Waste gas emission remote detection system in kitchen waste treatment process

Abstract

The invention provides an exhaust emission remote detection system in a kitchen waste treatment process, and relates to the technical field of data processing, and the method comprises a calculation module which is used for calculating an exhaust emission rate according to a correction parameter, a standard data frame, a chemical reaction conversion rate of a monitoring point A and a removal rate of an incinerator purification device of a monitoring point B; calculating pollutants of the monitoring point A and pollutants of the monitoring point B to generate a pollutant correction concentration matrix; and the early warning module is used for carrying out early warning according to the pollutant correction concentration matrix, the flow of the induced draft fan, the flow of the flue and a preset emission threshold value. Through the space correction coefficient, the concentration deviation caused in the waste gas transmission process is compensated in real time, and the data reliability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a remote detection system for exhaust gas emissions during the treatment of kitchen waste. Background Art

[0002] During the treatment of kitchen waste, the accurate monitoring and early warning of exhaust gas emissions are key links to ensure environmental safety. Traditional technical solutions generally use single-point fixed monitoring devices (such as spectrometers and electrochemical sensors) to detect the concentration at the outlet of the incinerator or the ventilation opening of the workshop, and trigger an alarm based on a preset threshold. However, such methods have some defects: For example, there are elbows and variable diameter structures in the pipeline between the pretreatment workshop and the incinerator flue. Computational fluid dynamics simulations show that the concentration of pollutants decays by 12% - 18% during transmission due to the turbulence effect. However, the traditional system does not consider this spatial correction and directly uses the concentration value at the monitoring point as the basis for calculating the total emissions, resulting in an overestimation of the actual total emissions by about 15%. The temperature gradient (300°C - 850°C) in the incinerator flue affects the spectral absorption characteristics. The traditional spectrometer does not perform dynamic temperature compensation, resulting in an NO x concentration measurement error of ±10%. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a remote detection system for exhaust gas emissions during the treatment of kitchen waste, which can compensate the concentration deviation caused during the exhaust gas transmission process in real time through a spatial correction coefficient, thereby improving the reliability of the data.

[0004] To solve the above technical problem, the technical solution of the present invention is as follows: A remote detection system for exhaust gas emissions during the treatment of kitchen waste, comprising: A detection module, configured to collect the spectral signals of the exhaust gas at monitoring point A and monitoring point B in real time, analyze the spectral absorption signals of characteristic pollutants, and generate a first concentration matrix and a second concentration matrix corresponding to monitoring point A and monitoring point B respectively; A processing module, configured to multiply the first concentration matrix by the flow rate of the induced draft fan at monitoring point A at each moment to obtain a first mass matrix; multiply the second concentration matrix by the flow rate of the incinerator flue at monitoring point B at each moment to obtain a second mass matrix; splice the first mass matrix and the second mass matrix to form a combined matrix; and preprocess the combined matrix to obtain a standard data frame; The correction module is used to analyze the physical environment between monitoring point A and monitoring point B, generate a spatial correction coefficient based on the physical environment based on the principle of fluid mechanics; estimate the theoretical total amount of pollutants between monitoring point A and monitoring point B based on the design parameters and operation principle of the incinerator flue; calculate the correction factor based on the measured total amount and theoretical total amount of pollutants at monitoring point A and monitoring point B in the standard data frame; and obtain the correction parameter based on the spatial correction coefficient and the correction factor; A calculation module is used to calculate the pollutants at monitoring point A and the pollutants at monitoring point B according to the correction parameters, the standard data frame, the chemical reaction conversion rate at monitoring point A and the removal rate of the incinerator purification device at monitoring point B, so as to generate a pollutant correction concentration matrix; The early warning module is used to issue early warnings based on the pollutant correction concentration matrix, induced draft fan flow, flue flow and preset emission thresholds.

[0005] The above solution of the present invention includes at least the following beneficial effects: The processing module couples the concentration matrix with the real-time flow data (draft fan flow, flue flow) moment by moment, generates a mass matrix and splices it into a joint matrix, breaking through the limitations of traditional single-point concentration detection. Compared with the solution that only monitors the concentration value, this system can accurately reflect the actual total amount of pollutant emissions (such as mass flow) and avoid misjudgment due to flow fluctuations (for example, when the induced draft fan flow increases suddenly, the traditional system only misjudges that the emission is reduced by the decrease in concentration, while this system captures the total amount changes in real time through the mass matrix).

[0006] The detection module simultaneously collects spectral signals from the pretreatment area (monitoring point A) and the incinerator flue (monitoring point B), constructs a dual-node monitoring network, and realizes full-process tracking of exhaust gas from the source (front-end workshop) to the terminal emission (incineration flue), solving the problem that traditional single-point monitoring cannot cover complex transmission paths.

[0007] The correction module analyzes the physical environment between monitoring points (such as pipeline structure, temperature gradient, turbulence effect) based on the principles of fluid mechanics, generates a spatial correction coefficient, and compensates for the concentration deviation caused by diffusion and attenuation during exhaust gas transmission in real time. For example, the measured error of the traditional system for concentration attenuation caused by eddy currents at pipe elbows is 15%-20%, while the error of this system can be controlled within 5% after correction, which improves the reliability of the data.

[0008] The theoretical total amount of pollutants is estimated through the incinerator design parameters (such as furnace volume, residence time) and operating principles, and the correction factor is generated by comparing it with the actual total amount, forming a closed-loop mechanism of "theoretical modeling-actual measurement verification-dynamic correction". Compared with the traditional static model that relies on manual regular calibration, this system can respond to operating fluctuations in real time (such as a decrease in the removal rate of the purification device and a change in the conversion rate of the chemical reaction), avoiding long-term monitoring deviations caused by equipment aging or sudden changes in operating conditions.

[0009] The calculation module integrates the chemical reaction conversion rate at monitoring point A (such as the generation rate of volatile organic compounds in the pretreatment stage) and the removal rate of the purification device at monitoring point B (such as the real-time change of the activated carbon adsorption efficiency), and performs secondary calibration on the pollutant concentration matrix in combination with the correction parameters, so that the system can adaptively process process fluctuations (such as the change of the incinerator load and the performance attenuation of the purification equipment). The error between the corrected concentration matrix and the actual emission is less than 8%.

[0010] Based on the corrected concentration matrix and real-time flow parameters, the early warning module uses a mass flow threshold (such as the total mass of particulate matter emissions per hour) to replace the traditional single concentration threshold, avoiding early warning delays or misjudgments caused by changes in air volume. For example, when the flow rate in the incinerator flue is uneven (the flow velocity difference between the edge and the center is 30%), the total amount calculation error of the traditional system reaches 15%-20%. However, through moment-by-moment flow coupling and spatial correction, the early warning response time of this system can be shortened to within 1 minute, and the early warning accuracy rate can be increased to over 95%. Description of the Drawings

[0011] Figure 1 It is a schematic diagram of a remote detection system for waste gas emissions during the treatment of kitchen waste provided by an embodiment of the present invention.

[0012] Figure 2 It is a schematic diagram of the process flow of a remote detection method for waste gas emissions during the treatment of kitchen waste provided by an embodiment of the present invention.

[0013] Explanation of the reference numerals in the figure: 11, detection module; 12, processing module; 13, correction module; 14, calculation module; 15, early warning module. Detailed Embodiments

[0014] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art.

[0015] As Figure 1 shown, an embodiment of the present invention provides a remote detection system for waste gas emissions during the treatment of kitchen waste, including: A detection module 11 for real-time collecting the spectral signals of the waste gas at monitoring point A and monitoring point B, analyzing the spectral absorption signals of the characteristic pollutants, and respectively generating a first concentration matrix and a second concentration matrix corresponding to monitoring point A and monitoring point B; Monitoring point A is the tail gas emission outlet of the pretreatment area of the kitchen waste treatment workshop; Monitoring point B is the outlet of the incinerator flue; A processing module 12 is configured to multiply the first concentration matrix by the flow rate of the induced draft fan at the collection monitoring point A moment by moment to obtain a first mass matrix; multiply the second concentration matrix by the flow rate of the incinerator flue at the monitoring point B moment by moment to obtain a second mass matrix; splice the first mass matrix and the second mass matrix to form a combined matrix; and preprocess the combined matrix to obtain a standard data frame. A correction module 13 is configured to analyze the physical environment between the monitoring point A and the monitoring point B, generate a spatial correction coefficient based on the principle of fluid mechanics according to the physical environment; estimate the theoretical total amount of pollutants between the monitoring point A and the monitoring point B according to the design parameters and operating principle of the incinerator flue; calculate a correction factor according to the measured total amount and the theoretical total amount of pollutants at the monitoring point A and the monitoring point B in the standard data frame; and obtain a correction parameter according to the spatial correction coefficient and the correction factor. A calculation module 14 is configured to calculate the pollutants at the monitoring point A and the pollutants at the monitoring point B according to the correction parameter, the standard data frame, the chemical reaction conversion rate at the monitoring point A, and the removal rate of the incinerator purification device at the monitoring point B, so as to generate a pollutant correction concentration matrix. An early warning module 15 is configured to perform early warning according to the pollutant correction concentration matrix, the induced draft fan flow rate, the flue gas flow rate, and a preset emission threshold.

[0016] In an embodiment of the present invention, spectral signals at the tail gas emission port (monitoring point A) of the pretreatment area and the incinerator flue outlet (monitoring point B) are collected in real time, and spectral absorption signals of characteristic pollutants (such as VOCs, particulate matter, SO 2 , NO x etc.) are analyzed to generate a first / second concentration matrix. It breaks through the limitation of traditional single-point monitoring (only measuring the end flue), simultaneously captures the concentration data of the waste gas generation source (volatile pollutants in the pretreatment workshop) and the end emission (pollutants after incineration), forms a "source - transmission - end" full-link monitoring network, and avoids the supervision blind spots in the intermediate links (such as pipeline leakage and transmission attenuation).

[0017] By comparing the concentration change trends at the monitoring points A / B, unorganized emissions that are not collected in the pretreatment stage can be quickly identified (such as a sudden increase in the concentration at the monitoring point A due to pipeline damage while there is no change at the monitoring point B), and such abnormal positioning cannot be achieved by traditional single-point monitoring.

[0018] Spectral signal analysis techniques (such as Fourier transform infrared spectroscopy, differential absorption spectroscopy) can distinguish the absorption characteristics of multi-component pollutants in real time. Compared with traditional electrochemical sensors (which are vulnerable to humidity and temperature interference), the detection accuracy is improved by more than 30%. In particular, the recognition ability for low-concentration pollutants (such as ppb-level VOCs) is significantly enhanced. It dynamically generates a concentration matrix (time series data), supports concentration fluctuation analysis with second-level resolution, and solves the problem of missed detection of transient over-standard caused by traditional minute-level sampling (such as the instantaneous concentration peak when the induced draft fan starts and stops).

[0019] Traditional systems only monitor concentration (such as mg / m 3 ), without combining flow rate (m 3 / h), resulting in misjudgment of the total amount when the air volume suddenly changes (for example, when the flow rate increases by 20% but the concentration decreases by 10%, the traditional system misjudges a reduction in emissions, while this system accurately captures an 8% increase in the total amount through the mass matrix (concentration × flow rate)); it performs coupled calculations at each moment to achieve second-level synchronous mass flow tracking (such as calculating the mass of pollutant emissions once per second). Compared with traditional minute-level average calculations, the response speed to sudden operating conditions (such as a sudden change in the incinerator load) is increased by more than 60 times. Pretreatment eliminates data asynchronous errors (such as differences in the sampling frequencies of different sensors), and through timestamp alignment and format unification, it ensures the parameter matching accuracy of subsequent calculation modules, avoiding correction deviations caused by data asynchronization in traditional systems (such as calculation errors in the total amount caused by time misalignment between flow rate and concentration data); through computational fluid dynamics (CFD) or empirical formulas, it quantifies the effects of pipe elbows, diameter changes, and temperature gradients on pollutant transmission (such as concentration attenuation caused by turbulence, the effect of temperature changes on the spectral absorption coefficient), generates spatial correction coefficients (such as concentration attenuation compensation factors, spectral wavelength shift correction values), and reduces the 15%-20% monitoring error caused by the physical environment in traditional systems to within 5%.

