A dual-channel threshold-compensated black carbon concentration determination method and system

Abstract

The application provides a double-channel threshold compensation black carbon concentration determination method, which synchronously collects light intensity signals and environmental parameters of a main measurement and a reference double channel, obtains two instantaneous absorbances through preprocessing, calculates a segmented dynamic threshold in combination with the environmental parameters and a real-time drift amount, generates a double-channel threshold difference vector, extracts multi-dimensional features to construct a feature matrix, identifies the interference type and the mixing degree through weighted similarity matching, corrects a decoupling coefficient matrix based on the interference type, solves to obtain a pure black carbon absorption contribution value, and finally calculates the final black carbon concentration through multi-dimensional correction. The method effectively suppresses environmental drift and multi-type aerosol cross interference, and significantly improves the full-range accuracy and anti-interference performance of black carbon concentration detection in a complex atmospheric scene.

Description

Technical Field

[0001] This invention relates to the field of atmospheric environment monitoring technology, and in particular to a method and system for determining black carbon concentration using dual-channel threshold compensation. Background Technology

[0002] Black carbon is a carbonaceous aerosol component produced by the incomplete combustion of fossil fuels and biomass. It is one of the key pollutants affecting global climate change, regional air quality, and human health. Its concentration level in the atmosphere is a core indicator for atmospheric environmental monitoring, pollution source tracing, and climate effect assessment. With the continuous improvement of my country's precise air pollution prevention and control system, increasingly stringent requirements have been placed on the rapid, high-precision, and highly interference-resistant on-site detection of black carbon concentration. Optical detection methods based on the Lambert-Beer light absorption law have become the mainstream technical solution for black carbon concentration detection due to their advantages of good real-time performance, wide detection range, and continuous online monitoring.

[0003] Currently, mainstream optical detection technology for black carbon is based on the filter membrane integral absorption method. It calculates absorbance by collecting the light attenuation signal of aerosol particles on the filter membrane, and then inversely determines the black carbon concentration. Early single-channel, single-wavelength detection schemes calculated black carbon concentration solely based on the light intensity attenuation in a single near-infrared band. This method could not distinguish between cross-interference between black carbon and other light-absorbing aerosols, and was significantly affected by optical baseline drift caused by changes in ambient temperature and humidity, resulting in extremely large detection errors in low-concentration scenarios. In recent years, the industry has gradually developed dual-channel, dual-wavelength detection technology. By adding a visible light reference channel, it utilizes the differences in absorption characteristics between black carbon and interfering substances such as brown carbon and dust at different wavelengths, and eliminates some scattering and absorption interference through dual-channel differential calculation. Simultaneously, some existing technologies effectively determine the light intensity signal using a fixed threshold. However, the following shortcomings still exist: 1. Existing schemes mostly use fixed thresholds for valid signal determination, and cannot perform dynamic threshold compensation based on environmental parameters and concentration ranges. Real-time threshold drift caused by changes in environmental temperature, humidity, and pressure cannot be effectively eliminated. Valid signals are easily missed in low concentration ranges, and interference signals are easily misjudged in high concentration ranges, resulting in poor detection consistency across the entire concentration range. 2. Existing dual-channel schemes only suppress interference through simple wavelength difference, without extracting the multi-dimensional optical and dynamic characteristics of the dual-channel signals. They cannot accurately identify the type and mixing degree of interfering substances, and the elimination of cross-interference is incomplete for complex aerosol samples with multiple types of substances such as brown carbon, dust, and sulfate. 3. Existing concentration inversion processes mostly use fixed absorption coefficients and single-dimensional range corrections, without considering the changes in absorption characteristics caused by the encapsulation effect of interfering substances on black carbon. This leads to systematic biases in the detection results and cannot meet the high-precision and high-interference-resistant detection requirements for black carbon concentration in complex atmospheric environments. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for determining black carbon concentration with dual-channel threshold compensation, so as to improve the full-range accuracy and anti-interference performance of black carbon concentration detection in complex atmospheric scenarios.

[0005] To achieve the above objectives, the present invention provides the following solution: A method for determining black carbon concentration with dual-channel threshold compensation includes the following steps: The light intensity signals and environmental parameters of the main measurement channel and the reference channel are acquired simultaneously, and filtering, baseline correction and data calculation are performed to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; Based on environmental parameters and real-time drift compensation, the segmented dynamic thresholds of the two channels are calculated respectively, resulting in a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. Feature parameters of the light intensity signal are extracted based on the dual-channel threshold difference vector and instantaneous absorbance, and feature matrices are constructed from the feature parameters in a preset order; The feature matrix is ​​matched with a pre-defined standard interference feature library to calculate the similarity and thus the interference type. Based on the type of interfering substance, an absorption contribution model is constructed and solved using the absorption contribution decoupling coefficient matrix and instantaneous absorbance to obtain the pure absorption contribution value. The black carbon concentration is calculated based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

[0006] Optionally, the light intensity signals and environmental parameters of the main measurement channel and the reference channel are simultaneously acquired, and filtering, baseline correction, and data calculation are performed to obtain the instantaneous absorbance of the two channels, including: Near-infrared laser diodes are used as the main measurement channel light source, and visible light laser diodes are used as the reference channel light source. Based on photodetectors and environmental sensors, light intensity signals and environmental parameters are collected at the same sampling frequency. The light intensity signal is filtered by moving average, and the filtered light intensity signal is baseline corrected by a second-order polynomial fitting algorithm to obtain the optimized signal. Instantaneous absorbance is calculated based on the optimized signal and the initial zero-air light intensity reference value.

[0007] Optionally, based on environmental parameters and real-time drift compensation, the segmented dynamic thresholds for the two channels are calculated separately to obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference, including: The black carbon concentration is divided into three continuous and non-overlapping intervals, and different correction coefficient matrices are determined according to different intervals. Based on environmental parameters, query the correction coefficient matrix to determine the threshold correction coefficient; The real-time drift compensation is calculated based on the average value of zero-air light intensity data over a continuous period prior to the current moment. The segmented dynamic threshold is obtained by multiplying the threshold correction coefficient by the baseline threshold and then adding the real-time drift compensation. Calculate the difference between the segmented dynamic threshold of the main measurement channel and the segmented dynamic threshold of the reference channel to obtain the dual-channel threshold difference. Then, arrange the segmented dynamic threshold of the main measurement channel, the segmented dynamic threshold of the reference channel, and the dual-channel threshold difference in columns to generate a dual-channel threshold difference vector.

