Particle size channel correction method based on trapping efficiency evaluation system

Abstract

The invention discloses a particle size channel correction method based on a capture efficiency evaluation system, and relates to the field of aerosol measurement, and the method comprises the steps: constructing a static box and a capture efficiency evaluation unit of an aerodynamic particle size spectrometer, and in a calibration state of the capture efficiency evaluation unit, carrying out the calibration of the capture efficiency evaluation unit; establishing a particle size channel reference database by measuring the monodisperse standard substance; the method comprises the following steps: preparing mixed aerosol of polydisperse test dust and monodisperse standard microspheres with known particle sizes, and respectively measuring particle size distribution original data of the mixed aerosol passing through the upstream and downstream of a cutter to be evaluated by utilizing a trapping efficiency evaluation unit; and inputting standard microsphere channel information in the particle size distribution original data of the particulate matters into a transfer learning model pre-trained based on a historical data set in the particle size channel reference database. According to the method, by analyzing the comprehensive correction quality factor calculated in the transfer learning model, quantitative evaluation of the reliability of the correction process is realized, and black box operation is converted into transparent and measurable quality control.

Description

Technical Field

[0001] This invention relates to the field of aerosol measurement, and in particular to a particle size channel correction method based on a trapping efficiency evaluation system. Background Technology

[0002] In the field of aerosol measurement, the evaluation system based on the static chamber-aerodynamic particle size analyzer is the mainstream technology for assessing the capture efficiency of PM2.5 and other particles. To address the channel drift problem that may occur in particle size analyzers due to long-term use and to improve measurement accuracy, the internal standard method is widely used. This method involves mixing monodisperse standard microspheres of known particle size with polydisperse test dust, observing the positional shift of the microspheres in the particle size channel to determine the degree of drift, and correcting the measurement data accordingly to obtain a more accurate capture efficiency curve.

[0003] While the internal standard method is effective, its correction strategy has limitations. Existing methods typically employ simple linear translation or fixed models for correction, making it difficult to accurately adapt to complex, nonlinear channel drift patterns. More importantly, this method lacks a quantitative evaluation mechanism for the reliability of the correction result; the correction process is like a black box, leaving users unable to ascertain the reliability of the correction. This introduces uncertainty and risk into the accurate determination of cutter performance and limits the application of this technology in metrology scenarios requiring high reliability. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a particle size channel correction method based on a trapping efficiency evaluation system to solve the problem of uncertainty risk in cutter performance evaluation caused by the simple correction model and inability to quantify the reliability of the correction results in the existing internal standard method.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a particle size channel correction method based on a trapping efficiency evaluation system, which includes constructing a trapping efficiency evaluation unit consisting of a static chamber and an aerodynamic particle size spectrometer, and establishing a particle size channel reference database by measuring monodisperse standard substances under the calibrated state of the trapping efficiency evaluation unit.

[0008] A mixed aerosol of polydisperse test dust and monodisperse standard microspheres with known particle size was prepared. Using a collection efficiency evaluation unit, the raw particle size distribution data of the mixed aerosol was measured at the upstream and downstream of the cutter to be evaluated.

[0009] The standard microsphere channel information in the particle size distribution raw data is input into a transfer learning model pre-trained based on a historical data set of the particle size channel reference database, non-linear channel drift characteristics are identified through small sample adaptive analysis, and a dynamic non-linear correction function is generated;

[0010] The particle size distribution raw data is corrected by using the dynamic non-linear correction function, and corrected particle size distribution data is obtained.

[0011] The correction quality factor is calculated based on the internal analysis data of the transfer learning model in the drift characteristic identification process.

[0012] The measurement confidence interval of the capture efficiency evaluation result is quantified based on the correction quality factor.

[0013] Based on the measurement confidence interval of the quantified capture efficiency evaluation result, a comprehensive evaluation report is output.

[0014] As a preferred scheme of the particle size channel correction method of the capture efficiency evaluation system based on the capture efficiency of the application, the capture efficiency evaluation unit of the static box and the aerodynamic particle sizer is constructed, and the particle size channel reference database is established by measuring monodisperse standard substances in the calibration state of the capture efficiency evaluation unit, including the following steps:

[0015] The hardware integration of the capture efficiency evaluation unit is obtained by connecting the static box, the aerodynamic particle sizer, the atomization dusting part, the descending mixing box and the industrial computer, and the capture efficiency evaluation unit is subjected to air tightness check and flow calibration;

[0016] The atomization dusting part of the capture efficiency evaluation unit is started, the monodisperse standard substance solution is atomized and then introduced into the measurement box of the capture efficiency evaluation unit, and the stable particle size channel information of the monodisperse standard substance aerosol in the measurement box of the capture efficiency evaluation unit is measured by using the aerodynamic particle sizer;

[0017] The known particle size of the monodisperse standard substance and the stable particle size channel information measured by the aerodynamic particle sizer are stored in the industrial computer as a corresponding relationship;

[0018] The new monodisperse standard substance with known aerodynamic diameter is selected, and the atomization, measurement and data storage processes are performed in the capture efficiency evaluation unit to obtain a new particle size-channel corresponding relationship;

[0019] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the present application, wherein: a mixed aerosol of polydisperse test dust and monodisperse standard microspheres with known particle size is prepared, and the particle size distribution raw data of the mixed aerosol passing upstream and downstream of the cutter to be evaluated are measured respectively by using the trapping efficiency evaluation unit, including the following steps:

[0020] Selecting polydisperse test dust and monodisperse standard microspheres with known aerodynamic diameter, placing them in a dispersion medium, and performing ultrasonic treatment to form a uniform mixed solution;