[0020] When the treatment plant modifies the pipeline (such as adding dust removal equipment resulting in changes in the flow field), the system can automatically recalculate the spatial correction coefficient without manual recalibration; based on the design parameters of the incinerator (such as the treatment capacity, air excess coefficient) and chemical reaction equations (such as C x H y O z combustion to generate CO 2 、H 2 O), it estimates the theoretical total amount of pollutant generation, compares it with the measured total amount to generate a correction factor (such as measured total amount / theoretical total amount), forms a closed loop of "modeling prediction - measured verification - error correction", and avoids long-term deviations caused by traditional systems relying on fixed emission factors (such as empirical values) (such as the change in the actual emission factor after the incinerator ages not being recognized).

[0021] The spatial correction coefficient (physical transmission influence) and the correction factor (process theoretical deviation) are combined to form comprehensive correction parameters, which can compensate for spatial transmission errors and process operation errors at the same time. The correction accuracy is improved by 40% compared with a single correction method.

[0022] The chemical reaction conversion rate at monitoring point A (e.g., the production of NH 3 The rate of change with temperature) is connected to the calculation module in real time to correct the concentration prediction error caused by fluctuations in reaction conditions in the pretreatment stage (such as increased volatilization due to high temperatures in summer), avoiding the deviation in source emission estimation caused by traditional systems ignoring changes in front-end processes. Through the data of the purification device differential pressure sensor, online penetration rate detection, etc., the removal rate is dynamically obtained (such as automatically updating parameters when the removal rate of the activated carbon adsorption tower decreases by 1% per hour). Compared with the traditional system that relies on manual regular detection (once a week), the correction delay is shortened from 48 hours to minutes, ensuring the real-time accuracy of the terminal emission calculation.

[0023] Taking into account multi-dimensional factors such as spatial transmission (correction module), process reaction (conversion rate), purification efficiency (removal rate), etc., the original concentration matrix is ​​calibrated again, so that the final pollutant corrected concentration matrix is ​​more than 92% consistent with the actual emissions, which is higher than the 70% consistency of traditional systems that only consider concentration or a single correction parameter.

[0024] The pollutant emission mass / hour (such as kg / h) is used as the core warning indicator to replace the traditional single concentration threshold (such as mg / m 3 ), eliminating the interference of air volume fluctuation on the warning results. For example, when the uneven flue flow (high center flow velocity and low edge flow velocity) leads to single-point concentration detection deviation, the total mass calculated by the full-section flow coupling of this system can still accurately reflect the actual emission, and the warning accuracy rate is increased from 75% of the traditional system to more than 95%. When the mass emission of monitoring point A continues to increase and there is no significant change in monitoring point B, the system can identify that the collection efficiency of the pretreatment workshop has decreased (such as pipeline leakage) and trigger equipment maintenance warning in a targeted manner. Traditional single-point monitoring cannot achieve such fault location. The second-level data processing capability shortens the warning response time to within 10 seconds (traditional systems require 2-5 minutes), especially for sudden working conditions of the incinerator (such as incomplete combustion causing a sudden increase in CO concentration), which can achieve real-time interception. Combined with the corrected concentration matrix and the distribution of sensitive points around the plant (such as the distance from residential areas), it supports dynamic adjustment of the warning threshold (such as increasing sensitivity at night), realizes refined supervision, and avoids "over-warning" or "delayed warning" caused by traditional fixed thresholds.

[0025] In another preferred embodiment of the present invention, the spectral signals of the exhaust gas at the monitoring point A and the monitoring point B are collected in real time, the spectral absorption signals of the characteristic pollutants are analyzed, and the first concentration matrix and the second concentration matrix corresponding to the monitoring point A and the monitoring point B are generated respectively, including: Obtain the original spectral sequences of monitoring point A and monitoring point B respectively, and denote them as the first spectral data stream and the second spectral data stream. Each data stream consists of spectral curves with continuous timestamps, specifically including: Deploy spectral acquisition devices (such as Fourier transform infrared spectrometers) at monitoring point A (the exhaust gas outlet of the pretreatment area) and monitoring point B (the flue gas outlet of the incinerator) respectively, and continuously collect the spectral signals of the exhaust gas at a fixed time interval (such as once per second); Each spectral signal contains a complete light intensity distribution curve from the ultraviolet to the infrared band (such as in the wavelength range of 200nm - 2000nm), and mark the acquisition timestamp (accurate to milliseconds) for each spectral curve to form the first spectral data stream (monitoring point A) and the second spectral data stream (monitoring point B); For example: At 10:00:00.000, monitoring point A collects an intensity curve containing 2000 wavelength points, and the corresponding timestamp is stored in the data stream.

[0026] Perform smoothing processing, baseline correction, and wavelength calibration on the first spectral data stream and the second spectral data stream respectively to obtain the first spectral curve and the second spectral curve, specifically including: For the light intensity value sequence of each wavelength point (such as 3.2μm, 7.8μm, etc.) in the original spectral data stream (each time point corresponds to a light intensity value), with the current time point as the center, select the light intensity values of its first 5 time points, the current time point, and the last 5 time points (a total of 11 time points, and calculate according to the actual existing time points at the boundary), and calculate the average value as the smoothed light intensity value of the current time point.

[0027] For example: At the 10th time point (the middle point), take the average of the light intensity values from the 5th to the 15th time points to obtain the smoothed light intensity; At the 1st time point, only take the average of the 1st to the 6th time points (take the available points when the first 5 do not exist).

[0028] Repeat this operation for the light intensity sequences of all wavelength points to obtain the smoothed light intensity sequence of each wavelength point, and combine them into a preliminarily smoothed spectral curve (the noise is reduced and the signal jitter is weakened).

[0029] According to the average width of the characteristic absorption peak (such as through previous experiments, it is known that the absorption peak of typical pollutants occupies a width of about 5 wavelength points on the spectral curve), select a rectangular window with a larger width as the structural element (for example, set the width to 15 wavelength points to ensure that it covers multiple absorption peak periods and only retains the low-frequency drift trend). Use the above structural element to perform "erosion" processing on each wavelength point of the preliminarily smoothed spectral curve, and take the minimum light intensity value within the structural element window as the output value of the current point. This step will "flatten" the peaks (i.e., the characteristic absorption signals) in the spectral curve and only retain the baseline drift trend that may be lower than or equal to the surrounding values (such as the overall light intensity decline trend caused by light source aging).

[0030] For each wavelength point of the eroded curve, perform "dilation" processing. Take the maximum light intensity value within the structural element window as the output value of the current point. This step will restore the compressed curve after erosion to a shape close to the original low-frequency trend, and finally obtain a smooth baseline trend curve (only reflecting the overall drift of light intensity over time, such as a slow decrease or increase, ignoring the high-frequency characteristic absorption fluctuations).

[0031] According to the low-frequency drift characteristics of the preliminary smoothed spectral curve (such as the light intensity attenuation caused by light source aging approximating a quadratic curve, and the drift caused by temperature change approximating a cubic curve), select an appropriate polynomial degree (for example, 3 or 5). Take the light intensity values at all time points in the preliminary smoothed spectral curve as fitting data, and calculate the polynomial coefficients by the least squares method to minimize the sum of the squares of the errors between the polynomial curve and the spectral curve in the low-frequency part (the overall trend after removing the high-frequency absorption signal). For example, if the spectral curve shows a gradually decreasing parabolic trend within 100 time points, the fitted quadratic polynomial curve will approximate this trend, ignoring the local fluctuations caused by pollutant absorption in the middle (such as a suddenly decreasing absorption peak); according to the fitted polynomial coefficients, calculate the baseline light intensity value corresponding to each time point to form a continuous baseline trend curve (for example, represented by y = ax 3 +bx 2 +cx+d, where x is the time point and y is the baseline light intensity), in the polynomial equation y = ax 3 +bx 2 +cx+d, this is a cubic polynomial equation, and the parameters a, b, c, d each have specific meanings, as follows: The constant term d represents the intercept of the curve on the y-axis. When x = 0, y = d. In the context of spectral curve fitting, this means that at the starting moment (time point x = 0), the light intensity value corresponding to the spectral curve is d, and this value may be related to factors such as the light intensity in the initial state of the instrument and the ambient background light intensity.

[0032] Example: Suppose that when starting to collect spectral data, even without pollutant absorption, the light intensity detected by the instrument is 500 units, then d may be equal to 500.

[0033] The coefficient c of the first-order term reflects the linear change trend of light intensity over time. If c is positive, it indicates that the light intensity increases linearly over time; if c is negative, the light intensity decreases linearly over time; if c = 0, it means that there is no linear change trend in light intensity. In the spectral curve, this linear change may be caused by factors such as the slow preheating of the instrument and slight ambient light changes.

[0034] Example: If c = 2, then for each increase of one unit in time x, the light intensity y will increase by 2 units on the original basis.

[0035] The quadratic coefficient b determines the quadratic trend of the light intensity change, that is, the acceleration of the light intensity change. If b is not 0, the change of the light intensity is not linear but shows a parabolic shape. When b > 0, the parabola opens upward, meaning that the growth rate of the light intensity will gradually increase or the attenuation rate will gradually slow down; when b < 0, the parabola opens downward, indicating that the growth rate of the light intensity gradually slows down or the attenuation rate gradually increases. In the spectral curve fitting, this may be related to factors such as the change of the light source aging rate and the non-linear influence of the ambient temperature.

[0036] Example: Suppose b = -0.1, which indicates that as time goes by, the change rate of the light intensity will gradually slow down. Maybe it is because the attenuation rate of the light intensity caused by the light source aging is gradually decreasing.

[0037] The cubic coefficient a represents the cubic trend of the light intensity change, that is, the further correction of the quadratic trend. The positive or negative value and magnitude of a will affect the bending degree and direction of the curve.

[0038] Example: If a = 0.01, it will cause the curve to have a more complex bend on the basis of the quadratic trend, so as to more accurately fit the actual change of the spectral curve.

[0039] When using the least squares method to perform polynomial fitting on the low-frequency drift part of the spectral curve, the purpose is to find the most suitable values of a, b, c, and d, so that the fitted cubic polynomial curve can be as close as possible to the low-frequency drift trend of the spectral curve, and then accurately estimate the baseline.

[0040] Subtract the light intensity value of the corresponding time point on the baseline trend curve from the light intensity value of each time point in the preliminarily smoothed spectral curve. For example: At the 10th time point, the preliminarily smoothed light intensity is 850, and the light intensity of the baseline trend is 800 (reflecting the attenuation caused by the light source aging). Then the corrected light intensity is 850 - 800 = 50 (only retaining the light intensity change caused by the absorption of pollutants and eliminating the overall attenuation trend). Finally, the spectral after baseline correction is obtained, and its light intensity value only reflects the absorption signal of the characteristic pollutants (such as the light intensity fluctuation caused by the change of the VOCs concentration in the waste gas), excluding the low-frequency drift interference caused by factors such as light source aging and ambient temperature change.

