Atmospheric pollutant concentration monitoring and early warning method and system

Abstract

The invention relates to an atmospheric pollutant concentration monitoring and early warning method and system, and belongs to the technical field of environmental protection informatization and environmental pollution prevention and control. The system comprises a multivariate data acquisition module, an analysis and early warning module, a linkage control module and a remote monitoring platform; the multivariate data acquisition module comprises an infrared spectrum absorption type gas sensor, an electrochemical gas sensor, a temperature and humidity sensor and a wind speed and direction sensor, and the multivariate data acquisition module is deployed in a monitoring area in a distributed manner and is used for acquiring pollutant concentration and environmental data. According to the invention, comprehensive and accurate monitoring can be realized, through cooperation of multi-principle sensors and dynamic weight fusion, limitation of a single sensor is overcome, accurate and reliable monitoring of multiple pollutants in a full concentration range is realized, monitoring blind areas are significantly reduced, advanced early warning and efficient closed-loop management and control can be realized, and the system has good application prospects. And the speed and the accurate management and control capability of the environmental pollution emergency response are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of environmental information technology and environmental pollution prevention and control, and in particular to a method and system for monitoring and early warning of atmospheric pollutant concentrations. Background Technology

[0002] With the acceleration of industrialization and urbanization, regional and complex air pollution problems are becoming increasingly prominent, posing a serious threat to the ecological environment and public health. Achieving accurate, real-time, and advanced early warning of air pollutants is crucial for improving the precision of environmental management and effectively carrying out pollution prevention and control. Existing atmospheric environmental monitoring and early warning technologies mainly have the following shortcomings:

[0003] First, the monitoring data sources are either singular or highly homogeneous. Most systems rely on isolated data from sensors based on a single principle or from standard monitoring stations. The former suffers from problems such as limited measurement range, cross-interference, and susceptibility to environmental influences. The latter, while providing accurate data, is costly and sparsely distributed, making it difficult to achieve comprehensive perception with high spatiotemporal resolution and effectively capture the spatial distribution and dynamic transmission process of pollutants.

[0004] Secondly, the early warning capabilities are lagging behind and the level of intelligence is low. Most existing early warnings are based on real-time concentration exceeding alarms with fixed thresholds, lacking forward-looking prediction of pollution trends. The few methods that introduce prediction models mostly use traditional statistical or linear models, which are insufficient in characterizing the nonlinearity and non-stationarity of pollutant concentration changes as well as the complex coupling relationship between multiple factors such as meteorology and traffic, resulting in short lead times and low accuracy of early warnings.

[0005] Furthermore, data quality and system reliability face challenges. Factors such as temperature and humidity changes, particulate matter pollution, and electromagnetic interference in open environments can degrade sensor performance and generate abnormal data. Existing methods often lack effective multi-source data verification and fusion mechanisms, as well as robust processing capabilities for abnormal data, affecting the reliability of monitoring data and the accuracy of early warning decisions.

[0006] Furthermore, the monitoring, early warning, and response processes are disconnected. Most existing solutions stop at data display and alarms, failing to form a closed-loop management system of "perception, analysis, decision-making, and control," which makes it difficult to support rapid and automated emergency response and control measures. Summary of the Invention

[0007] This invention provides a method and system for monitoring and early warning of atmospheric pollutant concentrations, in order to solve the problems of monitoring blind spots, range limitations, insufficient prediction capabilities, and poor anti-interference capabilities in existing environmental gas monitoring technologies.

[0008] The solution to the above-mentioned technical problems of the present invention is as follows: an atmospheric pollutant concentration monitoring and early warning method, comprising the following steps: S1, by deploying multiple monitoring nodes in the monitoring area, synchronously collecting concentration data of target atmospheric pollutants and environmental meteorological data;

[0009] The monitoring node integrates sensors based on different measurement principles, including at least an infrared spectroscopy absorption gas sensor, an electrochemical gas sensor, a temperature and humidity sensor, and a wind speed and direction sensor; the collected data is transmitted to the central processing unit via a communication network.

[0010] S2, the central processing unit receives data from different monitoring nodes and sensors based on different principles; it adopts a dual-channel data verification mechanism with main and backup, and dynamically assigns differentiated fusion weights to sensors based on different principles based on the historical measurement accuracy data of various types of sensors for different pollutants in different concentration ranges, and calculates the optimized pollutant concentration fusion value.