[0021] Injecting the prepared mixed solution into the atomization dust generation part of the trapping efficiency evaluation unit to generate a mixed aerosol of polydisperse test dust and monodisperse standard microspheres;

[0022] The mixed aerosol flows into the descending mixing box of the trapping efficiency evaluation unit for sufficient mixing and homogenization, and is introduced into the static box of the trapping efficiency evaluation unit;

[0023] Controlling the three-way electromagnetic valve of the trapping efficiency evaluation unit to switch to the path of extracting the mixed aerosol without passing through the cutter, and entering the aerodynamic particle sizer to measure and obtain the particle size distribution data of the mixed aerosol passing upstream of the cutter to be evaluated;

[0024] Controlling the three-way electromagnetic valve of the trapping efficiency evaluation unit to switch to the path of extracting the mixed aerosol passing through the cutter to be evaluated, and entering the aerodynamic particle sizer to measure and obtain the particle size distribution data of the mixed aerosol passing downstream of the cutter to be evaluated.

[0025] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the present application, wherein: the standard microsphere channel information in the particle size distribution raw data is input into the transfer learning model pre-trained based on the historical data set of the particle size channel reference database, and through small sample adaptive analysis, the nonlinear channel drift characteristics are identified to generate a dynamic nonlinear correction function, including the following steps:

[0026] From the particle size distribution raw data, the channel number corresponding to the characteristic peak signal generated by the monodisperse standard microspheres with known particle size is identified and extracted to form a set of standard microsphere channel information of the current measurement;

[0027] Using the particle size channel reference database to construct a historical data set of the corresponding relationship between normal and drift states, pre-training the transfer learning model, and obtaining the trained transfer learning model;

[0028] The standard microsphere channel information set is input into the trained transfer learning model, and the small sample data is subjected to field self-adaptive analysis based on the pre-training knowledge to identify the nonlinear channel drift characteristics deviating from the benchmark state in the current particulate matter particle size distribution original data;

[0029] Based on the identified nonlinear channel drift characteristics, a dynamic nonlinear correction function is output.

[0030] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the trapping efficiency, wherein: the dynamic nonlinear correction function is used to correct the particulate matter particle size distribution original data in batches to obtain corrected particle size distribution data, including the following steps:

[0031] The generated dynamic nonlinear correction function is read from the result storage location output by the transfer learning model after identifying and analyzing the drift characteristics of the current data, and the particulate matter particle size distribution original data to be corrected is simultaneously loaded;

[0032] The number of each channel in the particulate matter particle size distribution original data is input into the dynamic nonlinear correction function, and each channel value is operated to output the corrected particle size value;

[0033] The corrected particle size value is used to replace the corresponding original channel value in the particulate matter particle size distribution original data to form a new data pair set;

[0034] The new data pair set is integrated and arranged according to the particle size order to obtain the corrected particle size distribution data.

[0035] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the trapping efficiency, wherein: based on the internal analysis data of the transfer learning model in the process of identifying the drift characteristics, a correction quality factor is calculated, including the following steps:

[0036] From the intermediate calculation results and state variables generated by the transfer learning model in the process of identifying the drift characteristics, the internal analysis data set in the analysis process is extracted;

[0037] Based on the internal analysis data set, the transfer learning model fitting residual, classification confidence, nonlinear channel drift characteristic and known drift pattern characteristic matching degree, and different monodisperse standard microsphere independent correction parameter consistency indexes of the transfer learning model in the process of identifying the drift characteristics of the current input standard microsphere channel information are calculated to constitute an analysis confidence index;

[0038] The relative deviation between the independent correction parameters corresponding to different monodisperse standard microspheres in the internal analysis data set is calculated;

[0039] The analysis confidence index is weighted and fused with the consistency evaluation index of the multi-internal standard correction result to calculate a correction quality factor value.

[0040] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the present application, the trapping efficiency is calculated according to the corrected particle size distribution data, the trapping efficiency curve is fitted, and the measurement confidence interval of the trapping efficiency evaluation result is quantified based on the correction quality factor, including the following steps:

[0041] The particle number concentration of each particle size point upstream and downstream of the cutter in the corrected particle size distribution data is used to calculate the trapping efficiency value of each particle size point, and the data points are fitted using a reverse asymmetric S-shaped equation with particle size as the abscissa and trapping efficiency value as the ordinate to generate a trapping efficiency curve.

[0042] The aerodynamic diameter evaluation parameter is read from the trapping efficiency curve, and the measurement confidence interval is assigned to the aerodynamic diameter evaluation parameter based on the mapping relationship between the correction quality factor and the historical performance calibration.

[0043] As a preferred scheme of the particle size channel correction method based on the trapping efficiency evaluation system of the present application, the measurement confidence interval of the quantified trapping efficiency evaluation result is used to output a comprehensive evaluation report, including the following steps:

[0044] The fitted trapping efficiency curve and the measurement confidence interval data are associated and integrated to form a complete data set.

[0045] By comparing and analyzing the measurement confidence interval of the trapping efficiency evaluation result with the standard qualified range, a risk avoidance strategy is obtained, the actual value of the evaluation parameter in the integrated data set is compared and integrated, the performance qualification of the cutter is determined, and a visual chart is obtained.

[0046] Based on the corrected trapping efficiency curve and the actual value and measurement confidence interval of the evaluation parameter, a comprehensive evaluation report is obtained

[0047] The visual chart and the comprehensive evaluation report are integrated to obtain a cutter performance conclusion.