[0041] Introduce a standard gas with a known characteristic absorption wavelength (such as methane, whose theoretical characteristic peak is at 3.3 μm) into the spectral acquisition device, collect its spectral curve. In the spectrum after baseline correction, find the central wavelength position of the actual absorption peak (such as determining the peak position by finding the minimum point of light intensity or the extreme point of the second derivative). Compare the actual absorption peak position with the theoretical wavelength and calculate the offset (for example, the theoretical wavelength is 3.3 μm, the measured value is 3.31 μm, and the offset is +0.01 μm). For the wavelength values of each wavelength point in the spectrum after baseline correction, uniformly subtract the offset (or adjust according to the offset ratio, such as translating the entire wavelength axis), so that the characteristic peak of the standard gas returns to the theoretical wavelength position. For example: subtract 0.01 μm from the wavelength values of all wavelength points, correct the original 3.31 μm to 3.3 μm, and the original 7.82 μm to 7.81 μm, ensuring that the characteristic wavelengths for subsequent analysis are consistent with the theory. Finally, obtain the calibrated first spectral curve (monitoring point A) and the second spectral curve (monitoring point B).

[0042] Pre - establish a characteristic absorption wavelength library for target pollutants, specifically including: for typical pollutants in the exhaust gas from food waste treatment (such as ammonia, hydrogen sulfide, VOCs, and the characteristic wavelength of particulate scattering), through literature research and standard gas experiments, pre - determine the characteristic absorption wavelength of each pollutant (such as ammonia has a strong absorption peak at 9.6 μm, and hydrogen sulfide is at 7.8 μm), and organize it into a characteristic absorption wavelength library. For example: the library contains 10 pollutants, each corresponding to 2 - 3 characteristic wavelengths, forming a set of 25 wavelength points (such as 3.2 μm, 3.4 μm, 7.8 μm, etc.).

[0043] For the first spectral curve and the second spectral curve, within the wavelength range corresponding to the characteristic wavelength library, calculate the ratio of the actual light intensity to the light intensity of pure air to obtain the absorbance values at each characteristic wavelength, specifically including: for the obtained first spectral curve and the second spectral curve, within the wavelength range specified by the characteristic absorption wavelength library in the third step (such as only extracting the light intensity values of the 25 wavelength points included in the library), extract the actual light intensity value (denoted as I) at each timestamp one by one. At the same time, pre - collect the spectral curve of the same device under pure air (without pollutants) conditions as the reference spectrum, and extract the light intensity value (denoted as I0) at the same characteristic wavelength points; for each characteristic wavelength point, calculate the ratio of the actual light intensity to the light intensity of pure air (I / I0), and then obtain the absorbance value through logarithmic conversion (the absorbance reflects the degree of light absorption by pollutants, the smaller the ratio, the larger the absorbance, and the higher the pollutant concentration). For example: at a certain moment, at the 3.2 μm wavelength of monitoring point A, the actual light intensity is 500, and the light intensity of pure air is 1000, the ratio is 0.5, and the absorbance corresponds to the absorption intensity of a specific pollutant.

[0044] According to the absorbance values at each characteristic wavelength and the pre-established pollutant concentration-absorbance calibration model, calculate the pollutant concentration vectors at each moment for monitoring point A and monitoring point B, specifically including: For each target pollutant (such as VOCs, ammonia, hydrogen sulfide, etc.), pre-configure multiple standard gas samples with different concentrations (for example, set 5 concentration gradients covering the range from 0 to the maximum monitoring concentration, such as 0 mg / m 3 , 25 mg / m 3 , 50 mg / m 3 , 75 mg / m 3 , 100 mg / m 3 ). Each group of standard gases needs to ensure uniform and stable concentration, and record its accurate nominal concentration value (provided by the standard gas supplier or calibrated by high-precision equipment).

[0045] Collect the spectral data of the standard gas and calculate the absorbance: Under the same experimental conditions (such as constant temperature, air pressure, and avoiding external interference), sequentially introduce standard gases of each concentration into the monitoring equipment and collect the spectral curves at each concentration. For each characteristic wavelength point, first collect the light intensity value of pure air (i.e., at 0 concentration) as the reference value , and then collect the actual light intensity value of the standard gas at this concentration . According to the definition of absorbance (absorbance = negative logarithm of transmittance, transmittance = ), calculate the absorbance value at this wavelength under the current concentration (for example, for a standard gas of a certain concentration at 3.2 μm , then the absorbance is ). Collect each group of standard gases of each concentration multiple times (such as 3 times), and take the average absorbance value to reduce random errors.

[0046] Establish the mapping relationship between absorbance and concentration: For the pollutant corresponding to each characteristic wavelength (such as 3.2 μm corresponding to VOCs), correspond the absorbance values of the standard gases of different concentrations at this wavelength to the nominal concentration one by one to form a set of "absorbance-concentration" data pairs (for example, for VOCs at 3.2 μm: absorbance 0.05 corresponds to 25 mg / m 3 , absorbance 0.1 corresponds to 50 mg / m 3 , absorbance 0.15 corresponds to 75 mg / m 3 , etc.). By analyzing the distribution law of the data pairs (such as observing whether it shows a linear relationship), determine the mapping rule between the two (for example, for every 0.05 increase in absorbance, the corresponding concentration increases by 25 mg / m 3 ). Finally, form the calibration model of the pollutant at the corresponding wavelength (that is, clarify the rule of "how to uniquely determine the concentration value through the absorbance value", such as absorbance 0.2 corresponds to 100 mg / m3 ).

[0047] Introduce a set of standard gas concentrations that were not involved in the modeling (such as the intermediate concentration of 40 mg / m 3 ) into the device, collect its absorbance values, calculate the predicted concentration through the calibration model, and compare it with the actual nominal concentration. The error needs to be controlled within the allowable range (such as not exceeding ±5%). If the error exceeds the limit, re-check the standard gas configuration, data collection process, or adjust the mapping rules until the model accuracy meets the standard.

[0048] Structurally store the "absorbance-concentration" mapping relationships of each pollutant at each characteristic wavelength (such as the absorbance-concentration comparison table for VOCs corresponding to 3.2 μm, the comparison table for ammonia corresponding to 4.6 μm, etc.) to form a calibration model library. When processing real-time monitoring data subsequently, the corresponding calibration model can be quickly retrieved according to the characteristic wavelength to convert the absorbance into the pollutant concentration.

[0049] Arrange the concentration vectors at each moment vertically in chronological order to form a two-dimensional concentration matrix. Each row of the two-dimensional concentration matrix corresponds to a time point, and each column corresponds to a pollutant. Specifically, it includes arranging the pollutant concentration vectors corresponding to each obtained timestamp (such as [50, 10, 20] corresponding to 10:00:00, [55, 12, 18] corresponding to 10:00:01) vertically in chronological order. Each row of the matrix represents a time point, and each column represents a pollutant, finally forming a two-dimensional table (such as 1000 rows × 10 columns, representing the concentration data of 1000 time points and 10 pollutants), that is, the first concentration matrix (monitoring point A) and the second concentration matrix (monitoring point B). For example: The first row is the concentration of each pollutant at the first time point, the second row is the concentration at the second time point, and so on, forming a dynamic concentration sequence that changes over time.

[0050] In the embodiment of the present invention, the spectral curves of consecutive timestamps completely record the transient changes of the waste gas components (such as the sudden increase in concentration when the induced draft fan starts and stops, the spectral absorption peak drift caused by the load fluctuation of the incinerator). Compared with the traditional minute-level discrete sampling, it can capture more than 95% of the instantaneous abnormal signals (such as the peak exceeding the standard lasting for 2 seconds), avoiding missed detection of sudden pollution events. The unified timestamp mechanism ensures the strict synchronization of the spectral data of monitoring point A (pretreatment area) and B (incinerator flue gas duct), providing an accurate time reference for subsequent cross-node data fusion (such as analyzing the migration law of pollutants from the source to the end), and solving the correlation analysis error caused by clock asynchronization during traditional multi-device monitoring (such as when the time misalignment is more than 30 seconds, it is impossible to accurately match the concentration and flow fluctuations).

[0051] Smoothing processing reduces the influence of environmental noise (such as instrument circuit noise, vibration interference) on spectral signals, reducing the absorbance calculation error from ±5% to ±1.5%. Especially when detecting low-concentration pollutants (such as 10 ppb-level VOCs), the signal-to-noise ratio is increased by more than 3 times, avoiding false positive / false negative results caused by noise in traditional unprocessed spectra. Baseline correction eliminates long-term drift (such as light intensity attenuation caused by light source aging, baseline shift caused by environmental temperature changes), ensuring the comparability of spectral data at different times. A practical measurement case shows that the baseline of the spectrum without baseline correction drifts by up to 8% after 8 hours of operation, resulting in concentration misjudgment, while this step can control the drift within 1%. Wavelength calibration solves the problem of wavelength drift in spectrometers (such as the deviation of the grating diffraction angle caused by temperature changes, with a typical drift of 0.05 nm / °C). It is calibrated in real time through the characteristic absorption peaks of built-in standard gases (such as methane and carbon dioxide standard gases), making the wavelength positioning accuracy reach ±0.01 nm, ensuring the accuracy of absorbance calculation at characteristic wavelengths, and avoiding misidentification of pollutants caused by wavelength deviation (such as misjudging the absorption peak of NO 2 as that of SO 2 ).

[0052] The characteristic wavelength library enables simultaneous detection of multiple components for typical pollutants in the exhaust gas of food waste treatment (such as fatty acids, ammonia, hydrogen sulfide, dioxin precursors, etc.). More than 10 pollutants can be analyzed from a single spectral curve, with the detection dimension increased by more than 5 times compared to traditional single sensors (such as only measuring particulate matter or SO 2 ), avoiding the problems of high cost and difficult maintenance caused by stacking multiple devices. Absorbance calculation excludes the interference of background gases (such as N 2 , O 2 ), only focusing on the absorption signals of characteristic pollutants. For example, in a pretreatment workshop with a humidity as high as 90%, the detection error of VOCs by traditional sensors due to water vapor interference reaches 20%. However, this step reduces the humidity influence to less than 3% by selecting characteristic wavelengths without water vapor absorption (such as the 3.3 - 3.5 μm band). Following the Lambert-Beer law, a linear relationship between absorbance and concentration is established, combined with subsequent calibration models, reducing the concentration inversion error from ±15% of traditional empirical formulas to within ±5%, especially suitable for the precise quantification of low-concentration pollutants (such as the linearity R 3 within the range of 0 - 100 mg / m 2 > 0.995).

[0053] The calibration model takes into account the spectral overlap effect of multiple components (such as the overlap of the absorption peaks of NO and NO 2 in the visible spectral region). It analyzes the mixed spectrum through chemometric methods (such as PLS) to achieve independent quantification of coexisting pollutants, with the accuracy improved by 40% compared to traditional single-wavelength detection (assuming no cross-absorption). For example, when NO and NO 2When the concentration ratio is 1:1, the error of the traditional method reaches 30%, while the error of this model is <8%.

[0054] Dynamically update the calibration model (such as automatically importing the latest calibration data every week) to adapt to the performance drift of the instrument after long-term operation (such as the attenuation of detector sensitivity), avoid the error accumulation of the traditional fixed model over time (the typical annual error increases by 10%-15%), and ensure the stable accuracy of the system during long-term operation.

[0055] The two-dimensional concentration matrix integrates the time series and multi-pollutant dimensions, provides a standardized input for the subsequent processing module (such as multiplying with the flow data at each moment to generate a mass matrix), and supports spatio-temporal correlation analysis (such as cross-verifying the VOCs concentration matrix at monitoring point A and the removal rate matrix at monitoring point B to locate the failure period of the purification device).