[0008] Optionally, feature parameters of the light intensity signal are extracted based on the dual-channel threshold difference vector and instantaneous absorbance, and the feature parameters are constructed into a feature matrix in a preset order, including: The peak value of the light intensity signal is determined based on the instantaneous absorbance, and the peak value ratio between the two channels is calculated to obtain the absorbance amplitude ratio. The time difference between the peak absorbance values ​​of the two channels is calculated using a cross-correlation algorithm, and the time difference is converted into a two-channel phase difference. Calculate the time derivative of the light intensity signal of each channel from the single channel threshold to the peak value to obtain the rise slope and rise slope ratio of the two channels. The time it takes for the light intensity signals of the two channels to drop from their peak values ​​to preset values ​​is calculated separately to obtain the attenuation time constants and the ratio of the attenuation time constants of the two channels. The characteristic matrix is ​​obtained by arranging the absorbance amplitude ratio, dual-channel phase difference, rising edge slope ratio, and decay time constant ratio in order.

[0009] Optionally, the feature matrix is ​​compared with a pre-defined standard interference feature library to calculate similarity, thereby obtaining interference types, including: Based on the dual-channel threshold difference vector, the adaptive weight coefficients of the feature parameters are calculated, and the feature weight vector is generated. Based on the feature weight vector, the feature matrix and the standard interference feature library are weighted element by element to obtain the weighted matrix and the weighted standard matrix; Calculate the weighted Euclidean distance between the weighted matrix and the weighted standard matrix to obtain the initial similarity value; Calculate the cosine similarity between the dual-channel threshold difference vector and the pre-stored standard threshold difference feature vectors of each standard particulate matter, and correct the initial similarity value based on the cosine similarity to obtain the final similarity value; Based on the final similarity value, the type of interference and the level of mixing are determined by a preset hierarchical judgment threshold system.

[0010] Optionally, the identification rules of the preset hierarchical judgment threshold system include: Select the minimum and second minimum values ​​from the final similarity scores; If the minimum value is less than or equal to the single particulate matter determination threshold and the difference between the second minimum value and the minimum value is greater than or equal to the similarity difference threshold, then the current sample is determined to be a single particulate matter, and the standard particulate matter type corresponding to the minimum value is taken as the main component of the current sample. If the threshold for determining a single particulate matter is less than the minimum value and less than the threshold for determining a binary mixed particulate matter, and the difference between the second smallest value and the minimum value is less than the similarity difference threshold, then the current sample is determined to be a binary mixed particulate matter, and the two standard particulate matter types with the smallest final similarity value are selected as the mixed components. If the minimum value is greater than the threshold for determining binary mixed particulate matter, then the current sample is determined to be multi-component mixed particulate matter.

[0011] Optionally, based on the type of interfering substance, an absorption contribution model is constructed and solved using the absorption contribution decoupling coefficient matrix and instantaneous absorbance to obtain the pure absorption contribution value, including: Based on the type of interference, the absorption contribution decoupling coefficient matrix of the corresponding standard is retrieved, and the mass absorption coefficient in the absorption contribution decoupling coefficient matrix is ​​adaptively corrected by combining the dual-channel threshold difference vector to obtain the corrected decoupling coefficient matrix. Using the instantaneous absorbance peak values ​​of the main measurement channel and the reference channel as model inputs, a set of absorption contribution equations is constructed based on the modified decoupling coefficient matrix; The absorption contribution equations are solved by the least squares method, and invalid interference absorption components that exceed the piecewise dynamic threshold are removed to obtain effective solution results. The absorption component generated solely by black carbon in the main measurement channel is extracted from the effective solution results and determined as the pure absorption contribution value.

[0012] Optionally, the black carbon concentration is calculated based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient, including: The mass absorption coefficients in the absorption contribution decoupling coefficient matrix are adaptively corrected according to the type of interfering substance to obtain the correction coefficients; The initial concentration is obtained by dividing the pure absorption contribution value by the product of the correction factor and the optical path length of the optical system; Based on the range of the initial concentration and the type of interfering substance, the concentration correction coefficient is obtained by querying the correction coefficient matrix. The dual-channel absorption decoupling residual is calculated, and the initial concentration after correction by the concentration correction factor is calibrated again based on the dual-channel absorption decoupling residual to obtain the black carbon concentration.

[0013] Optionally, the formula for calculating the dual-channel absorption decoupling residual is: ;in, and These are the instantaneous absorbance peak values ​​for the main measurement channel and the reference channel, respectively. and These represent the absorption contribution values ​​of black carbon in the main measurement channel and the reference channel, respectively. and These represent the absorption contribution values ​​of the interfering material in the main measurement channel and the reference channel, respectively. The formula for calculating black carbon concentration is: ;in, This is the initial concentration. This is the residual correction factor. This is the concentration correction factor.

[0014] A dual-channel threshold-compensated black carbon concentration determination system includes: The channel signal acquisition module is used to synchronously acquire the light intensity signals and environmental parameters of the main measurement channel and the reference channel, and perform filtering, baseline correction and data calculation to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; The channel threshold calculation module is used to calculate the segmented dynamic thresholds of the two channels based on environmental parameters and real-time drift compensation, and obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. The feature extraction module is used to extract feature parameters of the light intensity signal based on the dual-channel threshold difference vector and instantaneous absorbance, and to construct a feature matrix by arranging the feature parameters in a preset order. The interference identification module is used to perform similarity matching calculations between the feature matrix and a preset standard interference feature library to obtain the interference type; The contribution quantification module is used to construct and solve an absorption contribution model based on the type of interference by using the absorption contribution decoupling coefficient matrix and instantaneous absorbance to obtain the pure absorption contribution value. The concentration detection module is used to calculate the black carbon concentration based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

[0015] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The method and system for determining black carbon concentration with dual-channel threshold compensation provided by the present invention include: synchronously acquiring light intensity signals and environmental parameters of the main measurement channel and the reference channel, and performing filtering, baseline correction and data calculation to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; calculating the segmented dynamic thresholds of the two channels according to the environmental parameters and the real-time drift compensation amount to obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference; extracting feature parameters of the light intensity signal according to the dual-channel threshold difference vector and the instantaneous absorbance, and constructing a feature matrix according to the feature parameters in a preset order; performing similarity matching calculation between the feature matrix and a preset standard interference feature library to obtain the interference type; based on the interference type, constructing an absorption contribution model through the absorption contribution decoupling coefficient matrix and the instantaneous absorbance and solving it to obtain the pure absorption contribution value; and calculating the black carbon concentration based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system and the concentration range correction coefficient. This method effectively suppresses environmental drift and cross-interference from multiple types of aerosols, significantly improving the full-range accuracy and anti-interference performance of black carbon concentration detection in complex atmospheric scenarios. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the method for determining black carbon concentration according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the black carbon concentration determination system according to an embodiment of the present invention. Detailed Implementation

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

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1As shown, this embodiment of the invention provides a method for determining black carbon concentration with dual-channel threshold compensation, comprising the following steps: Step 100: Synchronously acquire the light intensity signals and environmental parameters of the main measurement channel and the reference channel, and perform filtering, baseline correction and data calculation to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; Step 200: Based on environmental parameters and real-time drift compensation, calculate the segmented dynamic thresholds for the two channels respectively, and obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. Step 300: Extract the feature parameters of the light intensity signal based on the dual-channel threshold difference vector and instantaneous absorbance, and construct a feature matrix by arranging the feature parameters in a preset order; Step 400: Perform similarity matching calculation between the feature matrix and the preset standard interference feature library to obtain the interference type; Step 500: Based on the type of interfering substance, construct an absorption contribution model using the absorption contribution decoupling coefficient matrix and instantaneous absorbance, and solve it to obtain the pure absorption contribution value; Step 600: Based on the pure absorption contribution value, the black carbon concentration is calculated by combining the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