[0011] S3. Based on the fused historical and real-time concentration time series data obtained in step S2, the improved Prophet prediction model is used to predict the trend of pollutant concentration in the future period.

[0012] The improved Prophet model uses a logistic growth trend term combined with a mutation point adjustment function, and employs wavelet basis functions to fit non-stationary seasonal fluctuations.

[0013] S4. The concentration prediction results, real-time concentration change rate and environmental meteorological data obtained in step S3 are combined and input into the trained Cauchy function weighted least squares support vector machine classification model for auxiliary judgment.

[0014] Finally, the warning level is determined based on the model output and the preset multi-level concentration thresholds;

[0015] S5. Based on the warning level determined in step S4, generate and issue control commands to trigger the corresponding emergency response equipment or control system to execute preset emergency measures.

[0016] S6. Upload the raw data, processing results, early warning decisions and control execution status from steps S1 to S5 to the remote monitoring platform in real time for visualization, comprehensive analysis and storage recording.

[0017] Based on the above technical solution, the present invention can be further improved as follows.

[0018] Furthermore, the dynamic weight allocation in step S2 specifically involves: assigning weights to sensors based on different principles according to the pollutant type and the current concentration range; wherein, when the target pollutant is in a low concentration range, the infrared spectral absorption gas sensor is assigned a higher fusion weight than the electrochemical gas sensor; and when the target pollutant is in its electrochemical response characteristic concentration range, the electrochemical gas sensor is assigned a higher fusion weight.

[0019] This improvement enables data fusion to fully utilize the high precision and selectivity of infrared sensors in the low concentration range, as well as the sensitive response characteristics of electrochemical sensors to specific gases, thereby achieving optimal fusion accuracy across the entire concentration range of pollutants. This effectively avoids the performance limitations of single sensors in specific concentration ranges and significantly improves the reliability and accuracy of monitoring data.

[0020] Furthermore, the preset multi-level concentration thresholds and warning levels mentioned in step S4 include:

[0021] Level 1 warning: triggered when the predicted concentration reaches a set percentage of the ambient air quality standard limit;

[0022] Level 2 alarm: Triggered when the predicted concentration reaches or exceeds the ambient air quality standard limit;

[0023] Emergency control alarm: Triggered when the predicted concentration reaches the preset emergency control threshold, automatically initiating strong intervention measures;

[0024] For example, regarding fine particulate matter (PM2.5):

[0025] The Level 1 warning (attention) threshold can be set as 80% of the predicted concentration in the next hour > 75 μg / m³ (the secondary standard limit for the 24-hour average), i.e. > 60 μg / m³. 3 ;

[0026] The level 2 alarm (early warning) threshold is a predicted concentration of ≥75 μg / m³ within the next hour. 3 ;

[0027] The emergency control alarm threshold can be set to a predicted concentration ≥150 μg / m³ for the next hour. 3 ;

[0028] Regarding sulfur dioxide (SO2):

[0029] The primary warning threshold can be set as 80% of the predicted concentration in the next hour > 150 μg / m³ (the secondary standard limit for the 1-hour average), i.e. > 120 μg / m³. 3 ;

[0030] The level 2 alarm threshold is a predicted concentration of ≥150 μg / m³ within the next hour. 3 ;

[0031] The emergency control alarm threshold can be set to a predicted concentration of ≥500 μg / m³ within the next hour. 3 ;

[0032] The above values ​​are for illustrative purposes only. In actual applications, they can be dynamically configured according to the monitoring target, regional characteristics, and control requirements.

[0033] By setting up multi-level, progressive early warning thresholds and response mechanisms, refined hierarchical management from risk alerts to mandatory intervention has been achieved. Level 1 alerts provide forward-looking risk warnings, allowing time for manual prediction and preparation; Level 2 alarms activate standard emergency procedures; and emergency control alarms automatically execute mandatory interventions when facing serious pollution risks. This greatly shortens the delay in emergency response, improves the timeliness and effectiveness of pollution control, and helps prevent the pollution situation from further deteriorating.

[0034] An atmospheric pollutant concentration monitoring and early warning system includes a multi-source data acquisition module, an analysis and early warning module, a linkage control module, and a remote monitoring platform. The multi-source data acquisition module includes an infrared spectral absorption gas sensor, an electrochemical gas sensor, a temperature and humidity sensor, and a wind speed and direction sensor. The multi-source data acquisition module is distributed and deployed in the monitoring area to collect pollutant concentration and environmental data.