[0048] In a second aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the computer program is executed by the processor to implement any step of the particle size channel correction method based on the trapping efficiency evaluation system of the present application according to the first aspect of the present application.

[0049] In a third aspect, the present application provides a computer readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements any step of the particle size channel correction method based on the trapping efficiency evaluation system of trapping efficiency according to the first aspect of the present application.

[0050] The present application has the beneficial effects that: through small sample self-adaptive analysis of the internal standard data by using the transfer learning model pre-trained based on the historical data set, accurate identification and correction of complex nonlinear channel drift characteristics are realized, a high-precision dynamic nonlinear correction function is generated, the problem of insufficient precision of the traditional linear correction model is fundamentally solved, the comprehensive correction quality factor is calculated in the transfer learning model, quantitative evaluation of the reliability of the correction process itself is realized, and the black box operation is changed into transparent and measurable quality control. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0052] Fig. 1 The flowchart of the particle size channel correction method based on the trapping efficiency evaluation system of trapping efficiency.

[0053] Fig. 2 The flowchart of the particle size distribution measurement.

[0054] Fig. 3 The schematic diagram of the comprehensive evaluation report. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0056] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0057] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0058] Reference Figs. 1-3 This is one embodiment of the present invention, which provides a particle size channel correction method based on a trapping efficiency evaluation system, comprising the following steps:

[0059] S1. Construct a collection efficiency evaluation unit for a static chamber and an aerodynamic particle size spectrometer. Under the calibrated state of the collection efficiency evaluation unit, establish a particle size channel reference database by measuring monodisperse standard substances.

[0060] S1.1 By connecting the static box, aerodynamic particle size analyzer, atomizing dust generation section, descending mixing box and industrial control computer, the hardware integration of the collection efficiency evaluation unit is obtained, and the airtightness of the collection efficiency evaluation unit and flow calibration are performed.

[0061] Furthermore, the hardware integration of the capture efficiency evaluation unit is achieved through the physical piping and electrical signal connections between the static chamber, aerodynamic particle size analyzer, atomizing dust generation section, descending mixing chamber, and industrial control computer, as specified in the technical manuals and industry standards provided by the equipment manufacturer. Airtightness testing is a process of verifying the overall sealing performance of the connected capture efficiency evaluation unit using a pressure holding method. Flow calibration is the process of calibrating and adjusting the working flow rate through the key air paths of the aerodynamic particle size analyzer and cutter using a portable standard flow meter with metrological traceability. This controls the stability of the measurement environment and conditions from the source, avoiding systematic errors introduced by hardware leaks or flow deviations.

[0062] S1.2. Start the atomization and dust generation section of the collection efficiency evaluation unit, atomize the monodisperse standard substance solution and pass it into the measurement chamber of the collection efficiency evaluation unit. Measure the stable particle size channel information of the monodisperse standard substance aerosol in the measurement chamber of the collection efficiency evaluation unit using an aerodynamic particle size spectrometer.

[0063] Furthermore, the atomization and dust generation section of the collection efficiency evaluation unit is a crucial operation that converts a specific monodisperse standard substance solution into droplets and generates stable monodisperse aerosols through a drying process. The generated aerosols are then introduced into the measurement chamber of the collection efficiency evaluation unit, where they are uniformly distributed. An aerodynamic particle size analyzer is used to measure the monodisperse standard substance aerosols within the measurement chamber. This measurement utilizes the inherent time-of-flight principle of particles to acquire signals from a large number of particles and statistically analyze the peak particle size distribution. This records the stable particle size channel information corresponding to the monodisperse standard substance aerosol in the collection efficiency evaluation unit under the current calibration state. The known particle size-measured channel correspondence data under the baseline state is used to leverage the uniqueness and determinism of the monodisperse standard substance particle size.

[0064] S1.3, store the known particle size of monodisperse standard material and the stable particle size channel information measured by aerodynamic particle sizer as a corresponding relationship in industrial computer.

[0065] Further, the known aerodynamic diameter of monodisperse standard material and the stable particle size channel information measured by aerodynamic particle sizer are stored as a set of corresponding relationships. The two kinds of information are associated by the data processing software built in the industrial computer and written into the database file. The known aerodynamic diameter is a constant true value reference from the standard material certificate of monodisperse standard material. The stable particle size channel information is the measurement response of the trapping efficiency evaluation unit to the true value under the current ideal state. The purpose of this step is to form a traceable calibration record, and to establish a direct and quantitative relationship between the inherent properties of the material and the measurement response of the instrument, providing the most basic data unit for the entire particle size channel reference database.

[0066] S1.4, select a new monodisperse standard material with known aerodynamic diameter, execute the atomization, measurement and data storage process in the trapping efficiency evaluation unit, and obtain a new particle size-channel corresponding relationship.

[0067] Further, the new monodisperse standard material with known aerodynamic diameter is selected to expand the coverage of the particle size channel reference database. The atomization, measurement and data storage process is repeated in the trapping efficiency evaluation unit, and the monodisperse standard material is replaced with another material with different particle size, to obtain a new known particle size-measured channel corresponding relationship corresponding to different particle size points. By increasing the calibration points, the richness and representativeness of the database are improved. The response characteristics of the instrument at different particle size points may be different, and multi-point calibration can more comprehensively characterize the global performance of the trapping efficiency evaluation unit.

[0068] S1.5, integrate all particle size-channel corresponding relationships stored in the industrial computer to form a particle size channel reference database.