[0021] In the specific implementation process, the main measurement channel in step 100 uses a near-infrared laser diode with a center wavelength of 880nm as the light source, matched with a silicon-based photodetector with high response to the near-infrared band, to collect the light intensity signal of the characteristic absorption band of black carbon; the reference channel uses a visible light laser diode with a center wavelength of 405nm as the light source, matched with a high-sensitivity photodetector corresponding to the ultraviolet-visible band, to collect the light intensity signal of the characteristic band of interference. Environmental parameter acquisition uses an integrated meteorological sensor, which can simultaneously output three types of data: ambient temperature, relative humidity, and atmospheric pressure. The temperature measurement accuracy is ±0.2℃, the relative humidity measurement accuracy is ±2%RH, and the atmospheric pressure measurement accuracy is ±0.1hPa. Specifically, an FPGA chip is used to output a synchronous clock signal from the same source, with a unified sampling frequency of 10Hz, to simultaneously trigger the constant current drive circuits of the two laser diodes, the signal acquisition circuits of the two photodetectors, and the data reading circuit of the environmental sensor, ensuring that the timestamps of each set of sampling data are completely aligned, and the time synchronization error does not exceed 1ms. Within a single sampling period, 600 sets of synchronous data packets are continuously collected over 60 seconds. Each set of data packets contains the original light intensity and voltage values ​​of the main measurement channel, the original light intensity and voltage values ​​of the reference channel, as well as the ambient temperature, relative humidity, and atmospheric pressure collected synchronously.

[0022] Further processing, including moving average filtering and second-order polynomial fitting baseline correction, is performed. To address high-frequency interference from circuit noise, photon shot noise, and airflow fluctuations in the original light intensity signal, a symmetrical 5-point moving average filtering algorithm is employed for noise reduction. The filtering window length is set to N=5, corresponding to a time span of 0.5 seconds. This effectively suppresses high-frequency noise while preserving the transient characteristics of the light intensity signal to the greatest extent possible, thus avoiding signal peak distortion due to over-filtering. For the first four initial sampling points of the sampling sequence, a forward-filling boundary processing method is used, with the filtered value of the fifth sampling point serving as the initial filtered output for the first four points, ensuring that the length of the filtered data sequence is completely consistent with the original sampling sequence.

[0023] Baseline correction is used to eliminate baseline nonlinear shifts caused by changes in ambient temperature and humidity, slow drift in light source power, and fluctuations in dark current of the photodetector, restoring the true attenuation of the light intensity signal. Specifically, the baseline fitting reference data interval is first determined. In this embodiment, the first 10 seconds of high-purity zero gas (i.e., free of any aerosol particles) passing through the filter membrane and the last 10 seconds of zero gas purging after sampling are used as the baseline fitting reference interval, corresponding to 100 sets of filtered light intensity data points. Then, a second-order polynomial fitting model is constructed, with sampling time t as the independent variable and the filtered light intensity value as the dependent variable. The fitting model expression is: ,in , , These represent the second-order fitting coefficients, first-order fitting coefficients, and constant term for the corresponding channel. The fitting coefficients are then solved using the least squares method to minimize the sum of squared residuals between the fitted baseline and the light intensity data points within the reference interval. After fitting, a goodness-of-fit check is performed. When the goodness-of-fit R... 2 A value ≥ 0.95 indicates a valid fit; if R0.95... 2 If the value is less than 0.95, outlier data points exceeding 3 times the standard deviation within the reference interval are removed and the data is refitted to ensure that the corrected signal baseline is stable near the initial zero reference value, thereby completely eliminating the nonlinear drift of the baseline.

[0024] Furthermore, the initial zero-gas light intensity reference value is calibrated. After the equipment is powered on and stabilized, high-purity zero gas is continuously introduced for 30 minutes. When the continuous fluctuation range of the light intensity signals of the two channels is ≤±0.5%, light intensity data is collected for 10 consecutive minutes and the average value is taken. The initial zero-gas reference values ​​of the main measurement channel and the reference channel are determined respectively. These reference values ​​are only updated during equipment startup calibration or periodic calibration and are used consistently during daily testing. Subsequently, the instantaneous absorbance is calculated point by point. For each sampling time t within the sampling period, the formula for calculating the instantaneous absorbance of the main measurement channel is: ,in For the optimized light intensity signal, The initial zero-gas reference value is used for the main measurement channel; the instantaneous absorbance of the reference channel is the same as that of the main measurement channel. If a negative value is found in the calculated instantaneous absorbance due to noise or baseline overcorrection, the absorbance value at that moment is set to 0 to avoid invalid data interfering with subsequent processing. Finally, two continuous and synchronous instantaneous absorbance time series are output.

[0025] It should be noted that step 100, through a dual-channel acquisition mechanism triggered by a synchronous clock from the same source, achieves precise time alignment between the light intensity signal and environmental parameters, eliminating calculation errors caused by timing deviations between channels; through parameter-matched symmetrical moving average filtering, various high-frequency noise interferences are effectively suppressed while preserving the transient characteristics of the signal, significantly improving the signal-to-noise ratio of weak light intensity signals; the baseline correction algorithm based on second-order polynomial fitting solves the nonlinear baseline shift problem caused by environmental changes, slow drift of the light source and devices, ensuring the long-term stability of the absorbance calculation baseline; finally, the dual-channel instantaneous absorbance quantitative calculation based on the Lambert-Beer law ensures the low-concentration detection capability and full-range measurement stability of the entire detection method.

[0026] In the specific implementation process, step 200 divides the full detection range of black carbon into three intervals: the first interval is the low concentration background interval, with a concentration range of 0~2μg / m³. 3 The second interval is the medium-concentration conventional interval, with a concentration range of 2~50 μg / m³. 3 The third zone is the high-concentration pollution zone, with a concentration range of >50 μg / m³. 3 Each concentration range corresponds to a correction coefficient matrix, which is a three-dimensional matrix. The three dimensions correspond to three core environmental parameters: ambient temperature, relative humidity, and atmospheric pressure. The calibration process of the matrix is ​​completed in a standard environmental test chamber. Specifically, for each concentration range, under full parameter coverage conditions of -10℃ to 50℃ temperature, 0% to 100%RH relative humidity, and 80kPa to 110kPa atmospheric pressure, a standard concentration of black carbon aerosol sample is introduced. The threshold deviation rate of the light intensity signal of the two channels is collected under different combinations of environmental parameters. Using the benchmark threshold under zero-air standard conditions as a reference, the threshold correction coefficient calibration value under each combination of environmental parameters is calculated. Finally, the threshold correction coefficient matrices of the main measurement channel and the reference channel threshold correction coefficient matrix are constructed for the low, medium, and high concentration ranges, respectively. The dimensions of each matrix are 13×11×7, corresponding to the number of calibration nodes for temperature, humidity, and pressure, respectively. Each element in the matrix is ​​the measured threshold correction coefficient under the corresponding environmental conditions.

