A real-time monitoring system for air pollutants based on a gas sensor network

Abstract

This invention relates to the field of environmental monitoring and gas sensing technology, specifically to a real-time air pollutant monitoring system based on a gas sensor network. The system includes: a sensor front-end, a synchronous acquisition module, a data calibration module, a data processing unit, and a risk index assessment and feedback module. It collects the same airflow through a series-parallel gas path structure sharing a common intake manifold, calculates the airflow transmission delay based on the physical length of the gas path and the airflow velocity, and collaboratively triggers the sampling of particulate matter and gaseous pollutants, aligning the timestamps and outputting a synchronous original sampling sequence. This invention improves the accuracy of monitoring results for the same period and the same air mass.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and gas sensing technology, specifically to a real-time air pollutant monitoring system based on a gas sensor network. Background Technology

[0002] Currently, air pollutant monitoring systems based on gas sensor networks typically collect data separately using particulate matter sensors and gaseous pollutant sensors. When the intake paths and sampling times corresponding to different sensors are inconsistent, it can easily lead to different detection results corresponding to air samples from different time periods, thereby reducing the accuracy of pollutant concentration analysis and air quality assessment. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a real-time air pollutant monitoring system based on a gas sensor network. Specifically, the technical solution of this invention includes: The sensor front end is used to obtain the airflow in the target space through the series and parallel air path structure of the common intake manifold, obtain the airflow velocity through the flow velocity sensor set in the intake manifold, and output the particulate matter scattered light intensity signal, gaseous pollutant electrical signal, and environmental background parameters including measured temperature, measured humidity, measured wind speed and measured atmospheric pressure. The synchronous acquisition module is used to calculate the airflow transmission delay based on the system's preset physical length of the air path and the acquired airflow velocity, perform timing coordinated triggering on the sensor front end, and output a synchronous raw sampling sequence with sampling timestamps aligned. The data calibration module is used to receive the synchronous raw sampling sequence, convert the particulate matter scattered light intensity signal into primary particulate matter mass concentration, convert the gaseous pollutant electrical signal into volume concentration, and perform dimensional uniform conversion and water vapor condensation attenuation compensation based on environmental background parameters, and convert the volume concentration into primary gaseous mass concentration under standard operating conditions. The data processing unit is used to sequentially perform transient noise filtering and cross-interference correction on the primary gaseous mass concentration and the primary particulate matter mass concentration, and output the corrected final pollutant concentration. The risk index assessment and feedback module is used to calculate the comprehensive air quality index based on the final pollutant concentration and environmental background parameters, and to adjust the basic sampling cycle of the synchronous acquisition module based on the continuous state feedback of the comprehensive air quality index.

[0004] Preferably, the sensor front end includes an intake manifold, an airflow splitter module, and a sensor module; The airflow splitting module includes a main airflow and multiple parallel branch airflows. The outlet end of the intake manifold is connected to the main airflow, and the downstream end of the main airflow is connected to multiple parallel branch airflows. The sensor module includes a laser scattering particulate matter sensing unit connected in series on the main air duct, a gaseous pollutant sensing unit distributed in each branch air duct, and a capacitive temperature and humidity sensing unit, a wind speed sensing unit, and a pressure sensing unit located at the inlet end of the main air intake pipe. At the beginning of each basic sampling cycle, the synchronous acquisition module first wakes up the capacitive temperature and humidity sensing unit to collect environmental background parameters, and at the first sampling moment, it activates the laser scattering particulate matter sensing unit to generate particulate matter sampling timestamps. Then, at the second sampling moment after the airflow transmission delay, it triggers the gaseous pollutant sensing unit to generate gaseous sampling timestamps.

[0005] Preferably, the synchronous acquisition module constrains the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp through the following logic: The airflow transmission delay is derived based on the physical length of the air path and the airflow velocity in the intake manifold, and is used as the time difference between the first sampling time and the second sampling time. Among them, the physical length of the gas path is the length of the pipeline centerline from the detection cavity center of the laser scattering particulate matter sensing unit on the main airway to the surface of the gaseous pollutant sensing unit in the branch airway. The synchronous acquisition module compensates for airflow transmission delay, ensuring that the absolute value of the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp is less than a set time offset threshold.

[0006] Preferably, the data calibration module extracts the volume concentration obtained from the conversion of the gaseous pollutant electrical signal, as well as the mass concentration of primary particulate matter; The data calibration module uses measured temperature, measured humidity, and measured atmospheric pressure from the environmental background parameters to convert volumetric concentration into primary gaseous mass concentration. The conversion logic is as follows: Based on the molar mass and standard constant of the corresponding gaseous pollutant, combined with the absolute temperature converted from the measured temperature, the measured atmospheric pressure, and the humidity compensation coefficient, the volume concentration is dimensionally unified and compensated.

[0007] Preferably, the data calibration module constrains the humidity compensation coefficient using the following segmented conditions: When the measured humidity is less than or equal to the preset first humidity threshold, it is determined that there is no water vapor condensation effect, and the humidity compensation coefficient is set to one. When the measured humidity exceeds the first humidity threshold, the humidity compensation coefficient is set to a compensation value determined based on the preset condensation attenuation coefficient and the current humidity exceedance difference. The current humidity exceedance difference is the difference between the measured humidity and the first humidity threshold.

[0008] Preferably, the data processing unit includes a primary filter component and a fine calibration component connected in series. The primary filtration component independently applies a univariate recursive filter to the converted primary gaseous mass concentration and primary particulate matter mass concentration. By setting a state transition matrix and an observation matrix that characterize the constraints of the concentration sequence changes in a continuous period, the component uses the estimated value from the previous time step and the observed value from the current time step to perform smoothing, filter out transient noise caused by airflow disturbances, and outputs the primary filtration concentration sequence to the fine calibration component.

[0009] Preferably, the fine calibration component constructs a multi-layer calibration model to perform correlation calibration on the initial filtration concentration sequence; The input layer nodes of the multi-layer calibration model receive the initial filtration concentration sequence, measured temperature, measured humidity and measured wind speed, extract cross-interference information through the intermediate layer nodes, and output the final pollutant concentration after calibration by the output layer nodes. The fine calibration component has a built-in calibration interface, which is used to calculate the mean square error between the output value of the multi-layer calibration model and the standard concentration value when receiving calibration mode instructions and standard concentration gas data. The parameter weights of the multi-layer calibration model are adjusted by using the error backpropagation correction method until the mean square error is reduced to within the set threshold. In daily monitoring mode, the parameter weights are locked.