[0035] The analysis and early warning module includes a data fusion unit, a predictive analysis unit, and an early warning judgment unit. The data fusion unit is connected to the multi-source data acquisition module through a communication network and is used to receive and process data.

[0036] The data fusion unit executes a primary and backup dual-channel verification and dynamic weighted fusion algorithm to output a fusion concentration value. The prediction and analysis unit runs an improved Prophet prediction model, receives the fusion concentration value, and outputs a concentration prediction trend. The early warning judgment unit runs a Cauchy function weighted least squares support vector machine model, receives prediction results and related data, and outputs an early warning level signal.

[0037] The linkage control module is connected to the analysis and early warning module, receives early warning level signals, and automatically controls the actuator to respond.

[0038] The remote monitoring platform is connected to the analysis and early warning module for data visualization, analysis, management, and storage.

[0039] Furthermore, the infrared spectral absorption gas sensor uses a tunable semiconductor laser as a light source and utilizes the characteristic absorption lines of the target gas in the infrared band, combined with wavelength modulation and second harmonic detection technology for measurement.

[0040] This gives the sensor high selectivity, high sensitivity, and non-contact measurement capabilities. Utilizing characteristic absorption lines allows for precise identification of specific gases, effectively avoiding cross-interference; wavelength modulation and second harmonic detection techniques significantly suppress background noise and improve the signal-to-noise ratio, thus enabling stable and accurate measurement of low-concentration pollutants in complex open environments and extending the sensor's lifespan.

[0041] Furthermore, the monitoring and early warning system also includes an optical fiber sensor network, which uses space division multiplexing technology to connect multiple fiber optic grating sensors to a single trunk optical fiber and switches the acquisition of various signals through optical switches.

[0042] This greatly enhances the system's scalability and deployment flexibility. Through spatial multiplexing technology, a single optical fiber can connect a large number of sensor nodes, significantly reducing the number and complexity of cables required for traditional cabling, thus lowering deployment costs and maintenance difficulties. The inherent electromagnetic interference resistance, corrosion resistance, and long-distance transmission capabilities of fiber optic sensing make it particularly suitable for building stable and reliable distributed monitoring networks in complex environments with large-scale and strong interference, such as industrial parks and major transportation routes.

[0043] Furthermore, the data fusion unit dynamically assigns fusion weights to the infrared spectral absorption gas sensor and the electrochemical gas sensor based on the pollutant type and current concentration range. This enables it to intelligently select the most reliable data source for fusion based on different monitoring targets and real-time concentration levels. This not only improves data accuracy at the algorithm level but also ensures stable and reliable fusion results even when the performance of some sensors fluctuates or is limited, thus enhancing the overall robustness of the system.

[0044] Furthermore, the early warning judgment unit is configured to run a Cauchy function-weighted least squares support vector machine classification model. This model reduces the impact of abnormal samples on classification decisions by introducing a Cauchy weight function based on prediction error into its optimization objective function for the training samples. The logistic growth model is more in line with the physical laws of pollutant accumulation and saturation. The mutation point adjustment function can automatically adapt to drastic concentration changes caused by sudden emissions or sudden changes in meteorological conditions. By using wavelet basis functions instead of traditional models, it can more flexibly and accurately fit the non-stationary periodic fluctuations caused by human activities (such as morning and evening rush hours and holidays), thereby more realistically reflecting the evolution of pollution and achieving more reliable early warning.

[0045] Furthermore, the Cauchy function-weighted least squares support vector machine model configured in the early warning judgment unit reduces the impact of abnormal samples by introducing an error-based Cauchy weight function for each sample in the optimization objective.

[0046] In actual monitoring, sensor momentary failures, abnormal environmental interference, etc., can generate erroneous data (abnormal samples). The Cauchy weight function can automatically identify and reduce the weight of these abnormal data in model training and decision-making, preventing them from "misleading" the classification boundary. This ensures that the early warning judgment remains stable and accurate even when the data is noisy, effectively reducing the false alarm and false negative rates.

[0047] Furthermore, the linkage control module automatically triggers different levels of emergency measures, including information dissemination, activation of purification devices, or linkage traffic control, based on the received warning level signals.