[0056] In another preferred embodiment of the present invention, multiplying the first concentration matrix by the flow rate of the induced draft fan at the collection monitoring point A at each moment to obtain the first mass matrix, including: Collect the flow rate data of the induced draft fan at the collection monitoring point A. The flow rate data is a sequence that changes over time, and there is a corresponding induced draft fan flow rate value at each time point. Specifically, it includes: real-time collecting the flow rate data of the induced draft fan through the flow rate sensors installed at the monitoring point A (such as vortex flow meters, pitot tube flow meters). At each time point (strictly synchronized with the time stamp of the first concentration matrix), a flow rate value is recorded to form a one-dimensional sequence (for example: the flow rate at the 1st second is 2000 m 3 / h, the flow rate at the 2nd second is 2050 m 3 / h, and so on), denoted as the flow rate time series. For example: the flow rate data is stored as [Q1, Q2, Q3,..., Qn], where Qn represents the induced draft fan flow rate corresponding to the nth row (the nth time point) of the first concentration matrix.

[0057] Starting from the first row of the first concentration matrix, the first row represents the concentrations of various pollutants at the first time point; multiply each element in the first row, that is, the concentration of each pollutant, by the induced draft fan flow rate value corresponding to the time point to obtain the mass flow rate values of various pollutants at the first time point, and form a new row vector, specifically including: Extract the data of the first row of the first concentration matrix (corresponding to the first time point). This row contains the concentration values of multiple pollutants, such as [C11, C12, C13,..., C1m] (C1m represents the concentration of the mth pollutant at the first time point, unit mg / m 3 ); extract the flow rate value Q1 at the first time point from the flow rate time series in the first step (unit m 3 / h); multiply each pollutant concentration value in this row by Q1 to obtain the mass flow rate values of each pollutant (unit mg / h): The first type of pollutant: C11×Q1; the second type of pollutant: C12×Q1; the mth type of pollutant: C1m×Q1; Arrange the calculation results in the order of the original pollutants to form a new one-dimensional array [M11, M12, M13, …, M1m], that is, the mass flow row vector at the first time point (M1j = C1j×Q1).

[0058] And so on, multiply each row of the first concentration matrix until all rows are processed. Arrange the obtained row vectors in order to form the first mass matrix. The first mass matrix is two-dimensional, with each row corresponding to a time point and each column corresponding to a type of pollutant. The elements in the matrix represent the mass flow of the pollutant at the corresponding time point, specifically including: Starting from the second row of the first concentration matrix (corresponding to the second time point), repeat the following operations until the last row: Extract the concentration data [Ci1, Ci2, …, Cim] of the ith row and the corresponding ith flow value Qi; Multiply each concentration value by Qi according to the above method to obtain the mass flow values [Mi1, Mi2, …, Mim] at the ith time point; Arrange the row vectors generated at each time point in chronological order (i.e., the row order of the first concentration matrix) in sequence, ensuring that the ith row vector corresponds to the ith row of the matrix.

[0059] Stack the generated row vectors vertically in chronological order to form a two-dimensional matrix. The number of rows of this matrix is the same as that of the first concentration matrix (equal to the total number of time points), and the number of columns is the same as the number of pollutant types. The ith row of the matrix corresponds to the ith time point, and the jth column corresponds to the jth type of pollutant. The element Mij represents the mass flow of the jth type of pollutant at the ith time point (unit: mg / h). For example: If the first concentration matrix has 1000 rows (1000 time points) and 5 columns (5 types of pollutants), then the first mass matrix is also 1000 rows × 5 columns, and each row is obtained by multiplying the concentration at the corresponding time point by the flow rate.

[0060] In the embodiment of the present invention, the concentration of the pollutant (unit volume content, such as mg / m 3 3) only reflects the relative content of the pollutant in the local gas, and combined with the flow rate of the induced draft fan (gas volume per unit time, such as m 3After (h), the obtained mass flow rate (the mass of pollutants per unit time, such as mg / h) can directly reflect the absolute mass of pollutants actually emitted or transported per unit time. Each row of the first mass matrix corresponds to the mass flow rate of pollutants at a time point, completely retaining the coupling relationship between concentration and flow rate changing with time (such as the impact of flow rate change on the actual emission mass when the rotational speed of the induced draft fan fluctuates). By analyzing the changing trends of each column (pollutant types) in the matrix over time, the time periods with high pollution loads can be accurately located (such as a sudden increase in the mass flow rate of a certain pollutant when the flow rate suddenly increases). If only the concentration is concerned, misjudgments may occur, such as "higher total emissions when the concentration is low but the flow rate is large" or "the total emissions may not be prominent when the concentration is high but the flow rate is small". The mass flow rate directly reflects the result of the combined action of the two through the product of concentration and flow rate, avoiding the one-sidedness of a single indicator. For example: At monitoring point A, the concentration of a certain pollutant at time point 1 is 50 mg / m 3 and the flow rate is 1000 m 3 / h, and the mass flow rate is 50000 mg / h; at time point 2, the concentration drops to 40 mg / m 3 but the flow rate rises to 1500 m 3 / h, and the mass flow rate is even higher (60000 mg / h). Such changes can be visually identified through the mass matrix, avoiding decision-making biases caused by only looking at the concentration.

[0061] In another preferred embodiment of the present invention, the second concentration matrix is multiplied by the flow rate of the incinerator flue at monitoring point B at each moment to obtain a second mass matrix, including: Obtain the flow rate data of the incinerator flue at monitoring point B. The flow rate data of the incinerator flue is a sequence that changes with time, and each time point has a corresponding flue flow rate value. Specifically, it includes: Through flow rate monitoring devices (such as vortex flow meters, thermal mass flow meters) installed at monitoring point B (the outlet of the incinerator flue), the flue flow rate data is collected in real time. A flow rate value is recorded at each time point (strictly aligned with the time stamp of the second concentration matrix), forming a one-dimensional sequence (for example: the flow rate at the 1st second is 15000 m 3 / h, the flow rate at the 2nd second is 15200 m 3 / h, and so on), denoted as the flue flow rate time series. This sequence corresponds one-to-one with the time points of the second concentration matrix to ensure the time synchronization of concentration and flow rate during subsequent calculations (such as the concentration data at the i-th time point corresponds to the i-th flow rate value).

[0062] Starting from the first row of the second concentration matrix, multiply each element in the first row, that is, the concentration of each pollutant, by the incinerator flue gas flow value corresponding to the first time point to obtain the mass flow values of various pollutants at the first time point, forming a row vector. Specifically, extract the data of the first row of the second concentration matrix (corresponding to the first time point), which contains the concentration values of multiple pollutants, such as [C'11, C'12, C'13, …, C'1m] (C1j' represents the concentration of the jth pollutant at the first time point, with the unit mg / m 3 ) Extract the flow value Q1' at the first time point from the flue gas flow time series in the first step (with the unit m 3 / h). Multiply each pollutant concentration value in this row by Q1' to obtain the mass flow values of each pollutant (with the unit mg / h): For the first pollutant: C11' × Q1'; For the second pollutant: C12' × Q1'; For the mth pollutant: C1m' × Q1'; Arrange the calculation results in the original pollutant order to form a mass flow row vector [M'11, M'12, M'13, …, M'1m] at the first time point, where each element represents the mass flow of the corresponding pollutant at this time point.

[0063] Perform the same operation on the second row of the second concentration matrix, multiply its elements by the flue gas flow value at the second time point to obtain a mass flow row vector at the second time point. Specifically, starting from the second row of the second concentration matrix (corresponding to the second time point), process each row in chronological order:[[]] Extract the concentration data [C'i1, C'i2, …, C'im] of the ith row and the corresponding ith flue gas flow value Qi', repeat the above operation, multiply each concentration value by Qi' to obtain the mass flow values [M'i1, M'i2, …, M'im] at the ith time point, ensuring that each generated row vector is strictly arranged in chronological order (that is, the ith row vector corresponds to the ith time point, which is exactly the same as the row order of the second concentration matrix).

[0064] And so on, after processing all rows of the second concentration matrix, arrange the obtained row vectors in order to obtain the second mass matrix. The second mass matrix is two-dimensional, each row represents a time point, each column represents a pollutant, and the element is the mass flow of the pollutant at the time point. Specifically:[[]] Vertically stack all the generated row vectors in chronological order to form a two-dimensional matrix, and the structure of this matrix is the same as that of the second concentration matrix:[[]] Number of rows: equal to the total number of time points (the same as the number of rows in the second concentration matrix).

[0065] Number of columns: Equal to the number of pollutant types (same as the number of columns in the second concentration matrix).

[0066] Element Mij': Represents the mass flow rate of the jth pollutant at the ith time point (unit: mg / h), which is obtained by multiplying the concentration value Cij' at the corresponding time point by the flue gas flow rate value Qi'.

[0067] For example: If the second concentration matrix has 2000 rows (2000 time points) and 6 columns (6 pollutant types), then the second mass matrix is also 2000 rows × 6 columns. Each element in each row is the product result of the concentration and flow rate at the corresponding time point.

[0068] In the embodiment of the present invention, by combining the flue gas flow rate (reflecting the real-time volumetric flow rate of the incinerator exhaust gas) and the concentration data, the concentration of the end pollutants (mg / m 3 ) is converted into the mass flow rate (mg / h), which directly reflects the absolute mass of the pollutants actually emitted per unit time at the incinerator outlet, avoiding the problem of "overestimating the total emissions at high concentrations and low flow rates and underestimating the total emissions at low concentrations and high flow rates" caused by relying solely on concentrations. The coupling relationship between the flow rate and concentration in the time series is retained (such as the immediate impact of the change in the flue gas flow rate on the mass emissions when the incinerator load fluctuates), supporting the analysis of the dynamic laws of pollutant generation and emissions during the incineration process (such as the sudden increase in the mass flow rate of incomplete combustion products due to a sudden drop in the furnace temperature), providing data support for the operation optimization of the incinerator. As the mass flow rate data of the end monitoring point (Point B), it can be combined with the mass matrix of the front-end monitoring point (Point A) to calculate the pollutant removal rate (such as "(mass flow rate at Point A - mass flow rate at Point B) / mass flow rate at Point A"), and the treatment effect of the incinerator purification device (such as activated carbon adsorption, bag dust removal) can be evaluated in real time, solving the problem that traditional single-point concentration detection cannot quantify the purification efficiency. It forms a unified "time-pollutant-mass flow rate" data structure with the first mass matrix (emissions from the pretreatment area), facilitating subsequent splicing into a joint matrix to achieve full-link mass emission tracking from the exhaust gas generation (front-end workshop) to the end treatment (incinerator flue).

[0069] In another preferred embodiment of the present invention, the first mass matrix and the second mass matrix are spliced to form a joint matrix; the joint matrix is preprocessed to obtain a standard data frame, including: The first mass matrix and the second mass matrix are spliced column by column, that is, all columns of the first mass matrix are placed in front, and all columns of the second mass matrix are placed behind to form a new two-dimensional matrix, namely the joint matrix. Each row of the joint matrix corresponds to a time point. The front columns correspond to the pollutant mass flow rates of monitoring point A, and the rear columns correspond to the pollutant mass flow rates of monitoring point B, specifically including: Confirm that the number of rows of the first mass matrix (monitoring point A, denoted as MA) and the second mass matrix (monitoring point B, denoted as MB) is the same (both equal to the total number of time points, e.g., n rows), and the timestamps of each row are strictly aligned (e.g., the i-th row corresponds to the i-th time point); arrange all columns of MA (e.g., m columns, corresponding to the mass flows of m pollutants at monitoring point A) in the original order on the left side of the combined matrix, and then sequentially splice all columns of MB (e.g., k columns, corresponding to the mass flows of k pollutants at monitoring point B) to form a two-dimensional matrix with n rows × (m + k) columns, denoted as the combined matrix; the first to m-th columns of the combined matrix correspond to the pollutants at monitoring point A (such as VOCs, ammonia, etc.), and the (m + 1)-th to (m + k)-th columns correspond to the pollutants at monitoring point B (such as NO x 、particulate matter, etc.); each row still represents a time point. For example, the first m elements of the i-th row are the mass flows of various pollutants at the i-th time point of monitoring point A, and the last k elements are the corresponding data of monitoring point B.