[0010] Preferably, the risk index assessment and feedback module calculates the individual air quality sub-index for each pollutant based on the final pollutant concentration, and extracts the maximum value among the individual air quality sub-indexes as the base air quality index; The risk index assessment and feedback module calculates the comprehensive air quality index. The calculation logic of this index is as follows: based on the wind speed correction factor and the temperature and humidity correction factor, the basic air quality index is weighted and corrected to obtain the comprehensive air quality index. Among them, the wind speed correction factor is used to characterize the weakening effect of increased wind speed on local pollutant retention. The wind speed correction factor is determined based on the preset wind speed coefficient, the measured wind speed and the preset wind speed compensation constant. The temperature and humidity correction factor is equal to the preset temperature and humidity coefficient when the measured temperature is greater than the preset high temperature threshold and the measured humidity is greater than the preset high humidity threshold; otherwise, the temperature and humidity correction factor is equal to zero.

[0011] Preferably, the synchronous acquisition module also has a built-in data anomaly detection mechanism: After acquiring the current sampled value of any sensing unit in the sensor module in the current sampling period, calculate the change between the current sampled value and the sampled value in the previous period; If the change is greater than the preset change threshold, the current sampled value is marked as suspicious data and cached and intercepted; otherwise, the current sampled value is marked as normal data and the continuous count of suspicious data in the current sensing unit is cleared. If the sampled values ​​for three consecutive sampling periods are marked as suspicious data, the current output of the corresponding sensing unit is determined to be abnormal, an abnormal alarm is triggered, and data interpolation compensation is performed on the data during the abnormal period using a polynomial interpolation method based on the historical normal data of the sensing unit.

[0012] Preferably, the risk index assessment and feedback module performs feedback control to switch to a low-power intermittent operating mode through the following logic: The comprehensive air quality index generated in real time is monitored. When the comprehensive air quality index value is less than the preset good or excellent level threshold for a number of consecutive basic sampling cycles, the basic sampling cycle is extended to the dormant sampling cycle. During the dormant sampling period, the output linkage command cuts off the power supply circuits of the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit, leaving only the capacitive temperature and humidity sensing unit to maintain environmental background monitoring; when the change of any environmental background parameter within the preset time window exceeds the preset environmental background parameter change threshold, the sampling period is immediately restored.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention, by setting up a series-parallel air path structure consisting of a shared intake manifold, a main air duct, and multiple parallel branch air ducts, ensures that the particulate matter sensing unit and the gaseous pollutant sensing unit complete sampling around the same airflow. In conjunction with the flow velocity sensor to obtain the airflow velocity and the airflow transmission delay calculated by combining the preset air path physical length, it achieves time-series compensation and alignment between the particulate matter sampling timestamp and the gaseous sampling timestamp. This effectively avoids the problem that different intake paths and different sampling times lead to different detection results corresponding to air samples in different time periods, thereby improving the accuracy of pollutant concentration analysis and air quality assessment. 2. This invention introduces temperature, humidity, wind speed and atmospheric pressure as environmental background parameters simultaneously. Through the conversion of volume concentration to standard operating condition mass concentration, compensation for water vapor condensation attenuation, filtering of transient noise and correction of cross-interference of multiple pollutants, the stability and reliability of monitoring results in complex plant boundary environments are improved. 3. This invention introduces a comprehensive air quality index feedback adjustment, sampling anomaly detection and interpolation compensation, as well as a low-power intermittent working mechanism, which not only ensures the availability of continuous monitoring data, but also takes into account the energy consumption control and device lifespan during long-term operation of the terminal. Attached Figure Description