[0069] Further, all independent calibration data records are extracted by the data management software, and are reorganized, arranged and stored according to a certain format. The particle size channel reference database formed is a structured data set, which completely defines the mapping relationship between a series of known particle sizes and the correct channel numbers that should be measured in theory under the calibration state of the trapping efficiency evaluation unit.

[0070] S2, prepare a mixed aerosol of polydisperse test dust and monodisperse standard microspheres with known particle size, and use the trapping efficiency evaluation unit to measure the particle size distribution raw data of the mixed aerosol passing through the upstream and downstream of the cutting device to be evaluated, respectively.

[0071] S2.1, Selecting polydisperse test dust and monodisperse standard microspheres with known aerodynamic diameter, placing them in a dispersion medium, and performing ultrasonic treatment to form a uniform mixed solution.

[0072] Furthermore, the polydisperse test dust is used to simulate the wide particle size range of particulate matter in the real environment, and the monodisperse standard microspheres are used as internal standards to provide a known and accurate particle size reference point. The role of ultrasonic treatment is to break the particle agglomeration using cavitation effect and ensure that the polydisperse test dust and monodisperse standard microspheres are uniformly distributed in the solution, preparing a composite system with stable physical and chemical properties containing the background to be tested and the known reference.

[0073] S2.2, Injecting the prepared mixed solution into the atomization dust generation part of the collection efficiency evaluation unit to generate a mixed aerosol of polydisperse test dust and monodisperse standard microspheres.

[0074] Furthermore, the atomization dust generation part breaks the mixed solution into small droplets through high-speed airflow or ultrasonic energy. The solvent evaporates rapidly as the droplets pass through the drying tube, leaving solid mixed particulate matter to form a mixed aerosol, which physically converts the uniform liquid mixture into a gas-phase suspended particulate matter with the same composition ratio that can be measured.

[0075] S2.3, The mixed aerosol flows into the descending mixing box of the collection efficiency evaluation unit for sufficient mixing and homogenization, and is introduced into the static box of the collection efficiency evaluation unit.

[0076] Furthermore, the descending mixing box utilizes the natural diffusion and turbulent effect of particles during gravitational settling in a larger space to further mix the aerosol from the atomization dust generation part, eliminating the transient concentration fluctuations that may occur during the atomization process, and forming a highly stable mixed aerosol environment in terms of concentration in time and space before the static box inlet.

[0077] S2.4, Control the three-way electromagnetic valve of the collection efficiency evaluation unit to switch to the path of extracting mixed aerosol without the cutter, and enter the aerodynamic particle sizer to measure and obtain the particle size distribution data of the mixed aerosol passing through the particles upstream of the cutter to be evaluated.

[0078] Furthermore, the measured is the particle size distribution of the original mixed aerosol without the cutter, obtaining the intrinsic distribution of the particle population input to the cutter, wherein the characteristic peak signal of the monodisperse standard microspheres should be near its theoretical channel position at this time.

[0079] S2.5, Control the three-way electromagnetic valve of the collection efficiency evaluation unit to switch to the path of extracting mixed aerosol through the cutter to be evaluated, and enter the aerodynamic particle sizer to measure and obtain the particle size distribution data of the mixed aerosol passing through the particles downstream of the cutter to be evaluated.

[0080] Further, the particle size distribution of the mixed aerosol after being screened by the cutter is measured. Compared with the upstream data, the concentration of particles of different sizes in the downstream data will change characteristically. The distribution of the polydisperse test dust will reflect the trapping characteristics of the cutter, and the peak position of the monodisperse standard microspheres may reflect the channel drift of the particle size spectrometer itself. By comparing the differences between the upstream and downstream data, while decoupling the performance information of the cutter and the state information of the instrument, the two goals of simultaneously completing the performance evaluation of the cutter and the state diagnosis of the instrument in a single measurement process are achieved.

[0081] S3, input the standard microsphere channel information in the particle size distribution raw data into the transfer learning model pre-trained based on the historical data set of the particle size channel reference database, identify the nonlinear channel drift characteristics through small sample adaptive analysis, and generate a dynamic nonlinear correction function.

[0082] S3.1, from the particle size distribution raw data, identify and extract the channel number corresponding to the characteristic peak signal generated by the monodisperse standard microspheres with known particle sizes, and form a standard microsphere channel information set of the current measurement.

[0083] Further, based on the known aerodynamic diameter of the monodisperse standard microspheres, the range of channels in which the standard microspheres should theoretically appear in the spectrum is located, and the local maximum point of the particle number concentration in the range is found through a peak finding algorithm. The channel number corresponding to the peak point is recorded. Each monodisperse standard microsphere with a known particle size corresponds to an extracted channel number. The channel number constitutes a standard microsphere channel information set. The key information reflecting the current state of the instrument is accurately captured from the complex polydisperse background using the distinctness and knownness of the internal standard signal. The potential drift of the instrument is converted into a quantifiable channel number offset.

[0084] S3.2, use the particle size channel reference database to construct a historical data set corresponding to the normal and drift states, pre-train the transfer learning model, and obtain the trained transfer learning model.

[0085] Further, the particle size channel reference database stores a large amount of known particle size-standard channel correspondence data measured under the calibration state of the instrument, and may also include known particle size-drifted channel data measured after the instrument is artificially introduced to drift under certain conditions. A historical data set containing multiple state labels is built. The historical data set is used to pre-train the transfer learning model, so that the transfer learning model learns the complex and nonlinear mapping relationship between the channel observation value, the instrument state, and the true particle size.

[0086] S3.3, input the standard microsphere channel information set into the trained transfer learning model, perform domain adaptation analysis on the small sample data based on the pre-training knowledge, and identify the nonlinear channel drift characteristics deviating from the benchmark state existing in the current particulate matter particle size distribution raw data.

