Nested particulate matter concentration calibration method and device, terminal and storage medium

Abstract

The invention relates to the technical field of environment monitoring, in particular to a nested particulate matter concentration calibration method and device, a terminal and a storage medium. The method comprises the following steps: acquiring particulate matter concentration original data of a plurality of particle size channels acquired by each micro station, and respectively establishing independent inner layer consistency calibration models for the plurality of particle size channels, so as to benchmark the particulate matter concentration original data acquired by each micro station to be calibrated to reference data of a quality control micro station, then, an outer-layer transfer learning model is constructed based on corresponding data of the quality control micro station and the gold label station, so that cross-equipment difference mapping is realized; after the outer-layer transfer learning model is nested in the inner-layer consistency calibration model, a nested double-layer model is formed; and outputting finally calibrated multi-channel particulate matter concentration data based on the nested double-layer model. According to the method, the measurement accuracy, the model generalization ability and the deployment efficiency of the micro station in diversified field environments can be remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring technology, and in particular to a nested particulate matter concentration calibration method, device, terminal, and storage medium. Background Technology

[0002] With the acceleration of industrial automation and the upgrading of environmental monitoring needs, particulate matter concentration monitoring has become a key link in industrial production quality control and urban air quality management. Low-cost multi-channel particulate matter sensors are widely used in dense monitoring networks due to their flexible deployment and wide coverage. The accuracy of their data needs to be ensured through calibration, which has become a basic requirement for the industry to achieve efficient monitoring.

[0003] Current mainstream calibration schemes use data from gold standard monitoring stations (hereinafter referred to as gold standard stations) as the core reference and calibrate sensors through models such as linear regression and simple polynomial fitting. Some technologies rely on collecting a large amount of synchronous data between sensors and gold standard stations on-site to train dedicated calibration models, or periodically sending sensors to the standardized environment where gold standard stations are located to complete calibration in order to reduce measurement deviations.

[0004] However, existing technologies have significant limitations: the temperature and humidity fluctuations and dust interference in the field environment differ significantly from the standardized environment of the gold standard station, resulting in a substantial decrease in the accuracy of the calibrated data; under new monitoring scenarios or new sample types, it is difficult to quickly generate an effective calibration model, and it is often impractical to collect a sufficient number of synchronous samples on-site; at the same time, existing systems cannot effectively reuse historical calibration data, and each calibration must start from scratch, which not only wastes computing resources but also prolongs the deployment and response time of sensors. Summary of the Invention

[0005] This invention provides a nested particulate matter concentration calibration method, apparatus, terminal, and storage medium to address the problems of low calibration data accuracy, difficulty in quickly generating calibration models, and low deployment efficiency of calibration models in the prior art.

[0006] In a first aspect, embodiments of the present invention provide a nested particulate matter concentration calibration method, comprising: Obtain raw particulate matter concentration data from multiple particle size channels collected by each micro-station; Independent inner-layer consistency calibration models are established for each of the multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station. An outer transfer learning model is constructed based on the corresponding data of the quality control micro-station and the gold standard station to achieve cross-device difference mapping; The outer transfer learning model is nested after the inner consistency calibration model to form a nested two-layer model. The nested two-layer model outputs the final calibrated multi-channel particulate matter concentration data.

[0007] In one possible implementation, an independent inner-layer consistency calibration model is established for each of the multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to the reference data of the quality control micro-station, including: Calculate the Pearson correlation coefficient between the microstation to be calibrated and the corresponding channel data of the quality control microstation; Detect whether the Pearson correlation coefficient meets the first preset condition; If the Pearson correlation coefficient meets the first preset condition, then the inner consistency calibration model is constructed using a linear regression model.

[0008] In one possible implementation, after detecting whether the Pearson correlation coefficient satisfies a first preset condition, the method further includes: If the Pearson correlation coefficient does not meet the first preset condition, the root mean square error and coefficient of determination of the corresponding channel data of the micro-station to be calibrated and the quality control micro-station are calculated, and the inner consistency calibration model is constructed using a multinomial regression model. The inner consistency calibration model must satisfy that the root mean square error is less than or equal to the first preset value and the coefficient of determination is greater than or equal to the second preset value.