[0014] Figure 1 This is a system block diagram of a real-time air pollutant monitoring system based on a gas sensor network, as described in an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0016] Example 1: Please see Figure 1 A real-time air pollutant monitoring system based on a gas sensor network includes: The sensor front end is used to obtain the airflow in the target space through the series and parallel air path structure of the common intake manifold, obtain the airflow velocity through the flow velocity sensor set in the intake manifold, and output the particulate matter scattered light intensity signal, gaseous pollutant electrical signal, and environmental background parameters including measured temperature, measured humidity, measured wind speed and measured atmospheric pressure. The synchronous acquisition module is used to calculate the airflow transmission delay based on the system's preset physical length of the air path and the acquired airflow velocity, perform timing coordinated triggering on the sensor front end, and output a synchronous raw sampling sequence with sampling timestamps aligned. The data calibration module is used to receive the synchronous raw sampling sequence, convert the particulate matter scattered light intensity signal into primary particulate matter mass concentration, convert the gaseous pollutant electrical signal into volume concentration, and perform dimensional uniform conversion and water vapor condensation attenuation compensation based on environmental background parameters, and convert the volume concentration into primary gaseous mass concentration under standard operating conditions. The data processing unit is used to sequentially perform transient noise filtering and cross-interference correction on the primary gaseous mass concentration and the primary particulate matter mass concentration, and output the corrected final pollutant concentration. The risk index assessment and feedback module is used to calculate the comprehensive air quality index based on the final pollutant concentration and environmental background parameters, and to adjust the basic sampling cycle of the synchronous acquisition module based on the continuous state feedback of the comprehensive air quality index. The sensor front end includes the main intake pipe, the air path splitter module, and the sensor module; The airflow splitting module includes a main airflow and multiple parallel branch airflows. The outlet end of the intake manifold is connected to the main airflow, and the downstream end of the main airflow is connected to multiple parallel branch airflows. The sensor module includes a laser scattering particulate matter sensing unit connected in series on the main air duct, a gaseous pollutant sensing unit distributed in each branch air duct, and a capacitive temperature and humidity sensing unit, a wind speed sensing unit, and a pressure sensing unit located at the inlet end of the main air intake pipe. At the beginning of each basic sampling cycle, the synchronous acquisition module first wakes up the capacitive temperature and humidity sensing unit to collect environmental background parameters, and at the first sampling moment, it starts the laser scattering particulate matter sensing unit to generate particulate matter sampling timestamps. At the second sampling moment after the airflow transmission delay, it triggers the gaseous pollutant sensing unit to generate gaseous sampling timestamps. The synchronous acquisition module constrains the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp through the following logic: The airflow transmission delay is derived based on the physical length of the air path and the airflow velocity in the intake manifold, and is used as the time difference between the first sampling time and the second sampling time. Among them, the physical length of the gas path is the length of the pipeline centerline from the detection cavity center of the laser scattering particulate matter sensing unit on the main airway to the surface of the gaseous pollutant sensing unit in the branch airway. The synchronous acquisition module compensates for airflow transmission delay, ensuring that the absolute value of the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp is less than a set time offset threshold. In practical application scenarios: the system is deployed at the fugitive emission monitoring point at the boundary of the industrial park to continuously monitor particulate matter and gaseous pollutants in the air near the boundary of the plant, and to conduct real-time assessment of air quality in combination with on-site temperature, humidity, wind speed and atmospheric pressure. In this scenario, outside air enters the monitoring terminal through the main intake pipe and flows through the main air duct and branch air ducts. Various sensing units complete sampling around the same airflow to avoid time misalignment caused by different air masses being detected separately under independent intake structures. This implementation method is mainly used to illustrate the overall working process of the system and the relationship between the air path structure and synchronous acquisition. Its basic design idea is to arrange the particulate matter detection path and the gaseous pollutant detection path on the same air intake link, and then arrange the triggering order of each sensing unit according to the transmission time of the airflow in the air path, so that the particulate matter scattered light intensity signal and the gaseous pollutant electrical signal correspond to the air sample entering the system at the same time. At the beginning of each basic sampling cycle, the synchronous acquisition module first calls the capacitive temperature and humidity sensing unit, wind speed sensing unit, and air pressure sensing unit to obtain the current environmental background parameters, and writes the set of parameters and the current cycle identifier into the sampling buffer; the laser scattering particulate matter sensing unit is activated at the first sampling moment to form a particulate matter sampling timestamp. The gaseous pollutant sensing unit does not sample immediately. Instead, it waits for the airflow to travel along the main and branch airways to the corresponding detection location before triggering the sampling at the second sampling time, thus forming a gaseous sampling timestamp. The particulate matter scattered light intensity signal, gaseous pollutant electrical signal, and environmental background parameters obtained in this way are combined into a synchronous raw sampling sequence within the same period, which can be directly read by the subsequent data calibration module. The configuration of the intake manifold, main air duct, and multiple parallel branch air ducts corresponds to the structure of the embodiment. A laser scattering particulate matter sensing unit is connected in series on the main air duct, which is mainly used to complete particulate matter detection before the airflow is split. Gaseous pollutant sensing units are arranged in each branch air duct so that the electrical signals of different gaseous pollutants can be collected independently in their respective channels. The capacitive temperature and humidity sensing unit is located at the inlet end of the main air intake pipe, which facilitates obtaining the current environmental conditions before pollutant sampling; the wind speed sensing unit and the air pressure sensing unit also output parameters at the beginning of the basic sampling cycle, so that they can participate in time compensation and dimension conversion in a unified manner later. For time offset control, in this embodiment, the synchronous acquisition module pre-stores the physical length of the air path from the center of the particulate matter detection chamber to the surface of the gaseous pollutant sensing unit; the flow rate sensor is installed in the main air intake pipe to provide the airflow velocity in the current sampling period; the synchronous acquisition module divides the physical length of the air path by the current airflow velocity to obtain the airflow transmission delay of the current period, and then uses the delay to correct the second sampling time. Therefore, the offset between the particulate matter sampling timestamp and the gaseous sampling timestamp is limited to within a set time offset threshold; the set time offset threshold is determined based on the inherent response time of the gaseous pollutant sensing unit, and the specific value range is 50 milliseconds to 200 milliseconds. For industrial park factory boundary scenarios, if the start-up and shutdown of the factory's fans causes a short-term change in the intake air velocity, the delay value is also recalculated according to the current flow velocity, instead of using the result of the previous cycle, thereby avoiding different air masses corresponding to particulate matter and gaseous pollutants in the same cycle. It should be noted that, in order to ensure that the airflow velocity in the intake manifold can be directly used to calculate the airflow transmission delay of the entire section including the split structure, and to ensure the correctness of the fluid dynamics principle, the air path split module adopts the equivalent cross-sectional area design rule; specifically, the effective cross-sectional area of ​​the main air passage is equal to the cross-sectional area of ​​the intake manifold, and the sum of the cross-sectional areas of multiple parallel branch air passages is also strictly equal to the cross-sectional area of ​​the main air passage. Based on the principles of mass conservation and continuity of incompressible fluids, under the condition that the effective cross-sectional area of ​​the entire air path is equivalent everywhere, the transmission linear velocity of the airflow in the intake manifold, main air passage and each branch air passage can remain constant; therefore, it is physically feasible and accurate to directly calculate the total transmission delay of the entire series-parallel air path by using the measured airflow velocity of the intake manifold as the denominator. In terms of data flow, the synchronous acquisition module does not directly participate in concentration conversion, but is only responsible for generating the time-aligned original sequence. In this original sequence, the particulate matter scattered light intensity signal retains the original optical output value, the gaseous pollutant electrical signal retains the original electrical response value, and the environmental background parameters retain the measured value of the current period and the corresponding time marker. The data calibration module extracts the above data from the cache periodically, and the subsequent conversion and compensation processes are based on the aligned data. Therefore, even if there are multiple gaseous pollutant sensing units in the same monitoring terminal, the sampling results of each branch airway can still be processed uniformly with the particulate matter detection results in the main airway in the same period, without having to re-trace the sensor triggering sequence. To ensure the continuity of the sampling process, if the flow velocity sensor fails to return a valid flow velocity within a certain basic sampling period, the synchronous acquisition module can temporarily call the flow velocity record of the previous valid period to complete the current period delay calculation, and add a replacement mark to the synchronous original sampling sequence of that period for subsequent modules to identify. If any of the environmental parameters, such as temperature, humidity, wind speed, or air pressure, fails to be collected normally in the current cycle, the synchronous acquisition module will still retain the sampling action for particulate matter and gaseous pollutants, but will mark the missing parameter location as empty, and hand it over to the subsequent data calibration module to decide whether to use the parameter from the previous valid cycle to replace it according to its processing rules. The upper limit of the sampling wait allowed by the system is set according to the length of the basic sampling period, and the specific value is 80% of the basic sampling period. If the airflow velocity is lower than the preset velocity threshold, causing the calculated airflow transmission delay to exceed the upper limit of the sampling wait allowed by the system, the gaseous pollutant sampling in this period will no longer be forcibly matched with the particulate matter sampling, and the current period will be recorded as an abnormal period in which synchronization is not completed, so as to prevent the continued output of pseudo-synchronization data in the case of mismatch where the airflow transmission delay exceeds the upper limit of the sampling wait allowed by the system. When continuously monitoring at the boundary of an industrial park, the above structure and control method are suitable for situations where the air outside the boundary is subject to short-term disturbances from emission plumes. Particulate matter and gaseous pollutants often change simultaneously as the emission plume passes by. If independent air paths and separate timed sampling are used, the system may analyze air samples from two different time periods in parallel. With this implementation method, the detection target is limited to the same airflow entering through the main intake pipe, and subsequent concentration conversion, noise processing and risk assessment are all based on the premise of the same cycle and the same air mass.