[0087] Further, based on the pre-training knowledge, the domain adaptation analysis is performed on the small sample data, and the nonlinear channel drift characteristics deviating from the benchmark state existing in the current particulate matter particle size distribution raw data are identified. After receiving the current standard microsphere channel information set, the trained transfer learning model starts the internal mechanism to perform domain adaptation analysis. The transfer learning model compares and fine-tunes the general drift pattern knowledge learned from the historical data set in the pre-training stage with the current limited small sample data, so as to accurately identify the specific nonlinear channel drift characteristics of the current instrument. By using the powerful transfer ability of the pre-training transfer learning model, the limitation that the small sample data is difficult to support the training of a complex model is overcome, and high-precision state diagnosis is realized.

[0088] S3.4, based on the identified nonlinear channel drift characteristics, output a dynamic nonlinear correction function.

[0089] Further, the transfer learning model converts the identified nonlinear channel drift characteristics into a specific, calculable mathematical function expression, i.e. a dynamic nonlinear correction function, which describes the correction mapping relationship between the current instrument channel number and the real aerodynamic particle size, and can effectively compensate for the identified specific nonlinear drift. The abstract characteristics are converted into concrete functions, and the diagnosis conclusion is converted into an executable correction tool.

[0090] S4, use the dynamic nonlinear correction function to correct the particulate matter particle size distribution raw data in batches, and obtain the corrected particle size distribution data.

[0091] S4.1, read the generated dynamic nonlinear correction function from the storage location of the result output by the transfer learning model after analyzing the drift characteristics of the current data, and simultaneously load the particulate matter particle size distribution raw data to be corrected.

[0092] Further, after the transfer learning model completes the analysis, the specific mathematical expression and parameters of the generated dynamic nonlinear correction function are written to a designated temporary storage location, and the particulate matter particle size distribution raw data obtained by measurement are loaded from the data storage area at a specific address in the memory of the industrial computer.

[0093] S4.2, input the number of each channel in the particulate matter particle size distribution raw data into the dynamic nonlinear correction function as a parameter, and perform operation on each channel value to output the corrected particle size value.

[0094] Further, the paradigm shift from empirical manual calibration to automated intelligent correction is achieved by inputting each original channel number as a parameter into the parameterized nonlinear function dynamically generated by the transfer learning model for the current instrument state, performing batch operations, and directly outputting the corrected physical particle size value. The dynamic nonlinear correction function is not a fixed formula, and its parameters (a, b, c, d, f) are dynamically determined by the transfer learning model according to the real-time diagnosed drift characteristics, so that the dynamic nonlinear correction function can not only perform simple linear translation, but also accurately compensate for complex and nonlinear channel drift, and its correction accuracy is much higher than that of the previous linear or polynomial fitting method.

[0095] The corrected particle size value expression is:

[0096] ;

[0097] wherein, is the corrected aerodynamic particle size value, is the dynamic nonlinear correction function, is the asymptote of the correction function curve, is the span between the upper and lower asymptotes of the correction function curve, is the number of each channel in the original particle size distribution data of particulate matter, is the slope of the curve, is the central position parameter, is the factor controlling the asymmetry of the curve.

[0098] S4.3, using the corrected particle size value, replacing the corresponding original channel value in the original particle size distribution data of particulate matter to form a new data pair set.

[0099] Further, the original data pair set of the original form of the particle size distribution data of particulate matter is replaced by the corresponding corrected particle size value with the new and accurate particle size value, thereby forming a new data pair set, updating the horizontal coordinate of the data, and keeping the vertical coordinate unchanged. Because instrument drift mainly affects particle size identification rather than concentration measurement, the original measurement statistical information is preserved, and the systematic error of particle size identification is fundamentally corrected.

[0100] S4.4, integrating and arranging the new data pair set according to the particle size size order to obtain the corrected particle size distribution data.

[0101] Further, the new data pair set formed after replacement is integrated and arranged according to the particle size size order to obtain the corrected particle size distribution data. Due to the nonlinear characteristics of the dynamic nonlinear correction function, the sequence of the corrected particle size values may not be completely consistent with the sequence of the original channel numbers in size order.

[0102] It should be noted that the new data pair set is reordered and integrated according to the modified particle size value from small to large, to generate a standard particle number concentration distribution spectrum with the real aerodynamic particle size as the increasing sequence, that is, the modified particle size distribution data, and the modified data is standardized.

[0103] S5, based on the internal analysis data of the migration learning model in the process of identifying drift characteristics, the correction quality factor is calculated.

[0104] S5.1, from the intermediate calculation results and state variables generated by the nonlinear channel drift feature identification process completed by the migration learning model, the internal analysis data set in the current analysis process is extracted.

[0105] Further, the migration learning model generates a large amount of intermediate data in the running process, such as the activation value, gradient information, reconstruction error of input data, and probability distribution on different predefined drift modes of each layer in the model, and the intermediate calculation results and state variables jointly constitute the internal analysis data set. They objectively record the complete thinking process and internal state of the migration learning model in the process of feature recognition and decision making. Deeply mining and utilizing the deep process information generated in the running process of the intelligent model far beyond the final output result provides rich, multi-dimensional original basis for the credibility of the current analysis of the subsequent migration learning model, so that the evaluation is no longer dependent on a single final output, but is based on the internal running mechanism of the migration learning model, greatly improving the depth and reliability of the evaluation.