[0070] Filter each element in the combined matrix through moving average filtering, and perform normalization processing on the filtered combined matrix so that the values of each column of data are mapped to the interval [0, 1] to obtain a standard data frame, specifically including: Select a sliding window that includes the current time point and several time points before and after (such as including the first 2, the current, and the last 2 time points, a total of 5 time points) to calculate the average value to smooth the data; for each column of the combined matrix (regardless of whether it belongs to monitoring point A or B, each column corresponds to the mass flow of a pollutant at all time points), starting from the first time point, calculate the average value of the elements within the window in sequence; for boundary time points (such as the first time point, without the first 2 time points), use the available time points to calculate the average (such as only the first 1, the current, and the last 2 time points, a total of 4 points). For example: for the element at the i-th time point and the j-th column, the filtered value is the average value of the elements in the j-th column within the window (such as from the (i - 2)-th to the (i + 2)-th time points); after all elements are filtered, a matrix with the same number of rows and columns as the combined matrix is obtained, in which high-frequency noise (such as instantaneous jitter of the sensor) is suppressed and the data curve is smoother.

[0071] For each column (a total of m + k columns) of the filtered matrix, find the maximum value (max j )and the minimum value (min j )of this column, where j represents the column index (1 ≤ j ≤ m + k), and linearly map it to the interval [0, 1]: for each element of each column, convert it according to the following rules: If max j = min j , the new value = (current value - min j ) / (max j - min j ); if maxj = min j (i.e., there is no change in the data of this column, which is a very special case), the default value is assigned as 0.5 (or the original value is maintained, and it is processed according to actual requirements). After all columns are normalized, the final two-dimensional matrix, that is, the standard data frame, is obtained. The data range of each column is [0, 1], eliminating the analysis deviation caused by differences in units and magnitudes of the mass flow rates of different pollutants (for example, the VOCs mass flow rate at monitoring point A is 1000 mg / h, and the particulate matter at monitoring point B is 10 mg / h. After normalization, their weights in the model are the same). Each row still corresponds to a time point, integrating the smoothed and dimensionless mass flow rate data of the two monitoring points, providing a unified input for the subsequent correction module and calculation module.

[0072] In the embodiment of the present invention, a joint matrix is formed by column splicing, unifying the pollutant mass flow rate data of the pretreatment area (point A) and the incinerator flue (point B) to the same time dimension, facilitating direct comparison of their dynamic correlations (for example, when the mass flow rate of a certain pollutant at point A suddenly increases, whether the corresponding pollutant at point B changes synchronously to determine whether the purification device fails), and solving the problem of missing correlations caused by traditional independent analysis of data at two nodes. Moving average filtering reduces the high-frequency noise of the data (such as instantaneous jitter of sensors and short-term disturbance of airflows), making the mass flow rate curve in the joint matrix smoother and avoiding interference of outliers on the correction module and calculation module (such as misinterpreting the noise peak as a real emission mutation), improving the reliability of subsequent modeling (such as fluid mechanics correction and purification efficiency calculation). Mapping the mass flow rate data of different pollutants and different monitoring points to the [0, 1] interval, eliminating the analysis deviation caused by different units and magnitudes (for example, the VOCs mass flow rate is 1000 mg / h, and ammonia is 10 mg / h. After normalization, their weights in the model are balanced), providing a standardized input for subsequent machine learning algorithms (such as calculation of correction factors and optimization of warning thresholds), and improving the model training efficiency and generalization ability.

[0073] In another preferred embodiment of the present invention, the physical environment between monitoring point A and monitoring point B is analyzed, and based on the principle of fluid mechanics, a spatial correction coefficient is generated according to the physical environment, including: Obtain the physical parameters of monitoring point A and monitoring point B. The physical parameters include the straight-line distance, vertical height difference, total pipeline length, pipeline inner diameter, pipeline inner wall roughness, and pipeline bending characteristics of monitoring point A and monitoring point B in three-dimensional space. Specifically, include: through engineering design drawings or on-site surveys, record the three-dimensional spatial positions of monitoring point A (tail gas emission port of the pretreatment area) and monitoring point B (incinerator flue outlet): Straight-line distance: The shortest distance between two points in space (such as 10 meters); Vertical height difference: The height difference between two points in the vertical direction (for example, monitoring point B is 5 meters higher than A, recorded as +5 meters; otherwise, recorded as -5 meters).

[0074] Measure the parameters of the exhaust gas pipeline connecting two points: Total pipeline length: The total geometric length including all straight pipe sections, elbows, and diameter-changing sections (such as 20 meters).

[0075] Inner diameter of the pipeline: The inner diameter of each straight pipe section (if the pipeline changes diameter, take the inner diameter of the main flow section, such as 0.8 meters).

[0076] Inner wall roughness: The roughness of the inner wall of the pipeline (such as obtained through a roughness measuring instrument or by looking up a table, recorded as 0.1 mm).

[0077] Bending characteristics: Record the number and geometric parameters of non-straight pipe sections (such as 3 90° elbows, with a curvature radius of 1.2 meters for each elbow; 1 45° elbow, with a curvature radius of 0.8 meters).

[0078] Finally, form a physical parameter table containing the above parameters.

[0079] Based on the physical parameters, calculate the frictional resistance along the way when the exhaust gas flows in the straight pipe, and calculate the local resistance in the non-straight pipe section components, specifically including: According to the inner diameter of the pipeline and the inner wall roughness, evaluate the frictional resistance between the fluid and the pipe wall when the fluid flows in the straight pipe (such as the rougher the inner wall and the longer the pipeline, the greater the resistance); according to the principles of fluid mechanics, divide the straight pipe section into segments by inner diameter (if there is a diameter change, divide it into multiple segments), and calculate the resistance for each segment (for example, a 10-meter long straight pipe with an inner diameter of 0.8 meters and a roughness of 0.1 mm, its frictional resistance along the way is manifested as a continuous pressure loss to the air flow, resulting in a slightly decreased flow velocity, and pollutants may partially settle or react due to an increased residence time).

[0080] Local resistance calculation (non-straight pipe section): For each bending characteristic (such as an elbow), according to its angle (90°, 45°) and curvature radius, look up a table or determine the local resistance coefficient according to experience (for example, when the curvature radius of a 90° elbow is 1.2 meters, the local resistance coefficient is 0.3; the coefficient of a 45° elbow is 0.15).

[0081] Accumulate the local resistances of all non-straight pipe sections (such as the total coefficient of 3 90° elbows is 0.9, and the coefficient of 1 45° elbow is 0.15, with a total local resistance coefficient of 1.05), which reflects the instantaneous resistance generated when the air flow passes through components such as elbows and diameter changes (such as energy loss caused by eddies, which may cause local fluctuations or attenuation of pollutant concentration), and finally obtain the total frictional resistance along the way and the total local resistance, which are key parameters for measuring the air flow resistance in the pipeline.

[0082] Generate a spatial correction coefficient according to the frictional resistance along the way, local resistance, and height difference, specifically including: The frictional resistance and local resistance together determine the total effect of the flow resistance of the exhaust gas from point A to point B (the greater the resistance, the slower the air flow velocity, and the longer the residence time of pollutants in the pipeline, which may lead to concentration attenuation due to sedimentation, adsorption or chemical reactions).

[0083] For example, if the total resistance is large, assuming the attenuation rate of pollutants in the pipeline is 10%, the correction factor needs to reflect this attenuation (for example, the correction factor of 0.9 means that the measured concentration at point B needs to be multiplied by 0.9 to compensate for the attenuation).

[0084] The vertical height difference affects the additional pressure generated by the air flow due to gravity (for example, when point B is higher than point A, the air flow needs to overcome gravity, which may lead to a slight decrease in the flow velocity and indirectly affect the pollutant transmission efficiency). Combining the direction (rising or falling) and magnitude of the height difference, the correction factor is adjusted (for example, when the height difference is +5 meters, the correction factor is slightly less than that of the horizontal pipeline; when it is -5 meters, it is slightly greater than the horizontal pipeline).

[0085] Pre-establish the pollutant attenuation rates corresponding to different resistance levels (characterized by the sum of the frictional resistance and local resistance) through historical data. For example: Low resistance (total resistance coefficient < 0.5): attenuation rate of 5% (that is, 5% of the pollutants are lost due to sedimentation, adsorption, etc. during pipeline transmission, and the remaining 95% reaches the monitoring point B); Medium resistance (0.5 ≤ total resistance coefficient < 1.0): attenuation rate of 10%; High resistance (total resistance coefficient ≥ 1.0): attenuation rate of 15%.

[0086] Match the current resistance level. According to the calculated sum of the frictional resistance and local resistance (for example, the total resistance coefficient is 1.05), match the corresponding attenuation rate (such as the high resistance level with an attenuation rate of 15%), and convert it into an attenuation effect factor (that is, the remaining ratio, 1 - attenuation rate = 0.85).

[0087] Convert the height difference into a pressure effect factor. The processing process is as follows: Establish the height difference - pressure correction rule: Preset the influence rule of the height difference on the air flow pressure (considering the additional effect of gravity on the air flow): When the monitoring point B is higher than A (positive height difference): For every 1 - meter height difference, introduce a pressure loss of 0.1%, and the correction factor = 1 - 0.001 × height difference (meters); When the monitoring point B is lower than A (negative height difference): For every 1 - meter height difference, introduce a pressure gain of 0.05%, and the correction factor = 1 + 0.0005 × |height difference| (meters); Horizontal with no height difference (0 meters): correction factor = 1.

[0088] Calculate the current pressure effect factor: For example, if the vertical height difference is +5 m (B is higher than A), then the pressure effect factor = 1 - 0.001 × 5 = 0.995; if it is -3 m (B is lower than A), then the factor = 1 + 0.0005 × 3 = 1.0015.

[0089] The composite space correction coefficient is processed as follows: Determine the composite rule: Preset the resistance attenuation and pressure effect as independent influencing factors, and use multiplicative composition (i.e., the total correction coefficient = attenuation effect factor × pressure effect factor) to reflect the pollutant transmission efficiency under the combined action of both.

[0090] Example calculation: If the attenuation effect factor is 0.85 (15% attenuation) and the pressure effect factor is 0.995 (B is 5 m higher than A), then the space correction coefficient = 0.85 × 0.995 ≈ 0.845 (indicating that the pollutant concentration at monitoring point A needs to be multiplied by 0.845 to be corrected to the theoretical concentration at point B considering pipeline resistance and height difference); if the attenuation effect factor is 0.95 (5% attenuation) and the height difference is 0 m (pressure effect factor = 1), then the correction coefficient = 0.95 × 1 = 0.95. If the composite correction coefficient exceeds the reasonable range (such as <0.5 or >1.5, in extremely special cases), then take the nearest boundary value (such as 0.5 or 1.5) to avoid interference from extreme values on subsequent calculations.

[0091] This coefficient is used in the subsequent correction module to compensate for the concentration deviation in the pollutant transmission process between monitoring points A / B (such as after the concentration matrix at point A is adjusted by the correction coefficient, it is closer to the concentration that should be detected theoretically at point B).