[0009] In one possible implementation, if the Pearson correlation coefficient satisfies a first preset condition, then a linear regression model is used to construct the inner-layer consistency calibration model, including: If the absolute value of the Pearson correlation coefficient is greater than or equal to the third preset value, then the inner consistency calibration model is constructed using a linear regression model.

[0010] In one possible implementation, the outer transfer learning model is constructed using a nonlinear supervised learning method such as a neural network, an XGBoost regression model, or support vector regression. The outer transfer learning model is trained based on paired data between the quality control micro-station and the gold standard station. The training input is the output data of the inner consistency calibration model, and the training output is the particulate matter standard concentration value of the gold standard station.

[0011] In one possible implementation, the outer transfer learning model is nested after the inner consistency calibration model to form a nested two-layer model, including: The outer transfer learning model is used as the outer structure and connected to the inner consistency calibration model of all channels to form the end-to-end nested two-layer model.

[0012] In one possible implementation, before outputting the final calibrated multichannel particulate matter concentration data based on the nested two-layer model, the method further includes: The nested two-layer model is deployed on the micro-site's local edge computing node, Docker container, or cloud platform.

[0013] Secondly, embodiments of the present invention provide a nested particulate matter concentration calibration device, comprising: The acquisition module is used to acquire raw particulate matter concentration data from multiple particle size channels collected by each micro-station; The inner layer model construction module is used to establish independent inner layer consistency calibration models for the multiple particle size channels, so as to calibrate the original particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station. The outer model building module is used to build an outer transfer learning model based on the corresponding data of the quality control micro-station and the gold standard station, so as to realize the difference mapping between devices; The processing module is used to nest the outer transfer learning model after the inner consistency calibration model to form a nested two-layer model; The processing module is also used to output the final calibrated multi-channel particulate matter concentration data based on the nested two-layer model.

[0014] Thirdly, embodiments of the present invention provide a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the nested particulate matter concentration calibration method as described in the first aspect or any possible implementation thereof.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the nested particulate matter concentration calibration method as described in the first aspect or any possible implementation thereof.

[0016] This invention provides a nested particulate matter concentration calibration method, apparatus, terminal, and storage medium. It acquires raw particulate matter concentration data from multiple particle size channels collected by each microstation, establishes independent inner-layer consistency calibration models for each particle size channel, and aligns the raw particulate matter concentration data collected by each microstation to the reference data of the quality control microstation. Then, based on the corresponding data from the quality control microstation and the gold standard station, an outer-layer transfer learning model is constructed to achieve cross-device difference mapping. The outer-layer transfer learning model is nested after the inner-layer consistency calibration model, forming a nested two-layer model. Finally, based on the nested two-layer model, the calibrated multi-channel particulate matter concentration data is output. This invention achieves consistency of channel data between the microstation and the quality control microstation through the inner-layer consistency calibration model, while the outer-layer transfer learning model effectively captures the nonlinear differences between the quality control microstation and the gold standard station, realizing concentration conversion mapping. Finally, the inner and outer models are merged to form a nested two-layer model, enabling the microstation calibration model to adapt to different deployment environments. This significantly improves the measurement accuracy, model generalization ability, and deployment efficiency of the microstation in diverse field environments. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the implementation of the nested particulate matter concentration calibration device method provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the particulate matter monitoring system provided in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the implementation of the method for establishing an inner-layer consistency calibration model provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of a nested two-layer model provided in an embodiment of the present invention; Figure 5 This is a schematic flowchart of the nested particulate matter concentration calibration method provided in the embodiments of the present invention; Figure 6 This is a schematic diagram of the structure of the nested particulate matter concentration calibration device provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the terminal provided in an embodiment of the present invention. Detailed Implementation