[0106] S5.2, based on the internal analysis data set, the migration learning model fitting residual, classification confidence, nonlinear channel drift feature and known drift mode feature matching degree, and the consistency index of independent correction parameters of different monodisperse standard microspheres when the migration learning model identifies the drift characteristics of the current input standard microsphere channel information are calculated, to constitute the analysis confidence index.

[0107] It should be noted that the internal analysis data set is deeply processed and information is extracted. The migration learning model fitting residual measures the accuracy of the model reconstruction or prediction; the classification confidence reflects the degree of grasp of the final judgment of the migration learning model; the feature matching degree quantifies the similarity between the current problem and historical experience; and the independent correction parameter consistency index evaluates the robustness of the conclusions based on different internal standards. These four indicators reflecting the analysis quality from different aspects are fused by weighting, and the analysis confidence index expression is The analysis confidence index is integrated into a comprehensive analysis. A multi-indicator fusion strategy is adopted to conduct a comprehensive and quantitative evaluation of the overall quality of this intelligent analysis from four orthogonal dimensions: accuracy, certainty, empirical conformity, and result robustness. This avoids the one-sidedness of a single indicator and enables the final analysis confidence index to more comprehensively and reliably represent the overall credibility level of this corrective analysis process.

[0108] The expression for the fitted residual is:

[0109] ;

[0110] in For mean absolute residual, Given the number of internal standard microspheres, For the first The theoretical channel values ​​corresponding to the internal standard microspheres with known particle sizes in the particle size channel reference database. For the first The known particle size of a microsphere with a known internal standard particle size point is input as the inverse function of the current dynamic nonlinear correction function. Points with known particle size.

[0111] Mean absolute residuals measure the average deviation between the predicted and theoretical values ​​of a transfer learning model. The smaller the residuals, the higher the model fit. and Inversely proportional, ;

[0112] The expression for classification confidence is:

[0113] ;

[0114] in, The confidence level for determining the drift mode. The probability of the first predefined drift pattern. The probability of the second predefined drift mode. For the first The probability of a predefined drift pattern.

[0115] The confidence score of the transfer learning model in identifying drift patterns indicates that the higher the confidence score, the more certain the recognition result of the transfer learning model. .

[0116] The expression for feature matching degree is:

[0117] ;

[0118] in, For cosine similarity, This is a nonlinear channel drift characteristic. the features of the known drift patterns selected from the historical data set that are most similar to the current features;

[0119] cosine similarity, the value range is [-1, 1], the closer to 1, the more matched the current features and historical experience, , mapped to the interval [0, 1].

[0120] The expression of the independent correction parameter consistency index is:

[0121] ;

[0122] wherein, is the coefficient of variation, is the standard deviation of the independent correction parameter, is the average value of the correction parameters obtained by different monodisperse standard microspheres;

[0123] The coefficient of variation measures the dispersion of the group of independent correction parameters. The smaller the CV, the higher the consistency .

[0124] The expression of the analysis confidence index is:

[0125] ;

[0126] wherein, is the analysis confidence index, is the weight of the fitting residual, is the classification confidence weight, is the feature matching degree weight, is the independent correction parameter consistency index weight.

[0127] S5.3, using the independent correction parameters corresponding to different monodisperse standard microspheres in the internal analysis data set, the relative deviation between the independent correction parameters is calculated.

[0128] Further, based on the consistency of the evaluation correction result, the internal analysis data set contains correction parameters independently calculated based on each kind of monodisperse standard microspheres. The relative standard deviation between the independent correction parameters can quantify the dispersion between them. Cross-validation is performed using multiple internal standards. If the correction parameters obtained by different internal standards are highly consistent, it indicates that the correction result is less affected by accidental errors of individual internal standard signals, and the result is stable and reliable. Otherwise, it indicates that there may be an anomaly.

[0129] The expression of the relative deviation is:

[0130] ;

[0131] wherein, is the relative deviation.

[0132] The relative standard deviation expression is:

[0133] ;

[0134] wherein, is the relative standard deviation.

[0135] S5.4 The analysis confidence index and the consistency evaluation index of the multi-internal standard correction result are weighted and fused to calculate the correction quality factor value.

[0136] Further, the analysis confidence index reflects the internal quality of the intelligent analysis process, and the consistency evaluation index of the multi-internal standard correction result reflects the external robustness of the correction result. By weighted fusion, the indexes evaluated from the process and the result are combined into a unified correction quality factor value, and the process quality and the result quality are double checked. Only when the analysis process is reliable and the result consistency is good, a high correction quality factor can be obtained, and a single and authoritative quantitative index that can fully represent the overall quality of this correction is generated.

[0137] The correction quality factor value expression is:

[0138] ;

[0139] wherein, is the correction quality factor value, is the analysis confidence, is the consistency evaluation index of the multi-internal standard correction result, is the analysis confidence index weight, is the consistency evaluation index of the multi-internal standard correction result weight.

[0140] S6, calculate the collection efficiency according to the corrected particle size distribution data, fit the collection efficiency curve, and based on the correction quality factor, quantify the measurement confidence interval of the collection efficiency evaluation result.

[0141] S6.1, using the particle number concentrations of the cutters upstream and downstream corresponding to each particle size point in the corrected particle size distribution data, calculate the collection efficiency value of each particle size point, take the particle size as the horizontal coordinate and the collection efficiency value as the vertical coordinate, use the reverse asymmetric S-shaped equation to fit the data points, and generate the collection efficiency curve.