[0017] Example 2: The data calibration module extracts the volume concentration obtained from the conversion of the electrical signals of gaseous pollutants, as well as the mass concentration of primary particulate matter. The data calibration module uses the measured temperature, measured humidity, and measured atmospheric pressure from the environmental background parameters to convert the volume concentration into the primary gaseous mass concentration. The conversion logic is as follows: based on the molar mass and standard constant of the corresponding gaseous pollutant, combined with the absolute temperature converted from the measured temperature, the measured atmospheric pressure, and the humidity compensation coefficient, the volume concentration is dimensionally unified and compensated. The data calibration module constrains the humidity compensation coefficient under the following segmented conditions: when the measured humidity is less than or equal to the preset first humidity threshold, it is determined that there is no water vapor condensation effect, and the humidity compensation coefficient is set to one; when the measured humidity is greater than the first humidity threshold, the humidity compensation coefficient is set to a compensation value determined based on the preset condensation attenuation coefficient and the current humidity excess difference, where the current humidity excess difference is the difference between the measured humidity and the first humidity threshold. The data processing unit includes a pre-filter component and a fine-calibration component connected in series. The pre-filter component independently applies a univariate recursive filter to the converted primary gaseous mass concentration and primary particulate matter mass concentration. By setting a state transition matrix and an observation matrix that characterize the change constraints of the concentration sequence in a continuous period, it uses the estimated value of the previous time step and the observed value of the current time step to perform smoothing, filter out transient noise caused by airflow disturbance, and output the pre-filtered concentration sequence to the fine-calibration component. The fine calibration component constructs a multi-layer calibration model to perform correlation calibration on the initial filtration concentration sequence. The input layer node of the multi-layer calibration model receives the initial filtration concentration sequence, measured temperature, measured humidity and measured wind speed, extracts cross-interference information through the intermediate layer node, and outputs the final calibrated pollutant concentration by the output layer node. The fine calibration component has a built-in calibration interface, which is used to calculate the mean square error between the output value of the multi-layer calibration model and the standard concentration value when receiving calibration mode instructions and standard concentration gas data. The parameter weights of the multi-layer calibration model are adjusted by using the error backpropagation correction method until the mean square error is reduced to within the set threshold. In daily monitoring mode, the parameter weights are locked.