[0092] In the embodiment of the present invention, by quantifying the influence of factors such as pipeline structure (such as bending, diameter change), inner wall roughness, and height difference on the exhaust gas flow (such as concentration attenuation caused by frictional resistance along the way, and turbulent mixing caused by local resistance), the generated space correction coefficient can compensate in real time for the concentration deviation caused by the physical environment between monitoring points (such as at the pipeline elbow, the measured concentration is 15%-20% lower than the actual value due to eddy currents), avoiding the total amount misjudgment caused by traditional monitoring ignoring transmission losses. The calculation process based on the principles of fluid mechanics (such as the frictional resistance formula along the way, and looking up the local resistance coefficient table) enables the system to automatically adapt to the pipeline layouts of different treatment plants (such as differences in the straight pipe length and bending angle), without the need for manual presetting of correction parameters. When the pipeline is modified (such as adding dust removal equipment to change the flow field), the correction coefficient can be adjusted in real time by updating physical parameters, solving the problem that traditional fixed models cannot adapt to changes in working conditions. For example, by combining the mass attenuation rate caused by frictional resistance along the way, the theoretical total amount of pollutants transmitted from monitoring point A to B can be calculated more accurately, providing a quantitative basis for the calculation of the correction factor at the space transmission level.

[0093] In another preferred embodiment of the present invention, calculating a correction factor based on the measured total amount and the theoretical total amount of pollutants at monitoring point A and monitoring point B in the standard data frame, includes: Extracting the measured total amount of pollutants at monitoring point A and monitoring point B from the standard data frame, specifically including: According to the splicing rule of the joint matrix, the first m columns of the standard data frame correspond to the mass flow rate of pollutants at monitoring point A (such as VOCs, ammonia, etc.), and the last k columns correspond to the mass flow rate of pollutants at monitoring point B (such as NOx, particulate matter, etc.). Calculate the measured total amount at each time point: For each row (each time point) of the standard data frame, add the values of the first m columns respectively to obtain the measured total amount of pollutants at monitoring point A at this time point (that is, the sum of the mass flow rates of all pollutants at point A at this time point). Similarly, add the values of the last k columns to obtain the measured total amount of pollutants at monitoring point B at this time point. Store the measured total amounts at point A and point B at each time point into two one-dimensional arrays respectively (such as measured total amount A = [S1, S2,..., Sn], measured total amount B = [S1, S2,..., Sn]), where n is the total number of time points.

[0094] Calculating the ratio of the measured total amount to the theoretical total amount to obtain the correction factor, specifically including: According to the "theoretical total amount of pollutants estimated based on the design parameters and operating principle of the incinerator" in the correction module (this theoretical total amount can be regarded as the theoretical transmission total amount of pollutants between monitoring point A and B, which is a fixed value, for example, the theoretical total amount Ti at each time point is obtained through material balance calculation).

[0095] Calculating the ratio at each time point: For each time point i, compare the measured total amount (assuming the measured total amount here is the sum of point A / B; according to the user's description "the measured total amount of pollutants at monitoring point A and monitoring point B", it is assumed to be the sum of the two here) with the corresponding theoretical total amount Ti, and arrange the ratios at each time point in order to form a correction factor sequence for calculating the final correction parameter in combination with the spatial correction coefficient later.

[0096] In the embodiment of the present invention, by extracting the measured total amount and comparing it with the theoretical total amount, the actual monitoring data of pollutant emissions is directly associated with the output of the theoretical model. The ratio of the measured total amount to the theoretical total amount (correction factor) can reflect the differences between the actual emission characteristics of the monitoring point and the theoretical assumptions, such as the unquantified impacts of pipeline resistance, environmental factors, etc., thereby providing parameters for dynamic adjustment for subsequent data correction; as a dimensionless index, the correction factor can unify the differences in the total amount of pollutants under different monitoring points and different time scales.

[0097] In another preferred embodiment of the present invention, obtaining a correction parameter according to the spatial correction coefficient and the correction factor, includes: Taking monitoring point A as the coordinate origin (0, 0), the coordinates of monitoring point B in the two-dimensional plane are (L, 0), where L is the straight-line distance between the two points; according to the actual path of the waste gas transmission pipeline, the transmission area between monitoring points A and B is divided into several sub-regions, and each sub-region corresponds to different components in the spatial correction coefficient, specifically including: Read the three-dimensional straight-line distance (such as 10 meters) between monitoring points A and B from the obtained physical parameter table and denote it as L.

[0098] Establish a two-dimensional coordinate system: Taking the position of monitoring point A as the coordinate origin, defined as (0, 0); set the position of monitoring point B along the horizontal straight-line direction (ignoring the vertical height difference, only for plane division) as (L, 0), forming a two-dimensional coordinate system with A as the origin and AB as the x-axis; through the pipeline layout drawing or on-site measurement, record the actual transmission path of the waste gas from A to B (for example: starting from point A, first passing through a 5-meter-long straight pipe (inner diameter 0.8 meters, roughness 0.1 millimeter); then passing through a 90° elbow (curvature radius 1.2 meters); then connecting a section of 8-meter-long variable-diameter straight pipe (inner diameter shrinking from 0.8 meters to 0.6 meters, roughness 0.08 millimeter); finally passing through a 45° elbow (curvature radius 0.8 meters) to reach point B.

[0099] Divide the sub-regions by component type: Sub-region 1: The first straight pipe section after point A (5 meters long, inner diameter 0.8 meters), marked as "Straight pipe section 1"; Sub-region 2: The first 90° elbow (curvature radius 1.2 meters), marked as "Elbow 90°-1"; Sub-region 3: The variable-diameter straight pipe section (8 meters long, inner diameter 0.6 meters), marked as "Variable-diameter straight pipe section"; Sub-region 4: The second 45° elbow (curvature radius 0.8 meters), marked as "Elbow 45°-1".

[0100] Record the geometric parameters of the sub-regions: Mark the starting and ending coordinates for each sub-region (based on the projection of the actual path in the two-dimensional coordinate system, such as the starting point of the elbow is (5, 0) and the ending point is (5, 1.2), simulating an upward bend), as well as parameters such as length, angle, and inner diameter.

[0101] Mark the components of the spatial correction coefficient, and the processing process is as follows: Associate physical parameters with correction components: Straight pipe sections (Sub-region 1, Sub-region 3): Correspond to the "friction factor" component, which is determined by the inner diameter, length, and inner wall roughness of the pipe (such as the roughness of Sub-region 1 is 0.1 millimeter, the inner diameter is 0.8 meters, and the friction has a greater impact; the inner diameter of Sub-region 3 shrinks, and the friction further increases).

[0102] Elbow (Sub-region 2, Sub-region 4): Corresponding to the "local resistance coefficient" component, which is determined by the bending angle (90° or 45°) and the curvature radius (for example, Sub-region 2 is a 90° elbow with a curvature radius of 1.2 meters, and its local resistance coefficient is higher than that of the 45° elbow in Sub-region 4).

[0103] Height difference influence section: If there is a vertical height change in a certain sub-region (such as rising from an altitude of 10 meters at point A to an altitude of 15 meters at point B), then an additional "height difference pressure coefficient" component is marked (in this example, it is assumed that Sub-region 3 has a 5-meter rising height and the gravity influence needs to be corrected).

[0104] Sample the spatial points in each sub-region and calculate their distances to monitoring points A and B, specifically including: For a straight pipe section with a length of S meters (such as Sub-region 1 is 5 meters long), sampling points are evenly selected at intervals of 1 meter (including the starting point and the ending point); taking Sub-region 1 as an example, the starting point coordinates are (0, 0) (coinciding with point A), the ending point coordinates are (5, 0), then the sampling point coordinates are successively (1, 0), (2, 0), (3, 0), (4, 0), (5, 0) (a total of 5 points, including the ending point); for a variable-diameter straight pipe section (such as Sub-region 3 is 8 meters long with an inner diameter change), it is divided equally according to the length, and the coordinates of each point are recorded (such as the starting point (5, 1.2), the ending point (13, 1.2), and the sampling points are (6, 1.2), (7, 1.2), …, (13, 1.2).

[0105] For a 90° elbow (with a curvature radius of 1.2 meters, the center coordinates (5, 1.2), the starting point (5, 0), and the ending point (6.2, 1.2)), the arc is evenly divided into 5 equal parts (each part is 18°) according to the curvature.

[0106] Starting from the starting point, calculate the coordinates of each equal division point along the arc counterclockwise (or clockwise): The 1st point: At a distance of 1 / 5 of the arc from the starting point, the coordinates are calculated through the central angle (such as the central angle is 18°, and the coordinates are 5 + 1.2×cos(18°), 0 + 1.2×sin(18°)); And so on until the last equal division point before the ending point, a total of 5 points (adjustment is required when including the starting point and the ending point to ensure uniform distribution).

[0107] Store the coordinates of the sampling points in each sub-region into a list (such as the sampling point coordinates of the elbow in Sub-region 2 are (5.38, 0.38), (5.71, 0.71), (6, 1.2), (5.71, 1.62), (5.38, 1.94), etc., and the specific coordinates are calculated according to the actual curvature).

[0108] For each sampling point (x, y), calculate its straight-line distance (i.e., Euclidean distance) to the coordinate origin (0, 0). For example, the distance from the point (5, 1.2) to point A is: horizontal distance 5 meters, vertical distance 1.2 meters, and the distance is "the square root of the sum of the squares of 5 meters horizontal distance and 1.2 meters vertical distance" (avoid using formulas, describe the geometric meaning in words).

[0109] Distance to point B, dB: Given that the coordinates of point B are (L, 0) (e.g., L = 10 meters), for the sampling point (x, y), calculate its straight-line distance to (10, 0). For example, the distance from the point (5, 1.2) to point B is: horizontal distance 10 - 5 = 5 meters, vertical distance 1.2 meters, and the distance calculation method is the same as above.

[0110] Record dA and dB for each sampling point (e.g., for the point (3, 0) in sub-region 1: dA = 3 meters, dB = 7 meters; for the point (5.71, 0.71) at the elbow: dA ≈ 5.75 meters, dB ≈ 4.31 meters).

[0111] Multiply the spatial correction coefficient component of each sub-region by the corresponding weight to obtain the contribution value of this component to the correction parameter, specifically including: The weight rule processing process according to the sub-region type is as follows: Straight pipe section / stepped straight pipe section weight: Calculate the total length of all straight pipe sections (including stepped sections), and use the length ratio of each straight pipe section as the weight (e.g., total pipeline length 20 meters, sub-region 1 (straight pipe section) length 5 meters, weight is 5 / 20 = 0.25; sub-region 3 (stepped straight pipe section) length 8 meters, weight is 8 / 20 = 0.4).

[0112] If the inner diameter change of the stepped section is significant (e.g., the inner diameter suddenly decreases from 0.8 meters to 0.6 meters), an additional weight coefficient is added (e.g., multiply by 1.2 on the basis of the length ratio, and the final weight is 0.4 × 1.2 = 0.48).

[0113] Elbow / non-straight pipe section weight: Preset angle weight factor: 1.0 for 90° elbow, 0.6 for 45° elbow, 1.5 for 180° elbow (the greater the angle, the higher the resistance effect).

[0114] Adjustment in combination with the radius of curvature: For elbows with a radius of curvature < 1 meter, the weight factor is multiplied by 1.1 (smaller radius of curvature increases resistance); if the radius of curvature ≥ 1 meter, the original value is maintained.

[0115] For example: The weight factor for sub-region 2 (90° elbow, radius of curvature 1.2 meters) is 1.0 × 1.0 = 1.0; the weight factor for sub-region 4 (45° elbow, radius of curvature 0.8 meters) is 0.6 × 1.1 = 0.66.