[0019] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0021] Figure 1 This is a flowchart illustrating the implementation of a nested particulate matter concentration calibration method provided in an embodiment of the present invention. This nested particulate matter concentration calibration method is primarily applicable to particulate matter monitoring systems with dense field deployments and requiring model transfer capabilities. See also... Figure 2 As shown, the particulate matter monitoring system consists of the following components: microstations deployed within the monitoring area, at least one gold-standard station, at least one relatively stable and high-performance microstation serving as a "quality control microstation," a nested calibration algorithm framework including an inner-layer channel consistency calibration model and an outer-layer migration calibration model, and a data acquisition and fusion platform for storage, modeling, inference, and model deployment. The inner-layer channel consistency calibration model is a model for achieving multi-channel consistency calibration between the microstations and the quality control microstations, while the outer-layer migration calibration model is a nonlinear calibration model for the quality control microstations to the gold-standard station.

[0022] The calibration method for nested particulate matter concentration is detailed below: Step 101: Obtain the raw particulate matter concentration data from multiple particle size channels collected by each micro-station.

[0023] Microstations can be low-cost multi-channel sensors, with each channel outputting particulate matter of a different size, collecting raw particulate matter concentration data from all channels. It should be noted that a microstation includes both a microstation to be calibrated and a microstation used for quality control.

[0024] Optionally, the sensor can be a 12-channel sensor; for example, multiple particle size channels may include 0.3... -0.5 0.5 -1.0 1.0 -2.5 2.5 -4.0 4.0 -6.0 6.0 -7.0 7.0 -8.0 8.0 -10.0 10.0 -12.5 12.5 -15 15.0 -20.0 20.0 The above 12 particle size channels.

[0025] To establish a multi-channel consistency calibration model between the microstation and the quality control microstation, it is necessary to collect 12 channels of raw particle size number concentration data of the microstation to be calibrated and the quality control microstation within the same time period, and to preprocess all the raw particulate matter concentration data obtained.

[0026] Optionally, preprocessing of all acquired raw particulate matter concentration data may include time alignment, missing value imputation, and outlier removal to obtain time-aligned, complete, and accurate particulate matter concentration data for the microstation to be calibrated and the quality control microstation.

[0027] Optionally, the obtained time-aligned particulate matter concentration data can be denoted as: , ; in, Indicates the first microstation to be calibrated The dataset corresponding to each channel This indicates the first quality control micro-station The dataset corresponding to each channel Indicates the first microstation to be calibrated The first channel Group of particulate matter concentration data, This indicates the first quality control micro-station The first channel Group of particulate matter concentration data, , This indicates the number of data sets for particulate matter concentration.

[0028] Step 102: Establish independent inner-layer consistency calibration models for multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station.

[0029] It should be noted that the quality control micro-station is a pre-set micro-station in the system with relatively stable configuration and excellent performance. Its core function is to provide a reference benchmark for data alignment for ordinary micro-stations to be calibrated, which is the basic premise for the reliability of its reference data.

[0030] The reference data of the quality control micro-station comes from the raw particulate matter concentration data collected by it from multiple particle size channels, and the particulate matter concentration data is obtained after a preprocessing process.

[0031] In this embodiment, to determine whether there is a significant linear relationship between the channels of the microstation to be calibrated and the quality control microstation, the system calculates the Pearson correlation coefficient of the corresponding channel based on the particulate matter concentration data collected during the synchronous time period, and makes a linear determination based on the Pearson correlation coefficient.

[0032] In one embodiment, see Figure 3 As shown, establishing independent inner-layer consistency calibration models for multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to the reference data of the quality control micro-station may include the following steps.

[0033] Step 301: Calculate the Pearson correlation coefficient between the corresponding channel data of the micro-station to be calibrated and the quality control micro-station.

[0034] Optional, according to Calculate the Pearson correlation coefficient between the corresponding channel data of the microstation to be calibrated and the quality control microstation; in, express The mean, express The mean, This represents the Pearson correlation coefficient between the corresponding channel data of the microstation to be calibrated and the quality control microstation. Used to measure the linearity between channels, with values ​​ranging from [-1, 1]. The closer it is to ±1, the stronger the linearity between the corresponding channels.