[0048] This ensures that early warning information can be translated into concrete actions in a timely and accurate manner, such as issuing alerts to managers and the public, automatically activating fog cannons or sprinkler systems to reduce dust, or coordinating with traffic signal systems to guide traffic flow to reduce mobile source emissions. This automated linkage greatly shortens the response time from risk detection to control measures, improving the efficiency and effectiveness of pollution emergency response.

[0049] The beneficial effects of this invention are as follows: This invention provides a method and system for monitoring and early warning of air pollutant concentrations, which has the following advantages:

[0050] 1. It can achieve comprehensive and accurate monitoring. Through the collaboration of multi-principle sensors and dynamic weight fusion, it overcomes the limitations of single sensors and realizes accurate and reliable monitoring of multiple pollutants in the full concentration range, significantly reducing monitoring blind spots.

[0051] 2. It can achieve early warning. By adopting the improved Prophet model, it effectively captures the complex spatiotemporal patterns of pollution changes. Combined with a robust classification and judgment algorithm, it can realize the transformation from "post-event alarm" to "pre-event warning", significantly improving the advance warning time and accuracy.

[0052] 3. It has reliable system robustness, and improves the system's anti-interference ability and fault tolerance in complex environments from multiple levels, from hardware (multi-channel, fiber optic network) to algorithms (dynamic fusion, Cauchy weighting), effectively improving data quality and system stability.

[0053] 4. It can achieve efficient closed-loop management and control, realizing a complete business closed loop from data perception, intelligent analysis, early warning decision-making to automatic linkage, which greatly improves the speed and accuracy of environmental pollution emergency response and control capabilities. Moreover, it is flexible and economical to expand. By adopting fiber optic multiplexing and wireless transmission technology, the system can be deployed flexibly and expanded conveniently, which is conducive to building a smart environmental monitoring Internet of Things with wide coverage and low cost.

[0054] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0055] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0056] Figure 1 A flowchart of a method and system for monitoring and early warning of air pollutant concentrations provided in an embodiment of the present invention;

[0057] Figure 2 The present invention provides a system architecture diagram of an atmospheric pollutant concentration monitoring and early warning method and system according to an embodiment of the present invention. Detailed Implementation

[0058] The following is in conjunction with the appendix Figure 1-2 The principles and features of the present invention are described below. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically in the following paragraphs by way of example with reference to the accompanying drawings. The advantages and features of the invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the invention.

[0059] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0061] like Figure 2As shown, the atmospheric pollutant concentration monitoring and early warning system in this embodiment is mainly deployed in an industrial park and surrounding sensitive areas.

[0062] Multi-source data acquisition module: Monitoring micro-stations are set up at key locations such as the park boundary, downwind of key enterprises, and residential areas. Each micro-station integrates multiple sensors: a laser infrared gas sensor (using TDLAS technology) for monitoring VOCs and CH4; an electrochemical sensor for monitoring SO2, NO2, and O3; and a meteorological instrument for measuring temperature, humidity, wind speed, and wind direction. In addition, a distributed temperature / strain sensor network based on fiber optic gratings is deployed on the main transmission channels to help identify abnormal heat sources or emissions. Data is transmitted back via 4G / 5G or environmental protection private network.

[0063] The analysis and early warning module, deployed in the Environmental Protection Bureau's data center, first performs quality control upon receiving data from its data fusion unit. For VOCs concentrations, in low concentration ranges (e.g., 0-10 ppm), the laser infrared sensor is assigned a weight of 0.8, and the electrochemical sensor 0.2; in high concentration ranges, the weights are dynamically adjusted. For SO2, the data is primarily based on the electrochemical sensor, with compensation based on temperature and humidity. The predictive analysis unit uses hourly fused data as input and runs the Prophet model. The model is preset with seasonal patterns such as weekday / weekend and daytime / nighttime, and can automatically detect concentration abrupt changes caused by accidental emissions, predicting the concentration trend over the next 672 hours. The early warning judgment unit inputs the predicted concentration, current concentration, wind speed, and wind direction into a Cauchy weighted LSSVM model trained with historical data. This model can effectively filter out abnormally high values ​​caused by temporary instrument malfunctions or localized dust, outputting classification results such as "normal," "concerned," "warning," and "alarm," which are then combined with set thresholds (e.g., 75 μg / m³ for PM2.5). 3 150μg / m 3 ), and finally determine the warning level;

[0064] Linkage control module: Receives warning level signals. If it is at the "attention" level, it sends a prompt message to the park management personnel through the platform. If it is at the "warning" level, it automatically starts the smart sprinkler system in the park to reduce dust. If it reaches the "alarm" level, in addition to strengthening the sprinkler system, it can also link with the traffic signal system and suggest diverting and controlling heavy trucks on the surrounding roads.