[0142] Furthermore, in the core calculation step of evaluating the cutter's performance, the corrected particle size distribution data ensures the accuracy of the particle size coordinates. The physical basis for calculating the collection efficiency value at each particle size point is the law of conservation of particulate matter, that is, the ratio of downstream concentration to upstream concentration directly reflects the cutter's ability to collect particles of different sizes. An inverse asymmetric S-shaped equation is used for fitting. The S-shaped characteristic of the equation can accurately describe the transition zone from low to high collection efficiency, and its asymmetry can better fit the non-ideal penetration characteristics that may exist in actual cutters.

[0143] The expression for the collection efficiency value of the particle size point is:

[0144] ;

[0145] in, This represents the collection efficiency value corresponding to the i-th particle size point. The first measurement upstream of the cutter The number concentration of particulate matter at a known particle size point The first measurement downstream of the cutter The number concentration of particulate matter at a known particle size point.

[0146] The expression for the capture efficiency curve is:

[0147] ;

[0148] in, This is the capture efficiency curve.

[0149] S6.2 Read the aerodynamic diameter evaluation parameters from the capture efficiency curve, and assign a measurement confidence interval to the aerodynamic diameter evaluation parameters based on the mapping relationship between the corrected quality factor and historical performance calibration.

[0150] Furthermore, the key to quantifying the uncertainty of the evaluation results lies in directly reading the numerical points of the aerodynamic diameter evaluation parameters from the capture efficiency curve. The correction quality factor is a quantitative indicator that comprehensively evaluates the quality of the entire measurement and correction process. Based on the mapping relationship between the correction quality factor and historical performance calibration, it refers to clarifying the typical uncertainty levels of the final evaluation parameters corresponding to different ranges of correction quality factor values ​​through a database established from numerous previous experiments.

[0151] S7. Based on the measurement confidence interval of the quantified capture efficiency evaluation results, output a comprehensive evaluation report.

[0152] S7.1. Correlate and integrate the fitted capture efficiency curve and the measurement confidence interval data to form a complete dataset.

[0153] Further, the fitted collection efficiency curve contains the continuous function relationship of collection efficiency with particle size, and the measurement confidence interval provides a quantitative uncertainty range for the evaluation parameters. The correlation and integration operation is to bind these two parts of information in the data structure, and in the data file storing the collection efficiency curve, the upper and lower boundary data of the measurement confidence interval corresponding to each key evaluation parameter are attached.

[0154] S7.2, by comparing the measurement confidence interval of the collection efficiency evaluation result with the standard qualified range, obtaining the risk avoidance strategy, comparing the actual value of the evaluation parameter in the integration data set with the measurement confidence interval, determining the performance qualification of the cutter, and obtaining the visualization chart.

[0155] Further, when the actual value of the evaluation parameter and its entire measurement confidence interval are completely within the standard qualified range, it is determined to be qualified; when the actual value is within the qualified range, but the measurement confidence interval overlaps with the boundary of the qualified range, it is determined to be risk uncertain, and retesting is recommended; when the actual value and its entire measurement confidence interval are outside the qualified range, it is determined to be unqualified. Based on the information in the integration data set, a visualization chart containing the collection efficiency curve, the key parameter value, its measurement confidence interval, and the qualified range can be automatically generated. Based on the uncertainty analysis, a conservative decision is made to preferentially avoid the risk of misjudging unqualified products as qualified.

[0156] S7.3, based on the corrected collection efficiency curve and the actual value and measurement confidence interval of the evaluation parameter, obtaining a comprehensive evaluation report.

[0157] Further, based on the actual value of the evaluation parameter, the performance level is explained, and based on the measurement confidence interval, the reliability of the level value is clarified. The report will clearly record the test conditions adopted, the value of the correction quality factor, the conclusion of the qualification determination and its reasons. Data, charts and determination rules are converted into a normative technical document that meets the specifications, has clear conclusions, and has a complete evidence chain.

[0158] S7.4, integrating the visualization chart and the comprehensive evaluation report to obtain the performance conclusion of the cutter.

[0159] Further, the visualization chart visually displays data and determination process in a graphical manner, and the comprehensive evaluation report elaborates the entire process, data and conclusion in detail in the form of text. The chart is embedded as an important part of the report in the corresponding position of the document, forming a final deliverable with pictures and text, and with solid evidence.

[0160] The embodiment also provides a computer device suitable for the particle size channel correction method based on the trapping efficiency of the trapping efficiency evaluation system, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the particle size channel correction method based on the trapping efficiency of the trapping efficiency evaluation system proposed in the above embodiment.

[0161] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0162] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to realize the particle size channel correction method based on the trapping efficiency of the trapping efficiency evaluation system proposed in the above embodiment. The storage medium can be realized by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0163] To sum up, by using the transfer learning model pre-trained based on the historical data set to perform small sample self-adaptive analysis on the internal standard substance data, the complex nonlinear channel drift characteristics are accurately identified and corrected, and a high-precision dynamic nonlinear correction function is generated, thereby fundamentally solving the problem of insufficient precision of the traditional linear correction model, analyzing the comprehensive correction quality factor in the transfer learning model, realizing quantitative evaluation of the reliability of the correction process itself, and changing the black box operation into transparent and measurable quality control.