[0027] Based on the black carbon concentration detection results of the previous sampling period output in step 100, the concentration range corresponding to the current sampling period is determined, and the correction coefficient matrix of the main and reference channels corresponding to this range is retrieved. If it is the first sampling after the device is powered on, the correction coefficient matrix of the medium concentration normal range is retrieved by default, and it will automatically switch to the corresponding range after the first concentration calculation is completed. Then, the calibration node matching of environmental parameters is completed. For the real-time collected ambient temperature, relative humidity, and atmospheric pressure, the two calibration nodes adjacent to the real-time parameters are located in the temperature, humidity, and pressure dimensions of the correction coefficient matrix, respectively. Similarly, the adjacent nodes in the humidity and pressure dimensions are located, resulting in 8 calibration nodes surrounding the real-time environmental parameters in three-dimensional space and their corresponding correction coefficient calibration values. Then, the threshold correction coefficients corresponding to the real-time environmental parameters are calculated using a trilinear interpolation algorithm, and the real-time threshold correction coefficients of the main measurement channel and the reference channel are obtained, respectively. The interpolation calculation process strictly follows the weight allocation rules of three-dimensional linear interpolation.

[0028] In the specific implementation process, a dual sliding time window system for drift calculation was determined, setting two sliding windows: a reference window and a real-time window. The reference window is a 30-minute sliding window before the current time t, corresponding to 18,000 sets of sampled data, used to characterize the long-term slow drift characteristics of the optical system; the real-time window is a 1-minute sliding window before the current time t, corresponding to 600 sets of sampled data, used to characterize the short-term drift characteristics caused by transient environmental changes. Both windows are updated in real time during the sampling process. Subsequently, the zero-gas reference light intensity data within the windows were screened. Based on the baseline correction results, zero-gas state data points with instantaneous absorbance ≤0.001 were extracted from both windows to eliminate effective signal data during the aerosol sampling process, retaining only the light intensity data under pure zero-gas conditions. The average value of the zero-gas light intensity within the reference window, the average value of the zero-gas light intensity within the real-time window, and the standard deviation of the zero-gas light intensity within the reference window were calculated respectively. Next, an environmental parameter change rate impact factor is introduced. Based on the collected environmental parameters, the changes in temperature, relative humidity, and atmospheric pressure between the current time and the start time of the reference window are calculated. Combined with the environmental sensitivity weights of the two channels, the environmental change rate impact factor is calculated. In this embodiment, the near-infrared light source of the main channel is less affected by temperature and humidity, so the weights for temperature change, relative humidity change, and atmospheric pressure change are set to 0.3, 0.4, and 0.3, respectively. The visible light source of the reference channel is more affected by temperature and humidity, so the weights are set to 0.4, 0.4, and 0.2, respectively. The formula for calculating the environmental change rate impact factor is: ,in , , The channel weights are temperature weight, relative humidity weight, and atmospheric pressure weight. , , This represents the maximum range of environmental parameter variation. The real-time drift compensation is then calculated using the following formula: ; in, The standard deviation of short-term noise drift is based on the 3σ principle. The theoretical signal of long-term slow drift, This is a true signal of long-term slow drift.

[0029] Furthermore, the baseline thresholds were calibrated under standard zero-gas conditions. After the equipment was powered on and stabilized, high-purity zero gas was continuously introduced for 2 hours, and light intensity and noise data from both channels were collected. Based on the 3σ principle, three times the standard deviation of light intensity and noise under zero-gas conditions was used as the baseline threshold, yielding the baseline thresholds for the main measurement channel and the reference channel, respectively. Subsequently, the core calculation of the segmented dynamic thresholds was performed. For the concentration range corresponding to the current sampling period, the segmented dynamic thresholds for both channels were calculated using the following formula: ;in Corresponding to the 880nm main channel and the 405nm reference channel, This is the real-time threshold correction coefficient for the corresponding channel. This is the baseline threshold for the corresponding channel. This is the real-time drift compensation amount for the corresponding channel. When a switch in concentration range is detected, a 10-second linear transition interval is used to smoothly switch the threshold, avoiding signal recognition abnormalities caused by threshold jumps. At the same time, upper and lower limits of the threshold are set, ensuring that the lower limit of the main measurement channel threshold is not lower than 50% of the benchmark threshold and the upper limit is not higher than 500% of the benchmark threshold, and the lower limit of the reference channel threshold is not lower than 60% of the benchmark threshold and the upper limit is not higher than 600% of the benchmark threshold. This ensures that the threshold is always within a reasonable range, preventing noise misjudgment due to excessively low thresholds and missing valid signals due to excessively high thresholds. Finally, the effective segmented dynamic thresholds of the two channels at the current moment are output.

[0030] Furthermore, for the current moment, the difference between the segmented dynamic threshold of the main measurement channel and the segmented dynamic threshold of the reference channel is calculated to obtain the dual-channel threshold difference. This threshold difference effectively characterizes the difference in environmental influence between the two channels. Subsequently, the three parameters—the segmented dynamic threshold of the main measurement channel, the segmented dynamic threshold of the reference channel, and the dual-channel threshold difference—are arranged in columns to generate a 3×1 dimensional dual-channel threshold difference vector. Finally, based on the pre-calibrated full-range threshold range, the three elements in the vector are subjected to min-max normalization, mapping all elements to the interval of 0~1, generating the dual-channel threshold difference vector.

[0031] It should be noted that step 200 achieves accurate correction of the threshold across the entire range and environment by dividing the three concentration intervals based on the actual detection scenario and using a correction coefficient matrix calibrated with all environmental parameters. Through the original dual-window adaptive weighted real-time drift compensation algorithm, it achieves synchronous and accurate compensation for long-term slow drift of the optical system and short-term drift caused by transient changes in the environment, eliminating the signal misjudgment and missed judgment problems caused by threshold drift. Through the standardized construction of the dual-channel threshold difference vector, it solves the problems of missed judgment of effective signals in the low concentration range and misjudgment of interference signals in the high concentration range, greatly improving the consistency and stability of detection across the entire concentration range.