[0065] Remote monitoring platform: Displays all monitoring point data, fusion results, prediction curves and early warning status in real time in the form of electronic maps, trend curves, dashboards, etc. All data is stored in the database and supports the generation of pollution situation analysis reports and source tracing reports.

[0066] refer to Figure 1 As shown, the monitoring and early warning method flow of this embodiment is as follows:

[0067] S1: Each monitoring micro-station and fiber optic network continuously collects pollutant concentration and meteorological data and uploads them in real time;

[0068] S2: The data fusion unit cleans and calibrates multi-source data, and dynamically calculates weighted fusion values ​​based on pollutant type and current concentration range to obtain the most reliable minute-level / hour-level concentration data;

[0069] S3: The predictive analysis unit inputs the fused concentration time series (e.g., hourly data from the past week) into the Prophet model. The model automatically learns its changing patterns and predicts the hourly concentration values ​​and confidence intervals for a future period (e.g., the next 24 hours).

[0070] S4: The early warning judgment unit combines information such as predicted concentration and real-time wind speed (affecting diffusion), uses a robust classification model to determine the pollution level, and cross-validates it with a fixed threshold to finally issue a specific early warning level (such as "ozone light pollution warning").

[0071] S5: The linkage control module automatically executes preset plans based on the issued warning level, such as triggering spraying in specific areas or notifying relevant enterprises to verify emission reductions;

[0072] S6: All process data is synchronized to the remote monitoring platform to achieve visualized supervision and decision support, and generate historical archives.

[0073] Example:

[0074] S1, System Deployment and Hardware Installation;

[0075] Monitoring node deployment: Plan and install multiple atmospheric monitoring micro-stations in the target monitoring area (such as the boundary of industrial parks, upwind of sensitive residential areas, and along main traffic arteries). Each micro-station serves as a physical carrier for a multi-dimensional data acquisition module.

[0076] Sensor Integration and Installation: Infrared Spectroscopy Absorption Gas Sensor: Select a laser gas analyzer such as SenseAir's LaserMini series or Beijing Aerospace Yilian's GLA series. This sensor has a built-in distributed feedback (DFB) tunable semiconductor laser with wavelengths customized for the target gas (e.g., 1653nm for CH4 and 2004nm for CO2). During installation, ensure that the optical lenses are clean and that the detection optical path is not continuously obstructed.

[0077] Electrochemical gas sensors: For SO2, NO2, O3, etc., select sensor modules such as Alpha Sense B4 series or City Technology specific models. During installation, ensure that the air inlet is unobstructed and avoid high humidity condensation.

[0078] Meteorological parameter sensors: Integrated weather instruments, such as the Vaisala WXT530 or the WS series from LUFFT in Germany, are used to simultaneously measure temperature, humidity, wind speed, and wind direction;

[0079] Data Acquisition and Transmission Unit: Each micro-station is equipped with an industrial-grade data acquisition unit (such as Advantech ADAM series or Yourenlian USRG780), which is responsible for acquiring and initially packaging the analog / digital signals from all the above sensors, and uploading the data to the central server through a 4G / 5G DTU (data terminal unit) or an environmentally friendly dedicated fiber optic network.

[0080] Fiber optic sensor network deployment (optional enhancement options):

[0081] Lay single-mode communication optical cables along potential pollution transmission paths or perimeters;

[0082] Connect fiber optic grating (FBG) gas sensors (such as palladium-coated hydrogen sensors or polymer-coated VOCs sensors) or FBG temperature and humidity sensors in series at key locations.

[0083] At the equipment room end, a 1xNMEMS optical switch (such as the series of products of Beijing Lingyun Optoelectronics) is installed and connected to the network through a fiber optic grating demodulator (such as MOIsm125 or equivalent products). The system controls the optical switch to switch sequentially and cyclically to realize the spatial multiplexing acquisition of signals from each FBG sensor.