[0018] In practical application scenarios: During long-term operation, monitoring points at the boundaries of industrial parks are often simultaneously affected by temperature changes, increased humidity, local wind disturbances, and the coexistence of multiple gaseous pollutants. Obtaining only the synchronous original sampling sequence is not enough to directly obtain concentration results that can be used for evaluation. It is also necessary to process the deviations caused by inconsistencies in dimensions, changes in environmental conditions, and cross-interference. This implementation method is mainly used to illustrate how the data calibration module and data processing unit sequentially perform conversion, compensation, initial filtration and fine calibration after synchronous sampling is completed. The basic design idea is to first convert the electrical signal output by the gaseous pollutant sensing unit into volume concentration, and then convert it into primary gaseous mass concentration under standard operating conditions based on the measured temperature, measured humidity and measured atmospheric pressure. The particulate matter scattered light intensity signal is converted into primary particulate matter mass concentration. After the two types of primary concentrations enter the data processing unit, they are first smoothed by the primary filter component, and then enter the fine calibration component. The component combines temperature, humidity and wind speed to perform correlation correction on the cross-interference between multiple pollutants, and outputs the final pollutant concentration. For the dimensionless process, the data calibration module reads the volume concentration and primary particulate matter mass concentration of the current period from the synchronous original sampling sequence. Since the particulate matter portion already corresponds to a mass concentration form, it is directly used as the object for subsequent processing. For the gaseous pollutant portion, its molar mass is matched according to the current pollutant type, and it is converted according to the standard gas molar volume constant, standard temperature constant, and standard atmospheric pressure constant. During the conversion, the measured temperature is first converted to the current ambient absolute temperature, and then a temperature correction ratio is formed with the standard temperature constant. The measured atmospheric pressure is compared with the standard atmospheric pressure constant to form a pressure correction ratio; then multiplied by the humidity compensation coefficient; after the conversion, the primary gaseous mass concentration that can be processed in parallel with the particulate matter concentration is obtained. This result, along with the primary particulate matter mass concentration, is written into the processing buffer as the current cycle input of the primary filter component. Based solely on temperature and air pressure conversion, deviations may still occur in factory boundary environments where humidity exceeds the preset first humidity threshold. Therefore, a humidity compensation coefficient is introduced. The reason is that when the measured humidity is high, the breathable membrane on the surface of the gaseous pollutant sensing unit may be affected by water vapor condensation, which will change the diffusion process of the target gas into the sensor, causing the concentration value corresponding to the electrical signal to deviate from the normal state. To this end, the data calibration module first determines whether the measured humidity is higher than the preset first humidity threshold, where the preset first humidity threshold is 80% relative humidity and the preset condensation attenuation coefficient ranges from 0.005 to 0.015. If the humidity does not exceed the threshold, the humidity compensation coefficient is directly set to one, indicating that no humidity attenuation correction is performed in this cycle; if it exceeds the threshold, the difference between the measured humidity and the first humidity threshold is calculated, and then the difference is multiplied by the preset condensation attenuation coefficient and subtracted from one to form the humidity compensation coefficient for the current cycle; this compensation coefficient directly participates in the conversion of volume concentration to primary gaseous mass concentration and is no longer calculated repeatedly by subsequent modules. Based on the above processing, a primary filter component is introduced. If the converted primary gaseous mass concentration and primary particulate matter mass concentration are directly sent to the fine calibration component, the instantaneous flow fluctuations, wind pressure pulsations and short-term sampling disturbances in the gas path can easily cause high-frequency fluctuations in the concentration sequence of the current period, affecting the stability of subsequent cross-interference correction. Therefore, the data processing unit sets the primary filter component and the fine calibration component in series in sequence. The primary filtration component independently establishes a univariate recursive filtering process for each primary gaseous mass concentration and particulate matter mass concentration. In each cycle, it combines the estimated value from the previous moment with the observed value at the current moment to complete the smoothing process. The corresponding state transition matrix and observation matrix are preset to constrain the change relationship of the concentration sequence in continuous cycles. The smoothed initial filter concentration sequence does not replace the original calibration result, but is written as a separate data item into the next processing area for the fine calibration component to use; in the specific data flow and calculation rule decomposition, the univariate recursive filter is specifically implemented as a one-dimensional Kalman filter model; since the target parameter is a univariate concentration sequence, the state transition matrix and the observation matrix are both simplified to the scalar constant 1; The filtering iteration process is divided into two steps: prediction and update. The optimal concentration estimate from the previous time step is used as the prior concentration prediction for the current time step, and the prior error covariance is obtained by adding the system process noise variance to the estimation error covariance from the previous time step. The system process noise variance is set based on the inherent noise floor statistical variance of the sensing unit in an interference-free calibration environment, and the observation noise variance is set based on the concentration variance caused by airflow disturbance in the field test. By calculating the Kalman gain, which is the sum of the prior error covariance and the observation noise variance, the deviation between the current observation value and the prior prediction value is weighted and corrected, thereby obtaining the updated initial filtration concentration estimate at the current time. For example, if airflow disturbances cause a transient step change in the current observation value, the calculated Kalman gain will approach zero due to the large pre-set observation noise variance. The system will assign a higher confidence weight to the prior prediction value, thereby effectively reducing the high-frequency fluctuations of the transient. In this way, when it is necessary to trace back the original calibration value, the system can still retain the unfiltered data without affecting subsequent verification. The reason for setting this additional feature for the fine calibration component is that the cross-response problem between pollutants still exists even after removing transient noise using only a single-variable recursive filter. For example, near the boundary of an industrial park, multiple gaseous pollutants may change simultaneously, and the output of a certain gaseous pollutant sensing unit will be affected by another pollutant and environmental conditions. It is difficult to maintain consistency under different operating conditions by relying on a single threshold or fixed linear correction. Therefore, the fine calibration component uses a multi-layer calibration model to process the initial filtration concentration sequence. The input layer node receives the initial filtration concentration sequence of each pollutant, as well as the measured temperature, measured humidity, and measured wind speed; the intermediate layer node is used to extract the correlation between different pollutant readings and environmental conditions; the output layer node provides the corrected final pollutant concentration; this final pollutant concentration is written into the reading area of ​​the risk index assessment and feedback module as the direct basis for subsequent air quality index calculation. The fine calibration component has a built-in calibration interface to distinguish between calibration mode and daily monitoring mode. In calibration mode, the system receives standard concentration gas data and compares the output value of the multi-layer calibration model with the standard concentration value to form the mean square error. Then, the error is corrected by stepwise backpropagation to adjust the parameter weights. Specifically, the multi-layer calibration model is implemented as a feedforward neural network containing at least one hidden layer; the intermediate layer nodes, i.e. the hidden layer, use a sigmoid function as the activation function to capture the nonlinear cross-interference effect between various pollutants and temperature, humidity, and wind speed, while the output layer nodes use a linear function to directly output continuous concentration prediction values. During the error backpropagation correction process, the mean squared error is used as the global loss function. The preset learning rate ranges from 0.001 to 0.05. According to the chain rule, the partial derivative of the loss function with respect to the weights of the output layer nodes is calculated as the gradient, multiplied by the preset learning rate, and the gradient value is subtracted from the current output layer weights to complete the update. The error signal of the output layer is multiplied by the derivative of the activation function of the hidden layer and propagated back to the hidden layer. The partial derivative gradient of the hidden layer weights is calculated accordingly, and the hidden layer weights are updated in the same way. Through multiple rounds of forward calculation and backward gradient propagation iterations, the mean square error is reduced to within a set threshold. This set threshold is set to be less than or equal to five percent based on the allowable error range of the standard concentration gas. After the weight update is completed, the new parameter set is saved as the current valid calibration result. In routine monitoring mode, parameter weights are kept locked and do not change automatically during operation to avoid model parameters drifting with sampled values ​​when the field environment fluctuates. In this way, the calibration and monitoring processes are separated, and routine monitoring only calls the fixed parameter set without affecting the calibration results. To ensure uninterrupted processing, if the data calibration module detects missing values ​​in the measured temperature, humidity, or atmospheric pressure in the current cycle, it first checks whether complete environmental background parameters exist in the previous valid cycle. If they do, the volume concentration conversion for the current cycle is temporarily completed using these parameters, and a substitution mark is added to the resulting primary gaseous mass concentration. If no complete substitution value exists, the corresponding gaseous pollutant conversion result for the current cycle will not enter the primary filter component. For the humidity compensation coefficient, if the result is less than zero after segmented calculation, the value will not be pushed down for that period. Instead, it will be limited to an effective range of not less than zero before being used in the conversion, so as to avoid negative concentration correction ratios that have no physical meaning. For the primary filter component, if the estimated value of a certain pollutant is missing at the previous moment, the current observation value will be used as the starting value for the smoothing treatment of that pollutant. For the fine calibration component, if complete standard concentration gas data is not received in calibration mode, the parameter weights will not be updated in this calibration process, and the previous set of effective weights will continue to be maintained to prevent writing unstable parameters when the calibration data is incomplete. During continuous monitoring at the boundary of industrial parks, the above-mentioned processing link can continuously adapt to the diurnal temperature difference, high humidity after rain, and airflow disturbances caused by vehicles on the factory roads. The original sampling sequence is first uniformly converted in the data calibration module, then the short-term fluctuations are reduced by the primary filtration component, and the cross-influence between multiple pollutants is processed by the fine calibration component. After this sequential processing, the final pollutant concentration output can be directly used for displaying the continuous monitoring results at the factory boundary, storing historical records, and subsequent risk index assessment.