[0035] Step 302: Detect whether the Pearson correlation coefficient meets the first preset condition.

[0036] Detect whether the Pearson correlation coefficient is greater than or equal to the third preset value.

[0037] The third preset value here is a value set according to requirements. In this embodiment, the value of the third preset value is not limited. For example, the third preset value can be 0.8, 0.85 or 0.9, etc.

[0038] Step 303: If the Pearson correlation coefficient meets the first preset condition, then the inner consistency calibration model is constructed using a linear regression model.

[0039] Optionally, if the absolute value of the Pearson correlation coefficient is greater than or equal to the third preset value, a linear regression model is used to construct the inner-layer consistency calibration model. That is, if the absolute value of the Pearson correlation coefficient is greater than or equal to the third preset value, it is considered that the consistency of the particulate matter concentration data for the corresponding channel is strong, the corresponding channel shows a significant linear relationship, satisfying the linearity criterion, and then an independent linear function is established for that particle size channel. The goal is to calibrate the same channel across different devices to the level of a quality control microstation. This represents the particulate matter concentration value after consistency calibration. Represents the channel coefficient. This represents the particulate matter concentration data corresponding to the original channel. It represents a constant.

[0040] Step 304: If the Pearson correlation coefficient does not meet the first preset condition, calculate the root mean square error and coefficient of determination of the corresponding channel data of the micro-station to be calibrated and the quality control micro-station, and construct an inner-layer consistency calibration model using a multinomial regression model. The inner-layer consistency calibration model must satisfy that the root mean square error is less than or equal to the first preset value and the coefficient of determination is greater than or equal to the second preset value.

[0041] If the absolute value of the Pearson correlation coefficient is less than the third preset value, then the root mean square error and coefficient of determination of the corresponding channel data of the micro-station to be calibrated and the quality control micro-station are calculated.

[0042] In As the independent variable, Using [variable name] as the dependent variable, fit a linear regression. Then, calculate the residuals: ; The root mean square error (RMSE) is defined as follows: ; The coefficient of determination is defined as: .

[0043] If the absolute value of the Pearson correlation coefficient is less than 0.85, it is determined that the corresponding channel does not have a significant linear relationship, and a multinomial regression method is used to build an inner-layer consistency calibration model. This method constructs the following model through least-squares fitting: ; Among them, order Based on the fitting results and validation errors, 2nd to 3rd order models are preferentially selected to balance fitting accuracy and inner-layer consistency. Model stability is then calibrated after fitting. The effectiveness of the inner-layer consistency calibration model is verified by the RMSE index. That is, the inner-layer consistency calibration model is effective when the root mean square error is less than or equal to the first preset value and the coefficient of determination is greater than or equal to the second preset value.

[0044] The values ​​of the first preset value and the second preset value can be set according to requirements. In this embodiment, the values ​​of the first preset value and the second preset value are not limited. For example, the first preset value can be... The second preset value can be 0.75. When the root mean square error is less than or equal to... Furthermore, the inner-layer consistency calibration model is effective when the coefficient of determination is greater than or equal to 0.75.

[0045] After processing by the inner layer consistency calibration model, the particulate matter concentration data of the corresponding channel after calibration is obtained, and the calibration is completed.

[0046] If the root mean square error is greater than If the coefficient of determination is less than 0.75, the inner-layer consistency calibration model is considered invalid, and calibration ends. In this case, staff need to check the micro-stations and / or quality control micro-stations before re-performing channel consistency calibration.

[0047] After completing the channel consistency calibration, a set of calibration values ​​calculated by fitting for the 12 particle size channels of the microstation will be output. This output data can be used as the input feature vector for the subsequent outer-layer transfer learning model. The constructed consistency calibration model improves the data consistency among sensors from the same source while meeting the system's requirement for reliable alignment of different microstations under the same environment.