[0084] S2, Data aggregation and fusion processing;

[0085] Data reception: The analysis and early warning module located in the monitoring center (deployed on a high-performance server, such as the Dell PowerEdge series or an industrial control computer) continuously receives real-time data streams from various monitoring micro-stations and fiber optic demodulators;

[0086] Data preprocessing and validation:

[0087] Data cleaning: The program automatically removes illegal values ​​that are clearly outside the physical measurement range;

[0088] Main and backup channel verification: For the same pollutant (such as VOCs) at the same monitoring point, if data from both the infrared sensor (main channel) and the electrochemical sensor (backup channel) are present, the system will compare them in real time. If the difference between the two exceeds the set tolerance (such as 20%), a data suspicion flag will be triggered.

[0089] Execution of dynamic weighted fusion algorithm (core algorithm of data fusion unit):

[0090] Algorithm logic: For each pollutant, its fused concentration value Cfused is calculated using the following formula:

[0091]

[0092] Wherein, CIR and CEC are the readings of the infrared and electrochemical sensors, respectively, wIR and wEC are dynamic weights, and wIR+wEC=1;

[0093] Dynamic weight allocation strategy:

[0094] The system has a built-in historical accuracy lookup table that records the long-term average error or standard deviation of each sensor in different concentration ranges (such as 0.1 ppm, 110 ppm, >10 ppm).

[0095] During real-time fusion, the system determines the concentration range based on the preliminary judgment values ​​of the current CIR and CEC.

[0096] Example of weight setting: When monitoring VOCs, in the 0.5ppm range, set...

[0097] wIR=0.85, wEC=0.15; in the >5ppm range, wIR=0.60, wEC=0.40, when monitoring SO2, high weights (e.g., 0.9) are mainly assigned to it in the linear range of the electrochemical sensor.

[0098] Output: The final output is an optimized, highly reliable time series fused concentration data, which is stored in a real-time database;

[0099] S3, intelligent prediction of concentration trends;

[0100] Data preparation: The predictive analysis unit extracts hourly fused concentration data of specified pollutants (such as PM2.5 or O3) from the database over a period of time (such as 7 days) to form a time series {yt};

[0101] Improved Prophet model predictions:

[0102] Model Structure: The model is developed using the Facebook Prophet open-source framework and is broken down into the following parts:

[0103] ;

[0104] Trend term g(t): Employs a logistic growth model, rather than a nonlinear growth model; its formula is:

[0105]

[0106] Where C is the carrying capacity, k is the growth rate, and m is the offset parameter. The model will automatically detect abrupt changes and allow the growth rate to be adjusted at abrupt changes to capture the impact of abrupt changes in emission sources or sudden weather changes.

[0107] Seasonal term s(t): The expansion of the Daubechies wavelet basis function (such as Db4) is used to replace the standard Prophet Fourier series in order to better fit the non-stationary daily and weekly fluctuations. Wavelet transform can provide localization information in both the time and frequency domains, and is more suitable for handling seasonal morphological fluctuations caused by changes in activity patterns.

[0108] Prediction execution: Input the historical sequence {yt} into the configured improved Prophet model, call its predict method to generate concentration predictions for the next 24 to 72 hours. and its confidence interval;

[0109] S4, robust judgment of warning level;

[0110] Feature vector construction: The early warning judgment unit constructs a feature vector x for the current time, including:

[0111] 1. Predicted concentration for the next 13 hours ;

[0112] 2. Current real-time fusion concentration ;

[0113] 3. Instantaneous rate of change of concentration:

[0114]

[0115] (Calculated by differentiating recent data);

[0116] 4. Current wind speed, wind direction, and temperature;

[0117] Cauchy-weighted LSSVM classification:

[0118] Algorithm core: The optimization problem of the standard LSSVM is:

[0119]

[0120] Where ei is the error and γ is the regularization parameter, Cauchy weighted improvement: to enhance robustness, the error term is changed to

[0121]

[0122] The weight vi is calculated by the Cauchy function:

[0123]

[0124] Where c is the scale parameter. When the error ei of a certain sample is too large (it may be an outlier), its weight vi will be automatically reduced, thereby reducing the impact of the outlier sample on the model decision surface.

[0125] As an example, the scale parameter c can be set according to the standard deviation of the historical error data distribution, for example, c=1.5σ, where σ is the standard deviation of the sample error;

[0126] Model application: A classification model trained using historical normal and warning event data is used to calculate the current feature vector x, and the model outputs a classification label (e.g., 0 Normal, 1 Attention, 2 Warning, 3 Alarm).