[0032] In the specific implementation process, step 300 sets strict triggering rules: only when the instantaneous absorbance of the main measurement channel and the reference channel simultaneously exceeds the absorbance threshold of their respective segmented dynamic threshold conversion, and this state continues for more than 3 consecutive sampling points, is it determined to be a valid aerosol signal trigger. Single-channel spike noise exceeding the threshold is directly determined as an invalid signal and discarded. The effective signal start time is defined as the first moment when both channels simultaneously exceed the threshold, and the absorbance of both channels at the first 2 sampling points before this moment is lower than the corresponding threshold; the effective signal termination time is the moment when the absorbance of both channels simultaneously falls back below the corresponding threshold, and this state continues for 5 consecutive sampling points, thereby avoiding interval truncation errors caused by signal tail fluctuations. Within the effective interval from the start time to the termination time, the maximum absorbance of the main measurement channel and its corresponding time, and the maximum absorbance of the reference channel and its corresponding time are extracted respectively; at the same time, a peak validity verification rule is set: only when the peak value is greater than 1.5 times the absorbance threshold of the corresponding channel is it determined to be a valid peak value; otherwise, it is determined to be a noise signal, and all feature values ​​are set to zero. Furthermore, based on the Lambert-Beer law, the segmented dynamic thresholds are converted into corresponding absorbance thresholds. Simultaneously, a dual-channel peak synchronicity weight is introduced to correct the ratio error caused by the asynchronous peak times of the two channels. The weight calculation formula is as follows: ,in The absolute time difference between the peak times of the two channels. The synchronization time constant is 0.5s in this embodiment, and the dual-channel absorbance amplitude ratio is calculated using the following formula: ;in , These are the instantaneous absorbance peak values ​​for the main measurement channel and the reference channel, respectively. , These are the absorbance thresholds for the main measurement channel and the reference channel, respectively.

[0033] Furthermore, taking the midpoint between the peak times of the two channels as the center, 20 sampling points before and after are taken as the calculation interval, and based on the weighted cross-correlation function, higher weight is given to the effective signal and lower weight to the noisy region. The function expression is as follows: ; in, This is the time offset. , These are the weighting coefficients for the two channels, when... ≥ hour =1, otherwise =0.2; , The average absorbance of the two channels within the interval is calculated. Then, the value is found... The optimal time difference corresponding to the maximum value is the actual timing offset of the two channel signals, which is converted into a two-channel phase difference, expressed as: ;in, For dual-channel phase difference, To find the optimal time difference for weighted cross-correlation, The duration of the valid signal.

[0034] Furthermore, for each channel, the rising edge start time is the moment when the absorbance of that channel first exceeds the corresponding absorbance threshold, and the rising edge end time is the moment when the absorbance of that channel first reaches its peak value. Only when the number of sampling points within an interval is ≥5 is it determined to be a valid rising edge interval; otherwise, the slope ratio is set to 1. Subsequently, the absorbance data within the rising edge interval are linearly fitted, and the rising edge slope and rising edge slope ratio are obtained by minimization.

[0035] Furthermore, for each channel, the attenuation start time is when the absorbance of that channel reaches its peak value, and the attenuation end time is when the absorbance of that channel drops back to 10% of the peak net amplitude. An effective attenuation interval is determined only when the number of sampling points within the interval is ≥10; otherwise, the attenuation time constant is set to 1. Subsequently, an exponential attenuation model with filter loading correction is constructed, expressed as: ; in, This represents the decay time constant for the corresponding channel. This is a linear drift correction term for the membrane loading effect. This represents the absorbance threshold for the corresponding channel. This represents the peak absorbance value for the corresponding channel. The attenuation start time is defined. The absorbance data within the attenuation interval is fitted using a nonlinear least squares method to obtain the attenuation time constant and attenuation time constant ratio of the main measurement channel and the reference channel. Finally, four original feature values ​​are extracted in a preset order to form a 1×4 dimensional original feature row vector. The feature order strictly follows fixed rules of absorbance amplitude ratio, dual-channel phase difference, rising edge slope ratio, and attenuation time constant ratio, completely consistent with the feature order of the standard interference feature library.

[0036] It should be noted that step 300, based on dual-channel synchronous positioning of effective signals, constructs a multi-dimensional feature system covering aerosol wavelength absorption characteristics, temporal synchronization characteristics, filter membrane deposition dynamics characteristics, and light intensity attenuation characteristics. The calculation process of each feature takes into account the sampling characteristics adapted to the transient state of aerosols, eliminating the interference of environmental noise, baseline drift, concentration scale differences, and filter membrane loading effects on feature extraction accuracy, ensuring the stability and high discriminativeness of feature values ​​across the entire concentration range and environmental parameter range; through the construction of a standardized feature matrix, the accuracy of interference identification in complex atmospheric scenarios is significantly improved.

[0037] In the specific implementation process, step 400 constructs two sets of standardized databases: the first set is a standard feature vector library, with each type of standard particulate matter corresponding to a set of 4×1-dimensional standard feature vectors, which are completely matched with the feature matrix dimension and feature order output in step 300, and the vector elements are the calibrated mean and fluctuation range of the corresponding features; the second set is a standard threshold difference feature vector library, with each type of standard particulate matter corresponding to a set of 3×1-dimensional standard threshold difference feature vectors, which are completely matched with the dual-channel threshold difference vector dimension and element order output in step 200, and are used to characterize the dual-channel threshold response characteristics of different particulate matter.

[0038] In the specific implementation process, step 400 first calculates the magnitude of the threshold difference vector; a larger magnitude indicates stronger environmental interference. Then, based on the four core features extracted in step 300, basic weights are set according to the ability to distinguish aerosol types: the basic weight for the absorbance amplitude ratio is 0.4, the basic weight for the dual-channel phase difference is 0.2, the basic weight for the rise edge slope ratio is 0.2, and the basic weight for the decay time constant ratio is 0.2. Next, an environmental adaptive correction factor is introduced. This factor is the ratio of the magnitude of the threshold difference vector to the magnitude of the threshold difference vector under standard zero-air conditions. Each basic weight is multiplied by the environmental adaptive correction factor to obtain the adaptive weight coefficient, and a feature weight vector is generated.

[0039] Furthermore, the feature matrix of the current sample is weighted using element-wise multiplication to obtain a weighted feature matrix. Then, for each type of standard particulate matter in the standard feature vector library, its corresponding 4×1 dimensional standard feature vector is retrieved and weighted element-wise using the same feature weight vector to obtain the corresponding weighted standard feature vector. Based on the weighted feature matrix and all weighted standard feature vectors, the weighted Euclidean distance between the current sample and each type of standard particulate matter is calculated as the initial similarity value; the smaller the distance value, the higher the feature matching degree. The values ​​are then sorted in ascending order to generate an initial matching degree sequence. If the initial similarity value of a certain type of standard particulate matter is greater than 1.0, it is directly determined as having no matching, and subsequent correction processes will not include it in the calculation to reduce interference from invalid data. Then, based on the dual-channel threshold difference vector and the 3×1 dimensional standard threshold difference feature vector corresponding to each type of standard particulate matter in the pre-constructed standard threshold difference feature vector library, the cosine similarity between the current threshold difference vector and the standard threshold difference feature vector of each type is calculated. The closer the value is to 1, the higher the matching degree of the dual-channel threshold response characteristics between the current sample and the corresponding standard particulate matter. The initial cosine similarity value is then corrected using the following formula: ; in, This is the initial similarity value. For cosine similarity, This represents the final similarity value.