[0048] Step 103: Construct an outer transfer learning model based on the corresponding data of the quality control micro-station and the gold standard station to achieve cross-device difference mapping.

[0049] To achieve accurate alignment between micro-station data and gold-standard station data, an outer-layer transfer learning model was constructed after completing the inner-layer channel consistency calibration. This model aims to improve the overall accuracy of concentration estimation and its cross-device versatility. The outer-layer transfer learning model uses the output of the "quality control micro-station" as input and the "gold-standard station" as the target, employing a nonlinear mapping model to calculate particulate matter mass concentrations (PM2.5, PM10).

[0050] In one embodiment, the outer transfer learning model is constructed using a nonlinear supervised learning method such as a neural network, XGBoost regression model, or support vector regression. The outer transfer learning model is trained based on paired data from quality control microstations and gold standard stations. The training input is the output data of the inner consistency calibration model, and the training output is the particulate matter standard concentration value of the gold standard station.

[0051] Optionally, the specific steps for constructing and training the outer transfer learning model include: Data preparation: The particulate matter concentration data of the 12 channels of the quality control micro-station were selected as the input feature vector, and the PM2.5 and PM10 or total suspended particulates (TSP) concentration data of the gold standard station were selected as the training target. Feature processing: Feature transformation is performed on the 12-channel particulate matter concentration data, including normalization, principal component analysis, or the introduction of environmental covariates (such as temperature and humidity) to enhance the model's expressive power; Model Construction and Training: The outer transfer learning model can employ nonlinear supervised learning algorithms, including but not limited to neural networks, XGBoost regression models, or support vector regression. The feature-processed data is input into the outer transfer learning model to train the mapping relationship between it and the particulate matter standard concentration values ​​from the gold standard monitoring stations. Model Evaluation and Selection: Cross-validation was used to evaluate the outer-layer transfer learning model, based on mean squared error, The optimal model is selected based on indicators such as the value. The optimal model can be transferred to other microstations, with the input replaced by the 12-channel consistency calibration results (forming a 12-dimensional channel concentration vector) output by the microstation after processing by the inner consistency calibration model, and the output being the estimated high-confidence standard particulate matter concentration value.

[0052] This outer-layer transfer learning model utilizes high-quality test samples from the source domain to construct a nonlinear mapping relationship, and achieves concentration correction and unified benchmark conversion for the target micro-stations through transfer applications.

[0053] Step 104: The outer transfer learning model is nested after the inner consistency calibration model to form a nested two-layer model.

[0054] In one embodiment, this step may include: connecting the outer transfer learning model as the outer structure to the inner consistency calibration model of all channels to form an end-to-end nested two-layer model, which is then packaged into a unified inference process to improve efficiency and stability. See also Figure 4 The diagram shows a nested two-layer model.

[0055] The nested two-level model can be represented as ; in, This represents the raw multi-channel particulate matter concentration data collected by any micro-station to be calibrated. This indicates the calibrated particulate matter concentration data. This represents the inner-layer consistency calibration model, used for calibration of each channel. This represents the outer transfer learning model, used to output the calibrated multichannel particulate matter concentration at the corresponding time point.

[0056] The nested two-layer model can be migrated and deployed on the local edge computing node or Docker container of the micro-site to achieve rapid on-site application, or it can be remotely performed on the cloud platform for centralized inference, with good scalability and platform adaptability.

[0057] Nested two-level model outputs can be configured as needed, supporting the following two types of results: 1. Pollutant concentration output (main output): Outputs target concentration values ​​for PM2.5, PM10, or TSP; 2. Channel Calibration Value Output (Optional): Simultaneously outputs the calibration concentration values ​​of 12 particle size channels generated by the internal consistency calibration layer model, facilitating intermediate state analysis, channel performance monitoring, and multi-scale data fusion processing. This meets the needs of various scenarios such as environmental monitoring, data dissemination, and pollution source tracing.