[0164] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. A method for correcting a size channel of a collection efficiency evaluation system based on a collection efficiency, characterized by: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

2. The particle size channel correction method based on the trapping efficiency evaluation system of trapping efficiency according to claim 1, characterized by: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

3. The particle size channel correction method based on the trapping efficiency evaluation system of trapping efficiency according to claim 2, characterized by: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the The mixed aerosols flow into the descending mixing chamber of the collection efficiency evaluation unit for thorough mixing and homogenization, and then are introduced into the static chamber of the collection efficiency evaluation unit. The three-way solenoid valve of the control collection efficiency evaluation unit is switched to the path of extracting mixed aerosols that have not been cutter, and enters the aerodynamic particle size spectrometer to measure and obtain the particle size distribution data of the mixed aerosols as they pass upstream of the cutter to be evaluated. The three-way solenoid valve of the control collection efficiency evaluation unit is switched to the path of extracting the mixed aerosol that has passed through the cutter to be evaluated, and enters the aerodynamic particle size spectrometer to measure and obtain the particle size distribution data of the mixed aerosol as it passes downstream of the cutter to be evaluated.

4. The particle size channel correction method based on trapping efficiency evaluation system of trapping efficiency according to claim 3, characterized by: The standard microsphere channel information from the raw particulate matter size distribution data is input into a transfer learning model pre-trained on historical datasets from a particle size channel benchmark database. Through small-sample adaptive analysis, nonlinear channel drift characteristics are identified, and a dynamic nonlinear correction function is generated. The process includes the following steps: From the raw data of particle size distribution, the channel numbers corresponding to the characteristic peak signals generated by monodisperse standard microspheres with known particle sizes are identified and extracted to form a set of standard microsphere channel information for the current measurement. A historical dataset corresponding to the normal and drift states was constructed using a particle size channel benchmark database. The transfer learning model was pre-trained to obtain the trained transfer learning model. The standard microsphere channel information set is input into the trained transfer learning model. Based on the pre-trained knowledge, domain adaptive analysis is performed on small sample data to identify the nonlinear channel drift characteristics that deviate from the baseline state in the original data of the current particle size distribution. Based on the identified nonlinear channel drift characteristics, a dynamic nonlinear correction function is output.

5. The particle size channel correction method based on trapping efficiency evaluation system of trapping efficiency according to claim 4, characterized by: The original particle size distribution data is batch-corrected using a dynamic nonlinear correction function to obtain corrected particle size distribution data. This process includes the following steps: The generated dynamic nonlinear correction function is read from the storage location of the output results after the transfer learning model identifies and analyzes the drift features of the current data, and the original particle size distribution data to be corrected is loaded at the same time. The input parameter of the channel number in the original particulate matter size distribution data is substituted into the dynamic nonlinear correction function, and the corrected particle size value is output after calculation for each channel value. The corrected particle size values ​​are used to replace the corresponding original channel values ​​in the original particulate matter size distribution data to form a new set of data pairs. The new dataset is integrated and arranged in order of particle size to obtain the corrected particle size distribution data.

6. The particle size channel correction method based on trapping efficiency evaluation system of trapping efficiency according to claim 5, characterized by: Based on the intrinsic analysis data of the transfer learning model in the process of identifying drift features, the corrected quality factor is calculated, including the following steps: The intrinsic analysis data set for this analysis process is extracted from the intermediate calculation results and state variables generated by the transfer learning model in the process of identifying nonlinear channel drift features. Based on the internal analysis dataset, the following parameters are calculated to form the analysis confidence index: the residual of the transfer learning model fitting, classification confidence, matching degree between nonlinear channel drift features and known drift pattern features, and consistency index of independent correction parameters for different monodisperse standard microspheres when the transfer learning model identifies drift features of the current input standard microsphere channel information. The relative deviations between the independent correction parameters are calculated using the independent correction parameters corresponding to different monodisperse standard microspheres in the intrinsic analysis dataset. The confidence index and the consistency evaluation index of the multi-internal standard correction results are weighted and fused to calculate the correction quality factor value.

7. The particle size channel correction method based on trapping efficiency evaluation system of trapping efficiency according to claim 6, characterized by: The trapping efficiency is calculated based on the corrected particle size distribution data, a trapping efficiency curve is fitted, and the measurement confidence interval of the trapping efficiency evaluation results is quantified based on the corrected quality factor, including the following steps: Using the particle number concentration at each particle size point upstream and downstream of the cutter in the corrected particle size distribution data, the collection efficiency value at each particle size point is calculated. Plotting particle size on the x-axis and collection efficiency value on the y-axis, the data points are fitted using an inverse asymmetric S-curve equation to generate a collection efficiency curve. The aerodynamic diameter evaluation parameters are read from the capture efficiency curve, and a measurement confidence interval is assigned to the aerodynamic diameter evaluation parameters based on the mapping relationship between the corrected quality factor and historical performance calibration.

8. The particle size channel correction method based on trapping efficiency evaluation system of trapping efficiency according to claim 7, characterized by, Based on the measurement confidence interval of the quantified capture efficiency evaluation results, a comprehensive evaluation report is output, including the following steps: The fitted capture efficiency curve and the measurement confidence interval data are correlated and integrated to form a complete dataset; By comparing and analyzing the measurement confidence interval and standard qualified range of the capture efficiency evaluation results, risk avoidance strategies are obtained. By comparing and integrating the actual values ​​of the evaluation parameters in the dataset with the measurement confidence interval, the performance qualification of the cutter is determined and a visual chart is obtained. A comprehensive evaluation report is obtained based on the corrected capture efficiency curve and the actual values ​​and measurement confidence intervals of the evaluation parameters. By integrating visualization charts and comprehensive evaluation reports, conclusions on the cutter's performance can be obtained. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: When the processor executes the computer program, it implements the steps of the particle size channel correction method based on the capture efficiency evaluation system according to any one of claims 1 to 8.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the particle size channel correction method based on the capture efficiency evaluation system according to any one of claims 1 to 8.

Images