[0040] Furthermore, from the final similarity value sequence, the minimum value is identified. and the second smallest value and the corresponding standard particulate matter types and And determine the threshold for identifying a single particulate matter as The similarity difference threshold is The threshold for determining binary mixed particulate matter is The specific identification rules for the graded judgment threshold system are as follows: If satisfied ≤ ,and( )≥ If two conditions are met, the current sample is determined to be a single particulate matter, with the main component being... The corresponding standard particulate matter has a mixing degree level of single purity. like It is black carbon, and is determined to be free of interfering substances. If it is other particulate matter, it will be directly identified as the current type of interference. If satisfied < ≤ ,and( )< If two conditions are met, the current sample is determined to be a binary mixed particulate matter, and the mixed components are: and The two corresponding standard particulate matter; Simultaneously, the mass proportion of the two types of components is calculated based on the inverse ratio of the final similarity value. and ,like ≥70% is a main component-dominant mixture, 50% ≤ <70% is an equivalent mixture; the type of interfering substance is determined to be another type of mixed component besides black carbon. If both types are non-black carbon particulate matter, it is determined to be a two-type composite interference. If satisfied > If the current sample is determined to be a multi-component mixed particulate matter, the degree of mixing is a complex multi-component mixed type, and the type of interfering substance is a multi-type aerosol complex interference; and the top three non-black carbon standard particulate matter types with the highest final similarity values ​​are extracted as the main interfering components.

[0041] It should be noted that step 400, based on the adaptive weight allocation mechanism of the dual-channel threshold difference vector, realizes dynamic optimization of feature weights under different environmental interference intensities, which greatly improves the anti-interference ability of feature matching in complex environments; through the dual matching correction of weighted Euclidean distance and cosine similarity of threshold difference vector, it takes into account both the core optical dynamic characteristics of aerosols and the dual-channel environmental response characteristics, and eliminates the matching deviation caused by environmental drift; the hierarchical judgment system that strictly follows the preset rules realizes the accurate classification of single particles, binary mixed particles, and multi-component mixed particles, and at the same time completes the accurate identification of interference types and the quantitative classification of mixing degree.

[0042] In the specific implementation process, step 500, based on the Lambert-Beer superposition law of light absorption, targets five core aerosol components—black carbon, brown carbon, dust, ammonium sulfate, and ammonium nitrate—and introduces standard aerosols of known concentrations of each component under standard environmental conditions. Through repeated measurements, the standard mass absorption coefficients of each component at the wavelengths of the main measurement channel and the reference channel are calibrated, and three types of decoupling coefficient matrices are constructed. The first type is a 2×2 matrix for a single interference scenario, with rows corresponding to dual wavelengths of 880nm and 405nm, and columns corresponding to black carbon plus one type of interference component. The second type is a 2×3 matrix for a binary mixed interference scenario, with columns corresponding to black carbon plus two types of interference components. The third type is a 2×5 matrix for a multi-component mixed interference scenario, with columns corresponding to black carbon plus four types of all types of interference components. The elements in the matrix are the standard mass absorption coefficients of the corresponding components at the corresponding wavelengths.

[0043] Furthermore, based on the determination of the type and degree of interference, the corresponding standard decoupling coefficient matrix is ​​retrieved. If the interference is determined to be single brown carbon, a 2×2 matrix is ​​retrieved, with columns representing black carbon and brown carbon; if the interference is determined to be a binary mixture of black carbon, dust, and ammonium sulfate, a 2×3 matrix is ​​retrieved, with columns representing black carbon, dust, and ammonium sulfate; if the interference is determined to be a multi-component mixture, a 2×5 matrix representing all components is retrieved. Subsequently, the mass absorption coefficient within the decoupling coefficient matrix is ​​adaptively corrected based on the dual-channel threshold difference vector. The correction formula is as follows: ; in, Let be the mass absorption coefficient of the k-th component at wavelength λ. The pre-calibrated environmental sensitivity coefficients for the k-th component are (0.1 for black carbon, 0.8 for brown carbon, 0.6 for dust, 0.3 for ammonium sulfate, and 0.3 for ammonium nitrate in this embodiment). This represents the magnitude of the current dual-channel threshold difference vector. This represents the magnitude of the threshold difference vector under standard zero-gas conditions. All corrected mass absorption coefficients are arranged according to the original matrix structure to generate a corrected decoupling coefficient matrix. The matrix has a dimension of 2×N, with 2 rows corresponding to dual wavelengths and N columns corresponding to black carbon + N-1 types of interference components.

[0044] Furthermore, based on the law of superposition of light absorption, using the peak instantaneous absorbance of the dual channels as input and combining it with the modified decoupling coefficient matrix, a set of absorption contribution equations is constructed, the expression of which is: ; in, Let be the mass concentration of black carbon to be determined. Let K be the mass concentration of the k-th type of interfering component to be solved. To correct the number of columns in the decoupling coefficient matrix, corresponding to the total number of aerosol components, a least squares method with non-negativity constraints is used. The core objective is to minimize the sum of squared residuals while ensuring that the concentrations of all components are non-negative. The constraint condition is that the concentrations are non-negative, and invalid interference absorption components exceeding the piecewise dynamic threshold are removed to obtain an effective solution. Finally, the effective concentration of black carbon is extracted from the effective concentration vector. Combined with the corrected black carbon mass absorption coefficient and optical path length, the pure absorption contribution of black carbon at 880 nm in the main measurement channel is calculated. The calculation formula is as follows: ; in This represents the effective concentration of black carbon.

[0045] It should be noted that step 500, through a standardized matrix retrieval mechanism linked to the interference identification results, achieves accurate adaptation of the decoupled model under different scenarios such as single interference, binary mixture, and multi-component mixture, avoiding the solution error introduced by redundant variables; combined with the adaptive correction of the mass absorption coefficient of the dual-channel threshold difference vector, it compensates for the influence of environmental temperature, humidity, and pressure changes on aerosol absorption characteristics, eliminating the systematic bias caused by traditional fixed coefficients; the least squares solution with non-negative constraints and the invalid component elimination mechanism not only ensure the physical rationality of the decoupled results, but also completely eliminate noise interference below the detection limit, greatly improving the solution accuracy; finally, the extracted pure absorption contribution value of black carbon eliminates the cross interference of multiple types of aerosols such as brown carbon, dust, and sulfate, accurately restoring the true light absorption contribution of black carbon.