[0019] Example 3: The risk index assessment and feedback module calculates the individual air quality sub-index for each pollutant based on the final pollutant concentration, and extracts the maximum value among the individual air quality sub-indexes as the base air quality index. The risk index assessment and feedback module calculates the comprehensive air quality index. The calculation logic of this index is as follows: based on the wind speed correction factor and the temperature and humidity correction factor, the basic air quality index is weighted and corrected to obtain the comprehensive air quality index. Among them, the wind speed correction factor is used to characterize the weakening effect of increased wind speed on local pollutant retention. The wind speed correction factor is determined based on the preset wind speed coefficient, the measured wind speed and the preset wind speed compensation constant. The temperature and humidity correction factor is equal to the preset temperature and humidity coefficient when the measured temperature is greater than the preset high temperature threshold and the measured humidity is greater than the preset high humidity threshold. Otherwise, the temperature and humidity correction factor is zero; The synchronous acquisition module also has a built-in data anomaly detection mechanism: after acquiring the current sampled value of any sensing unit in the sensor module in the current sampling period, it calculates the change between the current sampled value and the sampled value in the previous period. If the change exceeds the preset change threshold, the current sampled value will be marked as suspicious data and cached for interception. Otherwise, mark the current sampled value as normal data and clear the continuous count of suspicious data for the current sensing unit; If the sampled values ​​of three consecutive sampling periods are marked as suspicious data, the current output of the corresponding sensing unit is determined to be abnormal, an abnormal alarm is triggered, and data interpolation compensation is performed on the data during the abnormal period using a polynomial interpolation method based on the historical normal data of the sensing unit. The risk index assessment and feedback module performs feedback control to switch to a low-power intermittent operating mode through the following logic: The comprehensive air quality index generated in real time is monitored. When the comprehensive air quality index value is less than the preset good or excellent level threshold for a number of consecutive basic sampling cycles, the basic sampling cycle is extended to the dormant sampling cycle. During the dormant sampling period, the output linkage command cuts off the power supply circuit of the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit, leaving only the capacitive temperature and humidity sensing unit to maintain environmental background monitoring. When the change of any environmental background parameter within a preset time window exceeds the preset threshold for the change of environmental background parameter, the sampling period is immediately restored to the basic sampling period.