[0058] Step 105: Output the final calibrated multi-channel particulate matter concentration data based on the nested two-layer model.

[0059] See Figure 5 The flowchart of the nested particulate matter concentration calibration method shown above first constructs an inner-layer consistency calibration model between ordinary micro-stations and quality control micro-stations. Then, an outer-layer transfer learning model is constructed between the quality control micro-stations and gold standard stations. The inner-layer consistency calibration model and the outer-layer transfer learning model constitute a nested two-layer model, which is used to output the calibrated particulate matter concentration values ​​of channels such as PM2.5, PM10, or TSP.

[0060] To ensure the long-term effectiveness of the nested two-layer model, the following dynamic maintenance mechanism is also provided: 1. Performance monitoring: Continuously collect residual changes between the quality control micro-station and the gold standard station. If the residual exceeds the preset threshold, the nested two-layer model retraining will be triggered. 2. Model update mechanism: Supports updating the outer transfer learning model at preset times, and can also perform incremental training on the inner consistency calibration model; the preset time can be monthly, quarterly, etc.

[0061] 3. Version control and rollback: All nested two-level model versions are automatically stored, supporting model rollback, historical error tracking, and adaptation report generation.

[0062] This invention provides a nested particulate matter concentration calibration method. It acquires raw particulate matter concentration data from multiple particle size channels collected by each microstation, establishes independent inner-layer consistency calibration models for each particle size channel, and aligns the raw particulate matter concentration data collected by each microstation to the reference data of the quality control microstation. Then, based on the corresponding data of the quality control microstation and the gold standard station, an outer-layer transfer learning model is constructed to achieve cross-device difference mapping. The outer-layer transfer learning model is nested after the inner-layer consistency calibration model, forming a nested two-layer model. Finally, the calibrated multi-channel particulate matter concentration data is output based on the nested two-layer model. The inner-layer consistency calibration model achieves consistency of channel data between the microstation and the quality control microstation, while the outer-layer transfer learning model effectively captures the nonlinear differences between the quality control microstation and the gold standard station, realizing concentration conversion mapping. Finally, the inner and outer models are merged to form a nested two-layer model, enabling the microstation calibration model to adapt to different deployment environments. This significantly improves the measurement accuracy, model generalization ability, and deployment efficiency of the microstation in diverse field environments.

[0063] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0064] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0065] Figure 6 A schematic diagram of a nested particulate matter concentration calibration device provided by an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 6 As shown, a nested particulate matter concentration calibration device includes: an acquisition module 61, an inner model construction module 62, an outer model construction module 63, and a processing module 64.

[0066] The acquisition module 61 is used to acquire the raw particulate matter concentration data of multiple particle size channels collected by each micro-station; The inner layer model construction module 62 is used to establish independent inner layer consistency calibration models for multiple particle size channels, so as to calibrate the raw particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station. The outer model building module 63 is used to build an outer transfer learning model based on the corresponding data of the quality control micro-station and the gold standard station, so as to realize the difference mapping between devices; Processing module 64 is used to nest the outer transfer learning model after the inner consistency calibration model to form a nested two-layer model; The processing module 64 is also used to output the final calibrated multi-channel particulate matter concentration data based on a nested two-layer model.

[0067] In one possible implementation, the inner model building module 62 establishes independent inner consistency calibration models for multiple particle size channels, so as to calibrate the raw particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station, for the following purposes: Calculate the Pearson correlation coefficient between the corresponding channel data of the microstation to be calibrated and the quality control microstation; Detect whether the Pearson correlation coefficient meets the first preset condition; If the Pearson correlation coefficient meets the first preset condition, then a linear regression model is used to construct an inner-layer consistency calibration model.

[0068] In one possible implementation, after the inner model building module 62 detects whether the Pearson correlation coefficient meets the first preset condition, it is further used for: If the Pearson correlation coefficient does not meet the first preset condition, the root mean square error and coefficient of determination of the corresponding channel data of the micro-station to be calibrated and the quality control micro-station are calculated, and an inner consistency calibration model is constructed using a multinomial regression model. The inner consistency calibration model must satisfy that the root mean square error is less than or equal to the first preset value and the coefficient of determination is greater than or equal to the second preset value.