[0127] Threshold cross-validation and final classification: The classification results are cross-validated with the following preset thresholds to determine the final warning level:

[0128] Level 1 Warning (Attention): Forecast concentration for the next hour > National Level 2 Standard Limit (e.g., PM2.5 24-hour average is 75 μg / m³). 3 80% (i.e., >60 μg / m) 3 Furthermore, the classification model supports "attention" levels;

[0129] Level 2 Alarm (Warning): Predicted concentration within the next hour ≥ National Level 2 Standard Limit (≥75 μg / m³) 3 (or the classification model determines it as a "warning");

[0130] Emergency control alarm (critical alarm): Predicted concentration ≥ preset emergency threshold (e.g., 150 μg / m³) within the next 1 hour 3 (or the classification model determines it as an "alarm", and the wind speed conditions are conducive to the accumulation of pollutants;)

[0131] S5, automatic execution of linkage control;

[0132] Command issuance: The linkage control module (which can be implemented by Siemens S71200 PLC or Delta DVP series PLC) receives the final warning level signal from the analysis and warning module in real time via OPCUA or ModbusTCP protocol;

[0133] Execution mechanism trigger:

[0134] Level 1 warning: The PLC triggers the information publishing system (such as the connected LED display controller or SMS gateway) via serial port or network to publish a reminder message;

[0135] Level 2 alarm: The PLC automatically starts the emergency purification device, such as controlling the relay to open the high-pressure fog cannon (model such as Zhengzhou Weilan Environmental Protection's WP series) or the solenoid valve of the tower crane sprinkler system;

[0136] Emergency control alarm: While activating the purification device, the PLC sends suggested instructions to the intelligent traffic signal control system (such as the traffic signal controller of Hisense Network Technology) through the industrial communication gateway to trigger the "green wave" or adjust the signal timing at the intersection, guide traffic flow, and reduce mobile source emissions;

[0137] S6, panoramic monitoring and data management;

[0138] Data Upload: The analysis and early warning module pushes all raw data, fused data, prediction curves, early warning logs, and control feedback status to the remote monitoring platform in real time via the enterprise intranet or VPN;

[0139] Platform Demonstration and Analysis: The platform software (which can be developed using KingView, ForceControl, or a web-based Vue / React framework) implements the following functions:

[0140] Visualized large screen: Displays real-time concentration and warning level of each location in the form of a GIS map;

[0141] Trend analysis: Plot a curve comparing historical and predicted concentrations at any given location;

[0142] Alarm management: Record, query, and confirm all warning events;

[0143] Report generation: Automatically generates daily, monthly, and yearly monitoring reports and statistical analysis reports;

[0144] Data storage: All data is stored in a time-series database (such as InfluxDB) or a relational database (such as PostgreSQL) for long-term preservation for tracing and analysis;

[0145] By cyclically executing the above six steps, this invention can achieve a complete closed-loop business process from accurate perception, intelligent analysis, early warning to automatic linkage.

[0146] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Content not described in detail in this specification is prior art known to those skilled in the art.

[0147] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Those skilled in the art can readily implement the present invention based on the accompanying drawings and the above description. However, any modifications, alterations, or variations made by those skilled in the art without departing from the scope of the present invention, utilizing the disclosed technical content, are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A method for monitoring and early warning of air pollutant concentrations, characterized in that, Includes the following steps: S1, through multiple monitoring nodes deployed in the monitoring area, simultaneously collects concentration data of target air pollutants and environmental meteorological data; The monitoring node integrates sensors based on different measurement principles, including at least an infrared spectral absorption gas sensor, an electrochemical gas sensor, a temperature and humidity sensor, and a wind speed and direction sensor; the collected data is transmitted to the central processing unit through a communication network; S2, the central processing unit receives data from different monitoring nodes and sensors based on different principles; it adopts a dual-channel data verification mechanism with main and backup, and dynamically assigns differentiated fusion weights to sensors based on different principles based on the historical measurement accuracy data of various types of sensors for different pollutants in different concentration ranges, and calculates the optimized pollutant concentration fusion value. S3. Based on the fused historical and real-time concentration time series data obtained in step S2, the improved Prophet prediction model is used to predict the trend of pollutant concentration in the future period. The improved Prophet model uses a logistic growth trend term combined with a mutation point adjustment function, and employs wavelet basis functions to fit non-stationary seasonal fluctuations. S4. The concentration prediction results, real-time concentration change rate and environmental meteorological data obtained in step S3 are combined and input into the trained Cauchy function weighted least squares support vector machine classification model for auxiliary judgment. Finally, the warning level is determined based on the model output and the preset multi-level concentration thresholds; S5. Based on the warning level determined in step S4, generate and issue control commands to trigger the corresponding emergency response equipment or control system to execute preset emergency measures. S6. Upload the raw data, processing results, early warning decisions and control execution status from steps S1 to S5 to the remote monitoring platform in real time for visualization, comprehensive analysis and storage recording.