[0020] In practical application scenarios: industrial park boundary monitoring points usually need to operate continuously for a long time, in order to respond to emission fluctuations, control terminal operating energy consumption, and improve the stability of long-term continuous operation. After the aforementioned final pollutant concentration is formed, the system also needs to provide an index result suitable for on-site use and arrange the subsequent sampling frequency according to the index status. At the same time, the plant boundary points are affected by changes in wind direction, rain and fog, dust accumulation and sensor aging. A single sensing unit may experience abnormal step-like changes during operation. Therefore, the sampling side also needs to supplement abnormal detection and compensation processing. This implementation method is mainly used to illustrate the coordination relationship between the comprehensive air quality index, the anomaly detection mechanism, and the low-power intermittent working mode. The basic design idea is that the risk index assessment and feedback module first generates a comprehensive air quality index based on the final pollutant concentration, and then adjusts the basic sampling cycle according to the continuous changes of the index; at the same time, the synchronous acquisition module periodically compares the sensor sampling values ​​at a more advanced stage to prevent abnormal values ​​from directly entering the subsequent index calculation. The risk index assessment and feedback module first calculates the corresponding individual air quality sub-index based on the final pollutant concentration of each pollutant, and then takes the maximum value from each individual air quality sub-index as the base air quality index. Using only this maximum value can reflect the level of major pollutants, but for industrial park boundary points, changes in on-site wind speed will significantly affect the degree of pollutant retention, and high temperature and humidity conditions will also change the exposure experience of personnel. Therefore, environmental corrections are added in addition to the base air quality index. The module reads the measured wind speed, measured temperature and measured humidity of the same period. First, it calculates the wind speed correction factor according to the preset wind speed coefficient and the preset wind speed compensation constant. Then, it determines whether the measured temperature is higher than the preset high temperature threshold and whether the measured humidity is higher than the preset high humidity threshold. If both are satisfied, the temperature and humidity correction factor is the preset temperature and humidity coefficient. If not satisfied, the temperature and humidity correction factor is zero. The preset wind speed coefficient ranges from 1.0 to 2.5, the preset wind speed compensation constant is 1.0 m / s; the preset high temperature threshold is 35 degrees Celsius, the preset high humidity threshold is 85% relative humidity, and the preset temperature and humidity coefficient ranges from 0.1 to 0.3. The module multiplies the base air quality index by a sum of wind speed correction factors, and then multiplies it by a sum of temperature and humidity correction factors to obtain the comprehensive air quality index for the current period. This index, along with the environmental background parameters used during generation and the final pollutant concentration, is written into the results recording area for display, storage, and retrieval by the sampling cycle feedback logic. If the comprehensive air quality index generated each cycle is used directly as the control basis without checking the front-end sampling values, sudden changes in values ​​caused by individual sensor units due to malfunctions may drive the index to rise or fall abnormally. Therefore, the synchronous acquisition module has a built-in data anomaly detection mechanism. Whenever any sensor unit outputs the current sampling value in the current sampling cycle, the system compares it with the sampling value of the previous cycle to obtain the change. Here, the change specifically refers to the absolute value of the difference between the current sampling value and the sampling value of the previous cycle. To ensure physical rationality, preset change thresholds are configured independently for different types of pollutants and environmental parameter characteristics. For example, particulate matter is easily carried by wind and its change threshold is set higher than that of temperature and relative humidity. However, since the natural change gradient of temperature and relative humidity is smaller than that of particulate matter, their change thresholds are set lower than that of particulate matter. These preset change thresholds are set based on three times the maximum single-cycle change rate of the corresponding sensing unit under historical normal operating conditions. If the calculated absolute change is greater than its corresponding preset change threshold, the current sampled value will not immediately enter the normal processing link, but will be marked as suspicious data and intercepted in the cache; if the change does not exceed the threshold, the sampled value will be marked as normal data, and the existing suspicious data count of the sensing unit will be cleared to zero; this can distinguish between occasional mutations and continuous anomalies, and avoid one-time fluctuations directly triggering fault judgment. When suspicious data appears in three consecutive sampling periods, the system determines that the current output of the corresponding sensor unit is abnormal and triggers an alarm. After the alarm, the data generated by the sensor unit during the abnormal period is not directly used for exponential calculation. Instead, the system calls the historical normal data of the sensor unit for polynomial interpolation compensation to maintain the continuity of the data sequence. The compensation result still retains the abnormal mark, and subsequent modules can identify that it is not a real-time measured value. The reason for this approach is that industrial park boundary points typically require continuous uploading of monitoring data. Even if a single sensor unit experiences a short-term anomaly and is not involved in the actual concentration evaluation, a complete timeline data record is still required. By using polynomial interpolation based on historical normal data, the system can maintain the availability of the sequence before the fault is resolved, without directly inputting significant outliers into subsequent concentration processing and index calculation. In terms of the breakdown of the underlying computational logic, the polynomial interpolation method based on historical normal data is specifically configured as second-order Newton interpolation. When it is determined that the output of a certain sensing unit is abnormal, the synchronous acquisition module will automatically extract the sampling timestamps and normal sampling values ​​of the three consecutive valid periods immediately preceding the abnormal interval, and use them as the independent and dependent variables of the interpolation node to calculate the first-order and second-order difference quotients, respectively. The obtained difference quotient is used to construct a quadratic polynomial function with respect to time, and the sampling timestamp of the current abnormal period is substituted into the function to calculate the compensation estimate. Through this second-order extrapolation mechanism that combines the slope and curvature changes of historical curves, the physical characteristics of the continuous and smooth change of air pollutant concentration over time can be effectively reflected, ensuring that the interpolation compensation data has a reasonable transition trend. If the basic sampling cycle is used continuously, and the air quality at the factory boundary is stable for a long time and the comprehensive air quality index is consistently below the preset excellent level threshold, the particulate matter sensing unit and the gaseous pollutant sensing unit will maintain unnecessary working time, increasing energy consumption and accelerating the aging process of the devices. Therefore, the risk index assessment and feedback module performs feedback control based on the comprehensive air quality index of multiple consecutive basic sampling cycles. When the comprehensive air quality index is lower than the preset good and excellent level threshold for a preset number of consecutive basic sampling cycles, the system extends the basic sampling cycle to the dormant sampling cycle. The preset number of consecutive cycles is an integer between 5 and 10, and the preset good and excellent level threshold is set to a comprehensive air quality index of 50. After entering the dormant sampling cycle, the module issues a linkage command to cut off the power supply circuit of the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit, leaving only the capacitive temperature and humidity sensing unit to maintain environmental background monitoring. If the change in any environmental parameter exceeds the preset environmental parameter change threshold within the preset time window, the dormant sampling cycle will be immediately terminated and the basic sampling cycle will be resumed in order to restart pollutant detection. The preset time window is set to 10 to 30 minutes, and the preset environmental parameter change threshold is set to a temperature change of more than 2 degrees Celsius or a relative humidity change of more than 10%. In this feedback process, the comprehensive air quality index itself does not directly control the sensor output, but first enters the periodic state record, and the feedback module makes a unified judgment on the index state for several consecutive periods; this can avoid frequent switching of the sampling period caused by slight fluctuations in a certain period. After entering hibernation, only the capacitive temperature and humidity sensing unit continues to work because its power consumption is lower than that of the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit. Moreover, environmental changes often precede pollutant concentration anomalies, so it can serve as a preliminary criterion for resuming sampling. After resuming the basic sampling cycle, the system re-executes synchronous acquisition, conversion, filtering, and fine calibration according to the process, and the new final pollutant concentration is then used to update the comprehensive air quality index. To ensure that the abnormal and low-power logic do not conflict with each other, if the system is in a sleep sampling cycle, the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit will no longer perform the comparison of changes based on the current sampled value because they have been powered off; the corresponding abnormal detection mechanism will only continue to be effective for environmental parameter sampling channels that are still in operation. If the system is executing an abnormal alarm and a certain sensor unit is in the interpolation compensation period, the risk index assessment and feedback module can still calculate the comprehensive air quality index based on the data and compensation values ​​of other normal sensor units, but will retain the abnormal source mark in the current cycle results; If any sensor unit malfunctions during the continuous good / good determination process, the continuous count of good / good status can be paused at the current value and will continue to accumulate after the malfunctioning channel returns to normal, so as to avoid directly switching to the sleep sampling period during the period of incomplete data. In actual operation at the industrial park boundary, the above solution reduces the working time of relevant components when the comprehensive air quality index is consistently below the preset good level threshold, and can promptly resume normal sampling when the boundary environment changes. At the same time, through front-end anomaly detection, buffer interception and interpolation compensation, it avoids abnormal sampling values ​​from directly affecting the comprehensive air quality index, thereby maintaining the continuity of long-term monitoring records and the availability of current index results.