[0069] In one possible implementation, if the Pearson correlation coefficient meets a first preset condition, then when the inner model building module 62 constructs the inner consistency calibration model using a linear regression model, it is used for: If the absolute value of the Pearson correlation coefficient is greater than or equal to the third preset value, then a linear regression model is used to construct an inner-layer consistency calibration model.

[0070] In one possible implementation, the outer transfer learning model is constructed using a nonlinear supervised learning method such as a neural network, an XGBoost regression model, or a support vector regression. The outer transfer learning model is trained based on paired data from quality control microstations and gold standard stations. The training input is the output data of the inner consistency calibration model, and the training output is the particulate matter standard concentration value of the gold standard station.

[0071] In one possible implementation, when processing module 64 nests the outer transfer learning model after the inner consistency calibration model to form a nested two-layer model, it is used for: By connecting the outer transfer learning model as the outer structure to the inner consistency calibration model of all channels, an end-to-end nested two-layer model is formed.

[0072] In one possible implementation, before the processing module 64 outputs the final calibrated multichannel particulate matter concentration data based on the nested two-layer model, it is also used for: The nested two-layer model can be deployed on local edge computing nodes of the micro-site, in Docker containers, or on cloud platforms.

[0073] The above embodiment provides a nested particulate matter concentration calibration device. An acquisition module obtains raw particulate matter concentration data from multiple particle size channels collected by each micro-station. An inner model construction module establishes independent inner consistency calibration models for each particle size channel to align the raw particulate matter concentration data collected by each micro-station to the reference data of the quality control micro-station. Then, an outer model construction module constructs an outer transfer learning model based on the corresponding data from the quality control micro-station and the gold standard station to achieve cross-device difference mapping. The processing module nests the outer transfer learning model after the inner consistency calibration model, forming a nested two-layer model, and outputs the final calibrated multi-channel particulate matter concentration data based on the nested two-layer model. The inner consistency calibration model achieves consistency of channel data between micro-stations, while the outer transfer learning model effectively captures the nonlinear differences between the quality control micro-station and the gold standard station, achieving concentration conversion mapping. Finally, the inner and outer models are merged to form a nested two-layer model, enabling the micro-station calibration model to adapt to different deployment environments. This significantly improves the measurement accuracy, model generalization ability, and deployment efficiency of micro-stations in diverse field environments.

[0074] Figure 7 This is a schematic diagram of a terminal provided in an embodiment of the present invention. Figure 7 As shown, the terminal 7 in this embodiment includes a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70. When the processor 70 executes the computer program 72, it implements the steps in the various nested particulate matter concentration calibration method embodiments described above, for example... Figure 1 Steps 101 to 105 are shown. Alternatively, when processor 70 executes computer program 72, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 6 The functions of each module / unit are shown.

[0075] For example, computer program 72 can be divided into one or more modules / units, one or more of which are stored in memory 71 and executed by processor 70 to complete the present invention. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 72 in terminal 7. For example, computer program 72 can be divided into... Figure 6 The modules / units shown are shown.

[0076] Terminal 7 may include, but is not limited to, processor 70 and memory 71. Those skilled in the art will understand that... Figure 7This is merely an example of terminal 7 and does not constitute a limitation on terminal 7. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0077] The processor 70 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0078] The memory 71 can be an internal storage unit of the terminal 7, such as a hard disk or RAM of the terminal 7. The memory 71 can also be an external storage device of the terminal 7, such as a plug-in hard disk, Smart MediaCard (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal 7. Furthermore, the memory 71 can include both internal and external storage units of the terminal 7. The memory 71 is used to store computer programs and other programs and data required by the terminal. The memory 71 can also be used to temporarily store data that has been output or will be output.