[0116] Height difference section weight: Set according to the proportion of the absolute value of the vertical height difference to the total transmission height difference (for example, if the total height difference is 5 meters and a sub-region accounts for 3 meters, the weight is 3 / 5 = 0.6). The weight of the ascending section is slightly higher than that of the descending section (for example, an additional 0.1 is added to the ascending section, and the final weight is 0.7).

[0117] Calculate the weight value of the sub-region. The processing process is as follows: Example of a straight pipe section (sub-region 1, 5 meters long, total pipe length 20 meters): Weight = length proportion = 5 / 20 = 0.25 (if it is an ordinary straight pipe without diameter change, it is directly adopted).

[0118] Example of a stepped straight pipe section (sub-region 3, 8 meters long, inner diameter from 0.8 → 0.6 meters): Basic weight = length proportion = 8 / 20 = 0.4; Inner diameter change correction: Since the inner diameter decreases by 25%, the weight × 1.2 → 0.4 × 1.2 = 0.48; Example of an elbow (sub-region 2, 90° elbow, curvature radius 1.2 meters): Angle weight factor = 1.0 (90°); The curvature radius ≥ 1 meter, no additional correction is required → weight factor = 1.0; Combined with the number of elbows in the total non-straight pipe sections (a total of 2 elbows), the normalized weight = 1.0 / (1.0 + 0.66) ≈ 0.61 (assuming the weight factor of sub-region 4 is 0.66 and the total factor sum is 1.66).

[0119] Calculate the contribution value of the sub-region. The processing process is as follows: Extract the correction coefficient component, obtained from the pre-marked sub-region - correction component comparison table (for example: sub-region 1 (straight pipe section) corresponds to "friction factor 0.3"; sub-region 2 (elbow) corresponds to "local resistance coefficient 0.15").

[0120] Multiply to calculate the contribution value: Contribution value of sub-region 1 = friction factor × weight = 0.3 × 0.25 = 0.075 (reflecting the contribution of this straight pipe section to the friction resistance).

[0121] Contribution value of sub-region 2 = local resistance coefficient × weight = 0.15 × 0.61 ≈ 0.0915 (reflecting the contribution of this elbow to the local resistance).

[0122] Accumulate the contribution values of all components by physical characteristics, and finally form correction parameters, specifically including: classify the contribution values of all sub-regions into several categories (for example: frictional resistance along the path category: includes the contribution values of all straight pipe segments, variable diameter segments, etc. that generate continuous frictional resistance; local resistance category: includes the contribution values of all elbows, valves, etc. that generate instantaneous resistance; height difference category: includes the contribution value of the pressure effect caused by the vertical height difference (if the sub-region involves height changes).

[0123] Classification and accumulation: Sum the contribution values of each category separately (for example, the total contribution of the frictional resistance along the path category = the contribution of sub-region 1 + the contribution of sub-region 3, and the total contribution of the local resistance category = the contribution of sub-region 2 + the contribution of sub-region 4).

[0124] Merge the accumulated results of each category according to the preset logic (for example, the contribution of frictional resistance along the path accounts for 60% weight, the local resistance accounts for 30%, and the height difference accounts for 10%), and finally form one or more dimensionless correction parameters (such as correction parameter 1 reflects the total resistance effect, and correction parameter 2 reflects the height difference effect) for subsequent pollutant concentration correction.

[0125] In the embodiment of the present invention, by dividing the transmission region between monitoring points A and B into several sub-regions and combining the actual bending path of the exhaust gas pipeline (instead of the simple straight-line distance), the differential effects of different physical environments (such as pipeline bending, roughness change, height difference) on the exhaust gas transmission can be accurately captured. For example, for sub-regions with dense elbows or rough inner walls, the spatial correction coefficient component can reflect higher local resistance or frictional resistance along the path, making the model closer to the true transmission characteristics of the complex pipeline system.

[0126] Sample the spatial points of each sub-region and calculate the distances to A and B. Combine the spatial correction coefficient components (such as the frictional resistance coefficient along the path, local resistance coefficient) with the corresponding weights (such as setting weights according to the pipeline length, bending angle), and the contributions of different physical factors (such as the resistance of the straight pipe segment, the resistance of the elbow, the height difference pressure) to the pollutant transmission can be quantitatively evaluated. For example, the weight of the frictional resistance component of the long straight pipe segment is higher, while the weight of the local resistance component of the sharp bend pipeline is higher, enabling the correction parameter to dynamically reflect the dominant physical mechanism of each sub-region and improve the correction accuracy.

[0127] Classify and accumulate the contribution values according to physical characteristics (such as frictional resistance along the path, local resistance, height difference pressure effect), so that the correction parameter has clear physical meaning and interpretability. For example, by combining the frictional resistance contribution values of all sub-regions, the influence of pipeline friction on pollutant attenuation can be reflected separately; combining the local resistance contribution values can reflect the energy loss of components such as elbows and valves. This classification treatment facilitates targeted adjustment in engineering applications (such as optimizing the smoothness of the pipeline in high-resistance sub-regions), and also provides more reliable basic parameters for subsequent pollutant diffusion models, concentration inversion, etc.

[0128] Taking the monitoring point A as the coordinate origin and setting the coordinate of B in the two-dimensional plane as (L, 0), the complex three-dimensional space problem is simplified into a plane model that is easy to calculate. At the same time, through sub-region discretization (such as grid division), the sampling and calculation processes can be efficiently implemented with the help of numerical methods (such as finite element and finite difference). This standardization process reduces the difficulty of engineering implementation, especially suitable for scenarios where the pipeline layout is complex and the monitoring points are irregularly distributed in industrial sites, providing convenience for the rapid deployment and parameter iteration of the real-time monitoring system.

[0129] This step forms a complete correction system with the previous spatial correction coefficient (based on the fluid mechanics model) and correction factor (based on the ratio of measured value to theoretical value): the spatial correction coefficient describes the theoretical influence of the physical environment on the transmission process, and the correction factor reflects the deviation between the actual monitoring data and the theoretical prediction. Through sub-region division and weight assignment, the two are organically combined in the correction parameters (for example, the measured deviation in the high-resistance sub-region will further amplify the correction intensity through the correction factor), which not only retains the mechanistic interpretability of the physical model but also incorporates data-driven adaptive adjustment, ultimately forming a correction scheme closer to the actual working conditions and effectively improving the accuracy of pollutant total calculation, concentration inversion and other links.

[0130] In another preferred embodiment of the present invention, according to the correction parameters, standard data frames, the chemical reaction conversion rate of monitoring point A, and the removal rate of the incinerator purification device at monitoring point B, the pollutants at monitoring point A and the pollutants at monitoring point B are calculated to generate a pollutant correction concentration matrix, including: Extract dimensionless correction parameters from the previous calculation results (such as a comprehensive coefficient reflecting the influence of resistance attenuation and height difference, for example, 0.85). Analyze the standard data frame: separate the mass flow data of monitoring points A and B: the first m columns correspond to point A (such as MA = [MA1, MA2,..., MA m]), and the last k columns correspond to point B (such as MB = [MB1, MB2,..., MB k]).

[0131] Read process parameters: The chemical reaction conversion rate of monitoring point A (for example, the catalytic reaction conversion rate of VOCs in the pretreatment area is 30%, denoted as RA); the removal rate of the incinerator purification device at monitoring point B (such as the removal rate of NO x is 80%, denoted as RB).

[0132] The specific process of correcting the mass flow at monitoring point A is as follows: Multiply the original mass flow MA at point A by the correction parameter (such as 0.85) to compensate for the attenuation or gain during the pipeline transmission process (for example, if the correction parameter is 0.85, it means that theoretically 85% of the mass flow at point A remains after pipeline transmission to reach point B).

[0133] Chemical reaction correction, considering chemical reactions in the pretreatment area (e.g., some pollutants react and are converted into other substances in the treatment equipment before point A): If the conversion rate of a certain pollutant at point A is RA (e.g., 30% is converted into harmless substances), then the corrected mass flow rate = the spatially corrected mass flow rate × (1 - RA).

[0134] For example: The original mass flow rate of a certain pollutant at point A is 100 mg / h, the correction parameter is 0.85, and the conversion rate is 30%. Then the correction result = 100 × 0.85 × (1 - 0.3) = 59.5 mg / h.

[0135] The process of correcting the mass flow rate at monitoring point B is as follows: Purification device correction: Regarding the pollutant removal effect of the incinerator purification device (e.g., bag filter removes particulate matter, activated carbon adsorbs VOCs): If the removal rate of a certain pollutant at point B is RB (e.g., 80%), then the corrected mass flow rate = the original mass flow rate × (1 - RB).

[0136] For example: The original mass flow rate of a certain pollutant at point B is 50 mg / h, the removal rate is 80%, and the correction result = 50 × (1 - 0.8) = 10 mg / h.

[0137] Reverse verification correction, combined with the correction parameter, verify the consistency between the corrected mass flow rate at point B and the theoretical transfer amount at point A (e.g., after the corrected mass flow rate at point A is transferred through the pipeline, it should be close to the theoretical value of the mass flow rate before correction at point B).

[0138] The process of generating the corrected pollutant concentration matrix is as follows: Obtain flow data, and extract the flow sequences aligned with the time stamps of the standard data frames from the induced draft fan flow rate (at point A) and the incinerator flue gas flow rate (at point B) collected during the calculation of the mass matrix (e.g., the flow rate at point A QA = [QA1, QA2,..., QAn], and the flow rate at point B QB = [QB1, QB2,..., QBn]).

[0139] Convert mass flow rate to concentration: For each time point i at point A, the corrected concentration = the corrected mass flow rate MAi' ÷ the corresponding flow rate QA i (the unit is converted from mg / h to mg / m 3 )

[0140] For each time point i at point B, the corrected concentration = the corrected mass flow rate MBi' ÷ the corresponding flow rate QBi.

[0141] Matrix splicing, splice the corrected concentrations at point A and point B in the original column order (point A column first, point B column second) to form a two-dimensional corrected pollutant concentration matrix, where each row corresponds to a time point and each column corresponds to the corrected concentration of a certain pollutant.

[0142] In another preferred embodiment of the present invention, early warning is performed according to the corrected pollutant concentration matrix, the flow rate of the induced draft fan, the flue gas flow rate, and a preset emission threshold, including: Separate the corrected concentrations at monitoring points A / B: The first m columns of the pollutant corrected concentration matrix correspond to the corrected concentrations at monitoring point A (such as VOCs, ammonia, etc., unit mg / m 3 ), and the last k columns correspond to the corrected concentrations at monitoring point B (such as NO x , particulate matter, etc.).

[0143] Obtain real-time flow data: The flow rate of the induced draft fan (monitoring point A, unit m 3 / h) and the flue gas flow rate of the incinerator (monitoring point B, unit m³ / h) are strictly aligned with the rows of the corrected concentration matrix according to the time stamp (for example, the i-th row corresponds to the flow rates QAi and QBi at the i-th time point).

[0144] Load the preset emission threshold: For each pollutant (such as the VOCs threshold of 50 mg / m 3 at point A and the NOx threshold of 20 mg / m 3 at point B), a threshold list is formed (such as threshold A = [TA1, TA2,..., TAm], threshold B = [TB1, TB2,..., TBk]).

[0145] The processing procedure for calculating the real-time emission index of each pollutant is as follows: Concentration early warning index: Directly use the values in the corrected concentration matrix (mg / m 3 ) to reflect the real-time concentration of pollutants in the current gas (such as the VOCs concentration at the i-th time point at point A is 45 mg / m 3 , and the NO x concentration at point B is 25 mg / m 3 ).

[0146] Mass flow early warning index (if the threshold is mass flow): For monitoring point A, multiply the corrected concentration (the first m columns of the i-th row) by the corresponding induced draft fan flow rate QAi to obtain the mass flow of each pollutant (mg / h, such as the VOCs mass flow at point A = 45 mg / m 3 × 2000 m 3 / h = 90000 mg / h).