[0046] In the specific implementation process, step 600 is based on the adaptive correction process of the mass absorption coefficient within the absorption contribution decoupling coefficient matrix in step 500 to obtain the correction coefficient. The pure absorption contribution value is divided by the product of the correction coefficient and the optical path length of the optical system to obtain the initial concentration. If the pure absorption contribution value is lower than the main channel absorbance threshold or the initial concentration is lower than 0.1 μg / m², then... 3 If the initial concentration exceeds 2000 μg / m³, then set the initial concentration to 0; if the initial concentration exceeds 2000 μg / m³, then set the initial concentration to 0. 3 If the concentration is too high, it is marked as "out of range," and no further calculations are performed in either case. Then, based on the initial concentration's concentration range and the corresponding matrix column determined by the interfering substance type, the corresponding concentration correction coefficient is retrieved. Next, the dual-channel absorption decoupling residual is calculated using the following formula: ; in, and These are the instantaneous absorbance peak values ​​for the main measurement channel and the reference channel, respectively. and These represent the absorption contribution values ​​of black carbon in the main measurement channel and the reference channel, respectively. and These represent the absorption contributions of the interfering substance in the main measurement channel and the reference channel, respectively. A secondary calibration is then performed on the initial concentration after correction by the concentration correction factor, based on the dual-channel absorption decoupling residual. The calibration formula is as follows: ; in, This refers to the concentration of black carbon. This is the initial concentration. This is the residual correction factor. This is the concentration correction factor.

[0047] It should be noted that step 600 eliminates the systematic deviation caused by changes in aerosol physical properties and environmental fluctuations through dual mass absorption coefficient correction based on the interference encapsulation effect and environmental parameters; the segmented primary calibration based on concentration range and interference type takes into account both the detection capability in the low concentration range and the detection accuracy in the high concentration range; and the secondary calibration through decoupling residuals further compensates for the fitting deviation in the absorption decoupling process, suppressing all-dimensional error sources such as environmental drift, cross-interference of multiple types of aerosols, encapsulation effect, and instrument nonlinear response.

[0048] like Figure 2 As shown, this embodiment of the invention also provides a dual-channel threshold-compensated black carbon concentration determination system, comprising: The channel signal acquisition module is used to synchronously acquire the light intensity signals and environmental parameters of the main measurement channel and the reference channel, and perform filtering, baseline correction and data calculation to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; The channel threshold calculation module is used to calculate the segmented dynamic thresholds of the two channels based on environmental parameters and real-time drift compensation, and obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. The feature extraction module is used to extract feature parameters of the light intensity signal based on the dual-channel threshold difference vector and instantaneous absorbance, and to construct a feature matrix by arranging the feature parameters in a preset order. The interference identification module is used to perform similarity matching calculations between the feature matrix and a preset standard interference feature library to obtain the interference type; The contribution quantification module is used to construct and solve an absorption contribution model based on the type of interference by using the absorption contribution decoupling coefficient matrix and instantaneous absorbance to obtain the pure absorption contribution value. The concentration detection module is used to calculate the black carbon concentration based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

[0049] The beneficial effects of this invention are as follows: 1) By dividing the full detection range of black carbon into three continuous concentration ranges that are adapted to actual monitoring scenarios, a three-dimensional threshold correction coefficient matrix covering the full temperature and humidity range and the full atmospheric pressure range was constructed. Combined with the dual-window adaptive real-time drift compensation algorithm, the accurate calculation of the segmented dynamic threshold of the dual channels was realized, eliminating the influence of environmental drift and slow drift of the optical system on signal judgment, greatly improving the consistency of full-range detection, and significantly optimizing the detection capability of weak signals in low-concentration background scenarios. 2) Based on the dual-channel signal, four core features were extracted: absorbance amplitude ratio, dual-channel phase difference, rising edge slope ratio, and decay time constant ratio. This fully covers the wavelength absorption characteristics, temporal synchronization characteristics, filter film deposition kinetics, and light intensity attenuation characteristics of aerosols. Combined with the adaptive weight allocation of the dual-channel threshold difference vector, the dual matching correction combining weighted Euclidean distance and cosine similarity, and the hierarchical judgment threshold system, the types and mixing degrees of single particles, binary mixed particles, and multi-component mixed particles were accurately identified, improving the accuracy of interference identification. 3) Based on the interference identification results, the absorption contribution decoupling coefficient matrix of the corresponding system is retrieved. The mass absorption coefficient is adaptively corrected by the environment in combination with the dual-channel threshold difference vector. A set of dual-wavelength absorption contribution equations with clear physical meaning is constructed. The equations are solved by the least squares method with non-negative constraints. Invalid interference absorption components that exceed the dynamic threshold are eliminated. The light absorption contribution of black carbon and various interferences is accurately separated. The pure absorption contribution value generated only by black carbon is extracted. The accurate decoupling of the pure absorption contribution of black carbon is achieved, and the cross interference of multiple types of aerosols is eliminated. 4) A triple closed-loop calibration system was constructed: First, a secondary correction of the black carbon mass absorption coefficient based on the type of interfering substance and the encapsulation effect was implemented, eliminating the systematic deviation caused by the change in absorption characteristics due to the encapsulation of black carbon by interfering substances; second, a segmented concentration correction based on the concentration range and the type of interfering substance was implemented, solving the problem of nonlinear response of the instrument across the entire range; and third, a secondary calibration based on the decoupling residual of dual-channel absorption solved the problem of large deviation in detection results caused by the use of fixed absorption coefficient and single-dimensional correction in traditional schemes, suppressing all-dimensional error sources such as environmental fluctuations, changes in aerosol physical properties, and nonlinear response of the instrument, thereby improving the detection accuracy across the entire range.

[0050] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0051] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for determining black carbon concentration with dual-channel threshold compensation, characterized in that, Includes the following steps: The light intensity signals and environmental parameters of the main measurement channel and the reference channel are acquired synchronously, and then filtered, baseline corrected, and data calculated to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity, and atmospheric pressure data; Based on the environmental parameters and real-time drift compensation, the segmented dynamic thresholds of the two channels are calculated respectively to obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. The feature parameters of the light intensity signal are extracted based on the dual-channel threshold difference vector and the instantaneous absorbance, and the feature parameters are used to construct a feature matrix in a preset order; The feature matrix is ​​compared with a preset standard interference feature library to calculate the similarity and thus the interference type. Based on the type of interference, an absorption contribution model is constructed and solved using the absorption contribution decoupling coefficient matrix and the instantaneous absorbance to obtain the pure absorption contribution value. Based on the pure absorption contribution value, the black carbon concentration is calculated by combining the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

2. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, Simultaneously acquire the light intensity signals and environmental parameters of the main measurement channel and the reference channel, and perform filtering, baseline correction, and data calculation to obtain the instantaneous absorbance of the two channels, including: Using a near-infrared laser diode as the main measurement channel light source and a visible light laser diode as the reference channel light source, the light intensity signal and the environmental parameters are collected at the same sampling frequency based on a photodetector and an environmental sensor. The light intensity signal is subjected to moving average filtering, and the filtered light intensity signal is baseline corrected by a second-order polynomial fitting algorithm to obtain an optimized signal. The instantaneous absorbance is calculated based on the optimized signal and the initial zero-air light intensity reference value.

3. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, Based on the environmental parameters and real-time drift compensation, the segmented dynamic thresholds for both channels are calculated to obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference, including: The black carbon concentration is divided into three continuous and non-overlapping intervals, and different correction coefficient matrices are determined according to different intervals. Based on the environmental parameters, the threshold correction coefficient is determined by querying the correction coefficient matrix. The real-time drift compensation amount is calculated based on the average value of zero-air light intensity data over a continuous period prior to the current moment; The segmented dynamic threshold is obtained by multiplying the threshold correction coefficient by the baseline threshold and then adding the real-time drift compensation amount. Calculate the difference between the segmented dynamic threshold of the main measurement channel and the segmented dynamic threshold of the reference channel to obtain the dual-channel threshold difference. Then, arrange the segmented dynamic threshold of the main measurement channel, the segmented dynamic threshold of the reference channel, and the dual-channel threshold difference in columns to generate the dual-channel threshold difference vector.

4. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, Feature parameters of the light intensity signal are extracted based on the dual-channel threshold difference vector and the instantaneous absorbance, and the feature parameters are used to construct a feature matrix in a preset order, including: The peak value of the light intensity signal is determined based on the instantaneous absorbance, and the peak value ratio between the two channels is calculated to obtain the absorbance amplitude ratio. The time difference between the peak absorbance values ​​of the two channels is calculated using a cross-correlation algorithm, and the time difference is converted into a dual-channel phase difference. Calculate the time derivative of the light intensity signal of each of the two channels from the single-channel threshold to the peak value to obtain the rise slope and rise slope ratio of the two channels; The time it takes for the light intensity signal of each of the two channels to drop from the peak value to a preset value is calculated to obtain the attenuation time constant and the ratio of the attenuation time constants of the two channels. The characteristic matrix is ​​obtained by arranging the absorbance amplitude ratio, the dual-channel phase difference, the rising edge slope ratio, and the decay time constant ratio in order.

5. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, The feature matrix is ​​compared with a preset standard interference feature library to calculate similarity, thereby obtaining the interference types, including: Based on the dual-channel threshold difference vector, the adaptive weight coefficients of the feature parameters are calculated, and a feature weight vector is generated. The feature matrix and the standard interference feature library are weighted element-wise according to the feature weight vector to obtain the weighted matrix and the weighted standard matrix. Calculate the weighted Euclidean distance between the weighted matrix and the weighted standard matrix to obtain the initial similarity value; Calculate the cosine similarity between the dual-channel threshold difference vector and the pre-stored standard threshold difference feature vectors of each standard particulate matter, and correct the initial similarity value based on the cosine similarity to obtain the final similarity value; Based on the final similarity value, the type of interference and the level of mixing are determined by a preset hierarchical judgment threshold system.

6. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 5, characterized in that, The identification rules of the preset hierarchical judgment threshold system include: Select the minimum and second minimum values ​​from the final similarity values; If the minimum value is less than or equal to the single particulate matter determination threshold and the difference between the second smallest value and the minimum value is greater than or equal to the similarity difference threshold, then the current sample is determined to be a single particulate matter, and the standard particulate matter type corresponding to the minimum value is taken as the main component of the current sample. If the threshold for determining a single particulate matter is less than the minimum value and less than the threshold for determining a binary mixed particulate matter, and the difference between the second smallest value and the minimum value is less than the similarity difference threshold, then the current sample is determined to be a binary mixed particulate matter, and the two standard particulate matter types with the smallest final similarity value are selected as the mixed components. If the minimum value is greater than the binary mixed particulate matter determination threshold, then the current sample is determined to be a multi-component mixed particulate matter.

7. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, Based on the type of interference, an absorption contribution model is constructed and solved using the absorption contribution decoupling coefficient matrix and the instantaneous absorbance to obtain the pure absorption contribution value, including: Based on the type of interference, the absorption contribution decoupling coefficient matrix of the corresponding standard is retrieved, and the mass absorption coefficient in the absorption contribution decoupling coefficient matrix is ​​adaptively corrected by combining the dual-channel threshold difference vector to obtain the corrected decoupling coefficient matrix. Using the instantaneous absorbance peak values ​​of the main measurement channel and the reference channel as model inputs, an absorption contribution equation set is constructed based on the modified decoupling coefficient matrix; The absorption contribution equations are solved using the least squares method, and invalid interference absorption components that exceed the segmented dynamic threshold are removed to obtain an effective solution. The absorption component generated solely by black carbon in the main measurement channel is extracted from the effective solution results and determined as the pure absorption contribution value.

8. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 3, characterized in that, Based on the pure absorption contribution value, the black carbon concentration is calculated by combining the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient, including: Based on the type of interference, the mass absorption coefficient in the absorption contribution decoupling coefficient matrix is ​​adaptively corrected to obtain the correction coefficient. The initial concentration is obtained by dividing the pure absorption contribution value by the product of the correction coefficient and the optical path length of the optical system; Based on the interval of the initial concentration and the type of the interfering substance, the concentration correction coefficient is obtained by querying the correction coefficient matrix. Calculate the dual-channel absorption decoupling residual, and perform a secondary calibration on the initial concentration after correction by the concentration correction coefficient based on the dual-channel absorption decoupling residual to obtain the black carbon concentration.

9. The method for determining black carbon concentration with dual-channel threshold compensation according to claim 1, characterized in that, The formula for calculating the dual-channel absorption decoupling residual is as follows: ;in, and These are the instantaneous absorbance peak values ​​for the main measurement channel and the reference channel, respectively. and These represent the absorption contribution values ​​of black carbon in the main measurement channel and the reference channel, respectively. and These represent the absorption contribution values ​​of the interfering material in the main measurement channel and the reference channel, respectively. The formula for calculating the black carbon concentration is: ;in, This is the initial concentration. This is the residual correction factor. This is the concentration correction factor.

10. A dual-channel threshold-compensated black carbon concentration determination system, characterized in that, include: The channel signal acquisition module is used to synchronously acquire the light intensity signals and environmental parameters of the main measurement channel and the reference channel, and perform filtering, baseline correction and data calculation to obtain the instantaneous absorbance of the two channels; the environmental parameters include: ambient temperature, relative humidity and atmospheric pressure data; The channel threshold calculation module is used to calculate the segmented dynamic thresholds of the two channels respectively based on the environmental parameters and the real-time drift compensation amount, and obtain a dual-channel threshold difference vector containing the single-channel threshold and the threshold difference. The feature extraction module is used to extract feature parameters of the light intensity signal based on the dual-channel threshold difference vector and the instantaneous absorbance, and to construct a feature matrix by arranging the feature parameters in a preset order. The interference identification module is used to perform similarity matching calculations between the feature matrix and a preset standard interference feature library to obtain the interference type; The contribution quantification module is used to construct an absorption contribution model based on the type of interference by using the absorption contribution decoupling coefficient matrix and the instantaneous absorbance, and then solve the model to obtain the pure absorption contribution value. The concentration detection module is used to calculate the black carbon concentration based on the pure absorption contribution value, combined with the black carbon mass absorption coefficient, the optical path length of the optical system, and the concentration range correction coefficient.

Images