[0079] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0080] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0081] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0082] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0083] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0084] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0085] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various nested particulate matter concentration calibration method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0086] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A nested particulate matter concentration calibration method, characterized in that, include: Obtain raw particulate matter concentration data from multiple particle size channels collected by each micro-station; Independent inner-layer consistency calibration models are established for each of the multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station. An outer transfer learning model is constructed based on the corresponding data of the quality control micro-station and the gold standard station to achieve cross-device difference mapping; The outer transfer learning model is nested after the inner consistency calibration model to form a nested two-layer model. The nested two-layer model outputs the final calibrated multi-channel particulate matter concentration data.

2. The nested particulate matter concentration calibration method according to claim 1, characterized in that, Independent inner-layer consistency calibration models are established for each of the multiple particle size channels to calibrate the raw particulate matter concentration data collected by each micro-station to the reference data of the quality control micro-station, including: Calculate the Pearson correlation coefficient between the microstation to be calibrated and the corresponding channel data of the quality control microstation; Detect whether the Pearson correlation coefficient meets the first preset condition; If the Pearson correlation coefficient meets the first preset condition, then the inner consistency calibration model is constructed using a linear regression model.

3. The nested particulate matter concentration calibration method according to claim 2, characterized in that, After detecting whether the Pearson correlation coefficient meets the first preset condition, the method further includes: If the Pearson correlation coefficient does not meet the first preset condition, the root mean square error and coefficient of determination of the corresponding channel data of the micro-station to be calibrated and the quality control micro-station are calculated, and the inner consistency calibration model is constructed using a multinomial regression model. The inner consistency calibration model must satisfy that the root mean square error is less than or equal to the first preset value and the coefficient of determination is greater than or equal to the second preset value.

4. The nested particulate matter concentration calibration method according to claim 2, characterized in that, If the Pearson correlation coefficient meets the first preset condition, then the inner consistency calibration model is constructed using a linear regression model, including: If the absolute value of the Pearson correlation coefficient is greater than or equal to the third preset value, then the inner consistency calibration model is constructed using a linear regression model.

5. The nested particulate matter concentration calibration method according to any one of claims 1-4, characterized in that, The outer layer transfer learning model is constructed using a nonlinear supervised learning method such as a neural network, XGBoost regression model, or support vector regression. The outer layer transfer learning model is trained based on paired data between the quality control micro-station and the gold standard station. The training input is the output data of the inner layer consistency calibration model, and the training output is the particulate matter standard concentration value of the gold standard station.

6. The nested particulate matter concentration calibration method according to claim 5, characterized in that, The outer transfer learning model is nested after the inner consistency calibration model to form a nested two-layer model, including: The outer transfer learning model is used as the outer structure and connected to the inner consistency calibration model of all channels to form the end-to-end nested two-layer model.

7. The nested particulate matter concentration calibration method according to claim 6, characterized in that, Before outputting the final calibrated multi-channel particulate matter concentration data based on the nested two-layer model, the following steps are also included: The nested two-layer model is deployed on the micro-site's local edge computing node, Docker container, or cloud platform.

8. A nested particulate matter concentration calibration device, characterized in that, include: The acquisition module is used to acquire raw particulate matter concentration data from multiple particle size channels collected by each micro-station; The inner layer model construction module is used to establish independent inner layer consistency calibration models for the multiple particle size channels, so as to calibrate the original particulate matter concentration data collected by each micro-station to be calibrated to the reference data of the quality control micro-station. The outer model building module is used to build an outer transfer learning model based on the corresponding data of the quality control micro-station and the gold standard station, so as to realize the difference mapping between devices; The processing module is used to nest the outer transfer learning model after the inner consistency calibration model to form a nested two-layer model; The processing module is also used to output the final calibrated multi-channel particulate matter concentration data based on the nested two-layer model.

9. A terminal, comprising a memory and a processor, the memory for storing a computer program, the processor for calling and running the computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the nested particulate matter concentration calibration method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the nested particulate matter concentration calibration method as described in any one of claims 1 to 7.

Images