyze the characteristics of the particulate matter by using a multi-dimensional data fusion algorithm according to the particulate matter component characteristic spectrum to generate a comprehensive detection report on the particulate matter emissions from the flue.

[0045] In the embodiments of the present application, by using a multi-dimensional airflow parameter sensor array to comprehensively analyze the characteristics of the flue gas flow, three-dimensional parameter data including flow velocity, temperature, pressure, and turbulence intensity are obtained, comprehensively and accurately grasping the airflow state in the flue, laying a foundation for optimizing the sampling points. Based on these airflow parameter data, the system uses a particle size distribution prediction algorithm to scientifically and reasonably optimize the layout of sampling points, generating a sampling point matrix, greatly improving the scientificity and representativeness of the sampling process. The multi-functional particulate matter sampling head monitors in real time according to the sampling point matrix, and obtains the particulate matter concentration distribution map through dual-wavelength laser scattering technology, realizing an intuitive visual expression of the spatial distribution of particulate matter in the flue. The system innovatively adopts a gradient isokinetic sampling strategy, dynamically adjusts the sampling parameters according to the distribution characteristics of the concentration gradient, solves the limitations of traditional sampling methods when facing unevenly distributed particulate matter, and the formed original particulate matter sample is more representative. The micro-spectral analyzer conducts in-situ characteristic analysis of the sample in the ultraviolet-visible-near-infrared full wavelength band, generating a characteristic spectrum of the particulate matter components, providing a scientific basis for the accurate identification of the particulate matter composition. The multi-dimensional data fusion algorithm integrates multi-faceted information such as component characteristics, particle size distribution, and spatial distribution, generating a comprehensive and detailed comprehensive detection report. The system applies artificial intelligence algorithms, especially in the particle size distribution prediction and multi-dimensional data fusion links, effectively solving the complex non-linear relationship problems that are difficult to handle by traditional methods. The particle size distribution prediction algorithm accurately predicts the particulate matter characteristics in different regions by learning the complex relationship between airflow parameters and particulate matter behavior, optimizing the sampling strategy; the multi-dimensional data fusion algorithm integrates multi-source heterogeneous data, explores the internal connections between the data, and generates more scientific analysis results. The application of these algorithms greatly improves the adaptability of the system to the complex environment in the flue and the intelligent level of data processing, transforming the particulate matter sampling and analysis process from traditional manual experience judgment to data-driven intelligent decision-making, improving the sampling accuracy and efficiency.

[0046] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic multi-point sampling system for particulate matter in a large flue, characterized in that, The full-automatic multi-point sampling system for particulate matter in a large flue includes: An analysis module for analyzing the characteristics of the flue gas flow through a multi-dimensional gas flow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; An arrangement module for optimizing the arrangement of sampling points according to the three-dimensional parameter data of the flue gas flow by using a particle size distribution prediction algorithm to obtain a sampling point matrix; A monitoring module for monitoring the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix to obtain a particulate matter concentration distribution map; A collection module for synchronously collecting multi-parameters of particulate matter according to the particulate matter concentration distribution map by adopting a gradient isokinetic sampling strategy to form a raw particulate matter sample; A generation module for in-situ characteristic analysis of the raw particulate matter sample through a micro-spectral analyzer to generate a particulate matter component characteristic spectrum; A fusion module for comprehensively analyzing the particulate matter characteristics according to the particulate matter component characteristic spectrum by using a multi-dimensional data fusion algorithm to generate a comprehensive detection report of flue gas emission particulate matter.

2. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 1, characterized in that The analysis module is used for: Measuring the gas flow velocity at multiple positions in the flue cross-section through a pitot tube to obtain gas flow velocity distribution data; Converting the signal of the gas flow velocity distribution data through a differential pressure sensor to obtain a digital gas flow velocity signal; Substituting the digital gas flow velocity signal into the Reynolds number calculation formula for flow regime analysis to obtain a flue gas flow turbulence characteristic index; Scanning and measuring the gas temperature in the flue based on a K-type thermocouple temperature sensor to form temperature distribution gradient data; Comprehensively processing the temperature distribution gradient data and the flue gas flow turbulence characteristic index through a gas flow tomography algorithm to generate a gas flow density distribution map; Measuring the pressure distribution in the flue according to a silicon piezoresistive pressure sensor to obtain pressure field distribution data; Performing data fusion processing on the gas flow velocity distribution data, the temperature distribution gradient data, and the pressure field distribution data to generate three-dimensional parameter data of the flue gas flow including flow velocity, temperature, pressure, and turbulence intensity.

3. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 1, wherein The arrangement module is used for: Performing dynamic analysis of the flow field on the three-dimensional parameter data of the flue gas flow and calculating the flow velocity distribution gradient of the flue cross-section; Dividing the flow velocity distribution gradient according to the equal area principle to obtain an initial sampling area division map; Optimizing the initial sampling area division map according to the flue shape by applying geometric adaptation processing to generate a basic sampling grid; Substituting the three-dimensional parameter data of the flue gas flow into a particle size distribution prediction algorithm for particulate matter distribution estimation to obtain a particulate matter particle size distribution prediction map; Adjusting the density of the basic sampling grid based on the particulate matter particle size distribution prediction map, increasing the sampling point density in the high concentration gradient area to form an optimized sampling point distribution; Converting the optimized sampling point distribution into three-dimensional space coordinates through coordinate transformation processing to generate three-dimensional positioning data of sampling points; Allocating sampling parameters to each sampling point according to the three-dimensional positioning data of sampling points and preset standard requirement data to generate a sampling point matrix including spatial position, residence time, and sampling flow rate.

4. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 1, characterized in that, The monitoring module is used for: Convert the sampling point matrix into sampling control instructions to drive the sampling tube movement module to position the multi-functional particulate matter sampling head at the starting point of the preset sampling path; Continuously scan the flue gas through the dual-wavelength laser scatterer built into the sampling head to obtain a particulate matter scatter signal sequence; Perform noise filtering and gain adjustment on the particulate matter scatter signal sequence through a signal processing algorithm to obtain a scatter intensity data set; Calculate the particulate matter concentration based on the scatter intensity data set using the light scattering theory to generate a particulate matter concentration data array at multiple points. The formula for concentration calculation is as follows: ； Among them, is the particulate matter mass concentration, representing the mass of particulate matter in the flue gas per unit volume; is the transmitted light intensity, the light intensity measured after passing through the flue gas containing particulate matter; is the incident light intensity, the light intensity emitted by the initial light source, is the extinction coefficient of the i-th type of particulate matter, related to the wavelength , complex refractive index and particle size ; is the optical path length, the distance that the light beam passes through the flue gas; is the total number of particulate matter classifications, dimensionless; is the particulate matter radius, which is the integration variable; is the minimum particulate matter radius within the measurement range; is the maximum particulate matter radius within the measurement range; is the density function of particulate matter with a particle size of r; is the particulate matter size distribution function, representing the normalized number distribution of particulate matter with different particle sizes; Perform electrostatic interference compensation processing based on the multi-point particulate matter concentration data array to eliminate the influence of charged particulate matter on the measurement and form corrected concentration data; Perform spatial interpolation processing on the corrected concentration data to form particulate matter concentration spatial distribution data; Generate a particulate matter concentration distribution map through pseudo-color mapping technology based on the particulate matter concentration spatial distribution data; 5. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 1, characterized in that, The acquisition module is used for: Perform gradient analysis on the particulate matter concentration distribution map to identify high-concentration gradient regions and low-concentration gradient regions and generate a gradient partition map; Calculate the isokinetic sampling flow rate required for each sampling point based on the gradient partition map and the local flue gas flow rate measured by the pitot tube to form a flow control parameter table; Transmit the flow control parameter table to the sampling pump control system and drive the sampling pump through a stepper motor to adjust the sampling flow rate to reach the isokinetic sampling state; Dynamically allocate the sampling time based on the particulate matter concentration value at the sampling point, reduce the sampling time for high-concentration regions, and extend the sampling time for low-concentration regions to obtain a sampling time control strategy; Real-time monitor the change in filter membrane resistance through the differential pressure sensor before and after the filter membrane to form resistance change trend data; Perform predictive analysis on the resistance change trend data and perform advance compensation on the output pressure of the sampling pump to maintain the stability of the sampling flow rate and obtain flow compensation control data; Associate and store the collected particulate matter samples with the sampling process parameters to form particulate matter original samples with complete environmental information. Among them, the sampling process parameters include temperature, pressure, humidity, oxygen content, and carbon dioxide content.

6. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 1, characterized in that, The generation module is used for: Place the particulate matter original sample in the sample chamber of the micro-spectrometer and remove the surrounding air through nitrogen replacement to form an inert analysis environment; Irradiate the particulate matter original sample with an ultraviolet light source to obtain ultraviolet band absorption spectrum data; Irradiate the particulate matter original sample with a visible light source to obtain visible light band absorption spectrum data; Irradiate the particulate matter original sample with a near-infrared light source to obtain near-infrared band absorption spectrum data; Merge and process the ultraviolet band absorption spectrum data, the visible light band absorption spectrum data, and the near-infrared band absorption spectrum data to form full-band spectrum data; Classify the components of the particulate matter according to the full-band spectrum data through a characteristic spectrum recognition algorithm to distinguish organic carbon, elemental carbon, and metal oxides and generate a component classification table; Based on the component classification table and the results of micro X-ray fluorescence analysis, the content of each component is quantitatively calculated to generate a particulate matter component characteristic spectrum containing the percentage of the content of each component.

7. The fully automatic multi-point sampling system for particulate matter in a large flue according to claim 6, characterized in that The fusion module is used for: Associating the particulate matter component characteristic spectrum with the mass data measured by an ultra-micro electronic balance to calculate the mass concentration value of each component; Performing spatial distribution processing on the mass concentration value in combination with the sampling point position information to generate particulate matter component spatial distribution data; Performing a morphology scan on the particulate matter sample, extracting the particle size distribution information of the particulate matter to form particle size statistical data, and performing correlation analysis on the particle size statistical data and the particulate matter component spatial distribution data to determine the component composition within different particle size ranges and generate a particle size-component association table; Performing weighted calculation on the particle size-component association table in combination with the three-dimensional parameter data of the flue gas flow to form an estimation result of the total emission considering the flow characteristics; Comparing and analyzing the total emission estimation result with the preset particulate matter emission concentration limit value to generate compliance assessment data; Based on the compliance assessment data, the particulate matter component spatial distribution data, and the particle size statistical data, generating a comprehensive detection report of flue gas emission particulate matter including data tables, graphs, and conclusion analysis.

8. An automatic multi-point sampling method for particulate matter in a large flue, characterized in that, The fully automatic multi-point sampling method for particulate matter in a large flue includes: Analyzing the characteristics of the flue gas flow through a multi-dimensional gas flow parameter sensor array to obtain three-dimensional parameter data of the flue gas flow; Optimally arranging the sampling points according to the three-dimensional parameter data of the flue gas flow using a particle size distribution prediction algorithm to obtain a sampling point matrix; Monitoring the particulate matter in the flue in real time through a multi-functional particulate matter sampling head according to the sampling point matrix to obtain a particulate matter concentration distribution map; Performing multi-parameter synchronous collection of the particulate matter according to the particulate matter concentration distribution map using a gradient isokinetic sampling strategy to form a raw particulate matter sample; Performing in-situ characteristic analysis on the raw particulate matter sample through a micro-spectrometer to generate a particulate matter component characteristic spectrum; Comprehensively analyzing the characteristics of the particulate matter according to the particulate matter component characteristic spectrum through a multi-dimensional data fusion algorithm to generate a comprehensive detection report of flue gas emission particulate matter.

Images