[0147] For monitoring point B, multiply the corrected concentration (the last k columns of the i-th row) by the corresponding flue gas flow rate QBi to obtain the mass flow at point B (such as the NO x mass flow at point B = 25 mg / m 3 × 15,000 m3 / h = 375000 mg / h).

[0148] The process of comparing the threshold for each pollutant at each time point is as follows: Concentration comparison (taking the concentration threshold as an example): For the j-th pollutant at monitoring point A, the corrected concentration CAij' at the i-th time point is compared with the corresponding threshold TAj: If CAij' ≥ TAj, it is marked as "warning" (e.g., the VOCs threshold at point A is 50 mg / m 3 , and the current is 45 mg / m 3 then there is no warning. If it is 55 mg / m 3 then a warning is triggered).

[0149] For the l-th pollutant at monitoring point B, the corrected concentration CBil' at the i-th time point is compared with the threshold TBl, and the logic is the same.

[0150] Mass flow comparison (if the threshold is the mass flow): The calculated mass flow of point A / B is compared with the corresponding mass flow threshold (e.g., the VOCs mass flow threshold at point A is 100000 mg / h, and there is no warning for the current 90000 mg / h. If it is 110000 mg / h, then a warning is triggered).

[0151] Record the warning status: Generate a status label ("normal" or "warning") for each time point and each pollutant, and mark the excess multiple (e.g., current concentration / threshold - 1, for hierarchical warning).

[0152] The process of generating and outputting the warning information is as follows: Summarize all the pollutants exceeding the standard at each time point (e.g., at the i-th time point: VOCs warning at point A, NO at point B x normal), to form a warning list (including timestamp, monitoring point, pollutant name, measured value, threshold, excess multiple).

[0153] If there is a "warning" status, send notifications through system interface pop-up windows, text messages, emails, etc. (e.g., "At 14:30, the VOCs concentration at monitoring point A is 55 mg / m 3 (threshold 50 mg / m 3 ), exceeding the standard by 10%").

[0154] Store the warning log: Store the warning information in the database for subsequent traceability and statistics (e.g., analyzing the monthly warning times of a certain pollutant and the distribution of exceeding-standard time periods).

[0155] As Figure 2 shown, a remote detection method for waste gas emissions during the treatment process of kitchen waste includes: Step 1: Collect the spectral signals of the waste gas at monitoring point A and monitoring point B in real time, analyze the spectral absorption signals of the characteristic pollutants, and generate the first concentration matrix and the second concentration matrix corresponding to monitoring point A and monitoring point B respectively; Step 2: Multiply the first concentration matrix by the flow rate of the induced draft fan at monitoring point A at each moment to obtain the first mass matrix; multiply the second concentration matrix by the flow rate of the incinerator flue at monitoring point B at each moment to obtain the second mass matrix; splice the first mass matrix and the second mass matrix to form a combined matrix; preprocess the combined matrix to obtain a standard data frame; Step 3: Analyze the physical environment between monitoring point A and monitoring point B, generate a spatial correction coefficient based on the principle of fluid mechanics according to the physical environment; estimate the theoretical total amount of pollutants between monitoring point A and monitoring point B according to the design parameters and operating principle of the incinerator flue; calculate the correction factor according to the measured total amount and the theoretical total amount of pollutants at monitoring point A and monitoring point B in the standard data frame; obtain the correction parameter according to the spatial correction coefficient and the correction factor; Step 4: Calculate the pollutants at monitoring point A and the pollutants at monitoring point B according to the correction parameter, the standard data frame, the chemical reaction conversion rate at monitoring point A and the removal rate of the purification device of the incinerator at monitoring point B to generate a pollutant correction concentration matrix; Step 5: Issue a warning according to the pollutant correction concentration matrix, the induced draft fan flow rate, the flue flow rate and a preset emission threshold.

[0156] An embodiment of the present invention further provides a computer-readable storage medium storing instructions, which when run on a computer cause the computer to execute the method described above. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

Claims

1. A remote detection system for waste gas emissions during kitchen waste treatment, characterized in that: include: The detection module is used to collect the spectral signals of the exhaust gas at the monitoring point A and the monitoring point B in real time, analyze the spectral absorption signals of the characteristic pollutants, and generate the first concentration matrix and the second concentration matrix corresponding to the monitoring point A and the monitoring point B respectively; The processing module is used to multiply the first concentration matrix with the induced draft fan flow at the collection monitoring point A moment by moment to obtain a first mass matrix; and multiply the second concentration matrix with the incinerator flue flow at the monitoring point B moment by moment to obtain a second mass matrix; concatenate the first mass matrix and the second mass matrix to form a joint matrix; Preprocess the joint matrix to obtain a standard data frame; The correction module is used to analyze the physical environment between monitoring point A and monitoring point B, generate a spatial correction coefficient based on the physical environment based on the principle of fluid mechanics, and estimate the theoretical total amount of pollutants between monitoring point A and monitoring point B based on the design parameters and operating principles of the incinerator flue. Calculate the correction factor based on the measured total amount and theoretical total amount of pollutants at monitoring point A and monitoring point B in the standard data frame; According to the space correction coefficient and correction factor, the correction parameter is obtained; A calculation module is used to calculate the pollutants at monitoring point A and the pollutants at monitoring point B according to the correction parameters, the standard data frame, the chemical reaction conversion rate at monitoring point A and the removal rate of the incinerator purification device at monitoring point B, so as to generate a pollutant correction concentration matrix; The early warning module is used to issue early warnings based on the pollutant correction concentration matrix, induced draft fan flow, flue flow and preset emission thresholds.

2. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 1, characterized in that: Monitoring point A is the exhaust gas emission outlet of the pretreatment area of ​​the food waste treatment workshop; monitoring point B is the flue outlet of the incinerator.

3. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 2, characterized in that: The spectral signals of the exhaust gas at monitoring point A and monitoring point B are collected in real time, the spectral absorption signals of the characteristic pollutants are analyzed, and the first concentration matrix and the second concentration matrix corresponding to monitoring point A and monitoring point B are generated respectively, including: The original spectrum sequences of monitoring point A and monitoring point B are obtained respectively, which are recorded as the first spectrum data stream and the second spectrum data stream respectively, and each data stream consists of spectrum curves with continuous time stamps; Respectively performing smoothing, baseline correction and wavelength calibration on the first spectral data stream and the second spectral data stream to obtain a first spectral curve and a second spectral curve; Pre-establish a library of characteristic absorption wavelengths of target pollutants; For the first spectral curve and the second spectral curve, within the wavelength range corresponding to the characteristic wavelength library, the ratio of the actual light intensity to the pure air light intensity is calculated to obtain the absorbance value at each characteristic wavelength; According to the absorbance value at each characteristic wavelength and the pre-established pollutant concentration-absorbance calibration model, the pollutant concentration vector at monitoring point A and monitoring point B at each moment is calculated; The concentration vectors at each moment are arranged vertically in chronological order to form a two-dimensional concentration matrix. Each row of the two-dimensional concentration matrix corresponds to a time point, and each column corresponds to a pollutant.

4. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 3, characterized in that: The first concentration matrix is ​​multiplied by the induced draft fan flow rate at the acquisition monitoring point A moment by moment to obtain the first mass matrix, including: Collect the flow data of the induced draft fan at monitoring point A. The flow data is a sequence that changes with time, and each time point has a corresponding induced draft fan flow value; Starting from the first row of the first concentration matrix, the first row represents the concentration of various pollutants at the first time point; multiply each element in the first row, that is, the concentration of each pollutant, by the induced draft fan flow value corresponding to the time point to obtain the mass flow value of various pollutants at the first time point, and form a new row vector; Similarly, each row of the first concentration matrix is ​​multiplied until all rows are processed, and all the obtained row vectors are arranged in sequence to form a first mass matrix. The first mass matrix is ​​two-dimensional, each row corresponds to a time point, each column corresponds to a pollutant, and the elements in the matrix represent the mass flow rate of the pollutant at the corresponding time point.

5. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 4, characterized in that: The second concentration matrix is ​​multiplied by the flue flow rate of the incinerator at the monitoring point B at each moment to obtain the second mass matrix, including: Obtain the flow data of the incinerator flue at monitoring point B. The flow data of the incinerator flue is a sequence that changes with time, and each time point has a corresponding flue flow value; Starting from the first row of the second concentration matrix, each element of the first row, that is, the concentration of each pollutant, is multiplied by the flue flow value of the incinerator corresponding to the first time point to obtain the mass flow value of each pollutant at the first time point to form a row vector; The same operation is performed on the second row of the second concentration matrix, and its elements are multiplied by the flue flow value at the second time point to obtain the mass flow row vector at the second time point; Similarly, after processing all rows of the second concentration matrix, the obtained row vectors are arranged in order to obtain the second mass matrix. The second mass matrix is ​​two-dimensional, each row represents a time point, each column represents a pollutant, and the element is the mass flow rate of the pollutant at the time point.

6. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 5, characterized in that: concatenate the first mass matrix and the second mass matrix to form a joint matrix; Preprocess the joint matrix to obtain a standard data frame, including: The first mass matrix and the second mass matrix are concatenated by columns, that is, all the columns of the first mass matrix are placed in front, and all the columns of the second mass matrix are placed in the back, so as to form a new two-dimensional matrix, that is, a joint matrix, each row of the joint matrix corresponds to a time point, the front column corresponds to the pollutant mass flow rate of the monitoring point A, and the back column corresponds to the pollutant mass flow rate of the monitoring point B; Each element in the joint matrix is ​​filtered by moving average filtering, and the filtered joint matrix is ​​normalized so that the value of each column of data is mapped to the interval [0, 1] to obtain a standard data frame.

7. The remote detection system for waste gas emission during the kitchen waste treatment process according to claim 1 is characterized in that: Analyze the physical environment between monitoring point A and monitoring point B, and generate a spatial correction coefficient based on the principles of fluid mechanics and the physical environment, including: Obtaining physical parameters of monitoring point A and monitoring point B, including the straight-line distance between monitoring point A and monitoring point B in three-dimensional space, vertical height difference, total length of pipeline, inner diameter of pipeline, roughness of inner wall of pipeline and bending characteristics of pipeline; Based on physical parameters, calculate the resistance along the exhaust gas when it flows in a straight pipe, and calculate the local resistance of the exhaust gas in non-straight pipe sections; Generate a spatial correction factor based on along-the-way resistance, local resistance and height difference.

8. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 7, characterized in that: According to the measured total amount and theoretical total amount of pollutants at monitoring point A and monitoring point B in the standard data frame, the correction factor is calculated, including: Extract the measured total amount of pollutants at monitoring point A and monitoring point B from the standard data frame; Calculate the ratio of the measured total amount to the theoretical total amount to obtain the correction factor.

9. A remote detection system for waste gas emissions during kitchen waste treatment according to claim 8, characterized in that: According to the space correction coefficient and correction factor, the correction parameters are obtained, including: Taking monitoring point A as the coordinate origin (0, 0), the coordinate of monitoring point B in the two-dimensional plane is (L, 0), where L is the straight-line distance between the two points; according to the actual path of the exhaust gas transmission pipeline, the transmission area between monitoring points A and B is divided into several sub-areas, and each sub-area corresponds to a different component in the spatial correction coefficient; Sample the spatial points in each sub-area and calculate their distances to monitoring points A and B; Multiply the spatial correction coefficient component of each sub-region by the corresponding weight to obtain the contribution value of the component to the correction parameter; The contribution values ​​of all components are accumulated according to the physical characteristics and finally form the correction parameters.

Images