2. The method for monitoring and early warning of air pollutant concentrations according to claim 1, characterized in that, The dynamic weight allocation in step S2 specifically involves: assigning weights to sensors based on different principles according to the pollutant type and the current concentration range; where, when the target pollutant is in a low concentration range, the infrared spectral absorption gas sensor is assigned a higher fusion weight than the electrochemical gas sensor; and when the target pollutant is in its electrochemical response characteristic concentration range, the electrochemical gas sensor is assigned a higher fusion weight.

3. The method for monitoring and early warning of air pollutant concentrations according to claim 1, characterized in that, The preset multi-level concentration thresholds and warning levels mentioned in step S4 include: Level 1 warning: triggered when the predicted concentration reaches a set percentage of the ambient air quality standard limit; Level 2 alarm: Triggered when the predicted concentration reaches or exceeds the ambient air quality standard limit; Emergency control alarm: Triggered when the predicted concentration reaches the preset emergency control threshold, automatically initiating strong intervention measures.

4. An air pollutant concentration monitoring and early warning system for implementing the air pollutant concentration monitoring and early warning method according to any one of claims 1-3, comprising a multi-source data acquisition module, an analysis and early warning module, a linkage control module, and a remote monitoring platform, characterized in that: The multi-source data acquisition module includes an infrared spectral absorption gas sensor, an electrochemical gas sensor, a temperature and humidity sensor, and a wind speed and direction sensor. The multi-source data acquisition module is distributed and deployed in the monitoring area to collect pollutant concentration and environmental data. The analysis and early warning module includes a data fusion unit, a predictive analysis unit, and an early warning judgment unit. The data fusion unit is connected to the multi-source data acquisition module through a communication network and is used to receive and process data. The data fusion unit executes a primary and backup dual-channel verification and dynamic weighted fusion algorithm to output a fusion concentration value. The prediction and analysis unit runs an improved Prophet prediction model, receives the fusion concentration value, and outputs a concentration prediction trend. The early warning judgment unit runs a Cauchy function weighted least squares support vector machine model, receives prediction results and related data, and outputs an early warning level signal. The linkage control module is connected to the analysis and early warning module, receives early warning level signals, and automatically controls the actuator to respond. The remote monitoring platform is connected to the analysis and early warning module for data visualization, analysis, management, and storage.

5. The air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The infrared spectral absorption gas sensor uses a tunable semiconductor laser as a light source and utilizes the characteristic absorption lines of the target gas in the infrared band, combined with wavelength modulation and second harmonic detection technology for measurement.

6. The air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The monitoring and early warning system also includes an optical fiber sensor network, which uses spatial division multiplexing technology to connect multiple fiber optic grating sensors to a single trunk optical fiber and uses optical switches to switch and collect signals from each channel.

7. The air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The data fusion unit dynamically assigns fusion weights to the infrared spectral absorption gas sensor and the electrochemical gas sensor based on the pollutant type and the current concentration range.

8. The air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The early warning judgment unit is configured to run a Cauchy function-weighted least squares support vector machine classification model. The model reduces the impact of abnormal samples on classification decisions by introducing a Cauchy weight function based on prediction error into its optimization objective function for the training samples.

9. The air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The Cauchy function-weighted least squares support vector machine model configured in the early warning judgment unit reduces the impact of abnormal samples by introducing an error-based Cauchy weight function for each sample in the optimization objective.

10. An air pollutant concentration monitoring and early warning system according to claim 4, characterized in that, The linkage control module automatically triggers different levels of emergency measures, including information dissemination, activation of purification devices, or linkage traffic control, based on the different warning level signals received.

Images