[0021] It should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A real-time air pollutant monitoring system based on a gas sensor network, characterized in that, include: The sensor front end is used to obtain the airflow in the target space through the series and parallel air path structure of the common air intake manifold, obtain the airflow velocity through the flow velocity sensor set in the air intake manifold, and output the particulate matter scattered light intensity signal, gaseous pollutant electrical signal, and environmental background parameters including measured temperature, measured humidity, measured wind speed and measured atmospheric pressure. The synchronous acquisition module is used to calculate the airflow transmission delay based on the preset physical length of the air path and the acquired airflow velocity, perform timing coordinated triggering on the sensor front end, and output a synchronous raw sampling sequence with sampling timestamps aligned. The data calibration module is used to receive the synchronous raw sampling sequence, convert the particulate matter scattered light intensity signal into primary particulate matter mass concentration, convert the gaseous pollutant electrical signal into volume concentration, and perform dimensional uniform conversion and water vapor condensation attenuation compensation based on the environmental background parameters, and convert the volume concentration into primary gaseous mass concentration under standard operating conditions. The data processing unit is used to sequentially perform transient noise filtering and cross-interference correction on the primary gaseous mass concentration and the primary particulate matter mass concentration, and output the corrected final pollutant concentration. The risk index assessment and feedback module is used to calculate the comprehensive air quality index based on the final pollutant concentration and the environmental background parameters, and to adjust the basic sampling cycle of the synchronous acquisition module based on the continuous state feedback of the comprehensive air quality index.

2. The real-time air pollutant monitoring system based on a gas sensor network according to claim 1, characterized in that, The sensor front end includes an intake manifold, an air path splitter module, and a sensor module; The airflow splitting module includes a main airflow and multiple parallel branch airflows. The outlet end of the main intake pipe is connected to the main airflow, and the downstream end of the main airflow is connected to the multiple parallel branch airflows. The sensor module includes a laser scattering particulate matter sensing unit connected in series on the main air duct, a gaseous pollutant sensing unit distributed in each branch air duct, and a capacitive temperature and humidity sensing unit, a wind speed sensing unit, and a pressure sensing unit located at the inlet end of the main air intake pipe. The synchronous acquisition module first wakes up the capacitive temperature and humidity sensing unit at the beginning of each basic sampling cycle to collect environmental background parameters, and starts the laser scattering particulate matter sensing unit to generate particulate matter sampling timestamps at the first sampling time. Then, at the second sampling time after the airflow transmission delay, the gaseous pollutant sensing unit is triggered to generate gaseous sampling timestamps.

3. The real-time air pollutant monitoring system based on a gas sensor network according to claim 2, characterized in that, The synchronous acquisition module constrains the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp using the following logic: The airflow transmission delay is derived based on the physical length of the air path and the airflow velocity in the intake manifold, and is used as the time difference between the first sampling time and the second sampling time. The physical length of the gas path is the length of the pipeline centerline from the detection cavity center of the laser scattering particulate matter sensing unit on the main airway to the surface of the gaseous pollutant sensing unit in the branch airway. The synchronous acquisition module compensates for the airflow transmission delay, ensuring that the absolute value of the time offset between the particulate matter sampling timestamp and the gaseous sampling timestamp is less than a set time offset threshold.

4. The real-time air pollutant monitoring system based on a gas sensor network according to claim 3, characterized in that, The data calibration module extracts the volume concentration obtained from the conversion of the gaseous pollutant electrical signal, as well as the mass concentration of the primary particulate matter; The data calibration module uses the measured temperature, measured humidity, and measured atmospheric pressure from the environmental background parameters to convert the volume concentration into the primary gaseous mass concentration. The conversion logic is as follows: Based on the molar mass and standard constant of the corresponding gaseous pollutant, combined with the absolute temperature converted from the measured temperature, the measured atmospheric pressure, and the humidity compensation coefficient, the volume concentration is subjected to dimensional unification and compensation conversion.

5. The real-time air pollutant monitoring system based on a gas sensor network according to claim 4, characterized in that, The data calibration module constrains the humidity compensation coefficient using the following segmented conditions: When the measured humidity is less than or equal to the preset first humidity threshold, it is determined that there is no water vapor condensation effect, and the humidity compensation coefficient is set to one. When the measured humidity is greater than the first humidity threshold, the humidity compensation coefficient is set to a compensation value determined based on a preset condensation attenuation coefficient and the current humidity exceedance difference, where the current humidity exceedance difference is the difference between the measured humidity and the first humidity threshold.

6. The real-time air pollutant monitoring system based on a gas sensor network according to claim 5, characterized in that, The data processing unit includes a primary filter component and a fine calibration component connected in series. The initial filtration component independently applies a univariate recursive filter to the converted primary gaseous mass concentration and primary particulate matter mass concentration. By setting a state transition matrix and an observation matrix that characterize the constraints of the concentration sequence changes in a continuous period, the component uses the estimated value from the previous moment and the observed value from the current moment to perform smoothing, filtering out transient noise caused by airflow disturbances, and outputting the initial filtration concentration sequence to the fine-tuning component.

7. The real-time air pollutant monitoring system based on a gas sensor network according to claim 6, characterized in that, The fine-calibration component constructs a multi-layer calibration model to perform correlation calibration on the initial filtration concentration sequence; The input layer node of the multi-layer calibration model receives the initial filtration concentration sequence, measured temperature, measured humidity and measured wind speed, extracts cross-interference information through the intermediate layer node, and outputs the corrected final pollutant concentration through the output layer node. The fine calibration component has a built-in calibration interface, which is used to calculate the mean square error between the output value of the multi-layer calibration model and the standard concentration value when receiving calibration mode instructions and standard concentration gas data. The parameter weights of the multi-layer calibration model are adjusted by using an error backpropagation correction method until the mean square error is reduced to within a set threshold. In daily monitoring mode, the parameter weights are locked.

8. The real-time air pollutant monitoring system based on a gas sensor network according to claim 7, characterized in that, The risk index assessment and feedback module calculates the individual air quality sub-index for each pollutant based on the final pollutant concentration, and extracts the maximum value among the individual air quality sub-indexes as the base air quality index. The risk index assessment and feedback module calculates the comprehensive air quality index. The calculation logic of the index is as follows: the basic air quality index is weighted and corrected based on the wind speed correction factor and the temperature and humidity correction factor to calculate the comprehensive air quality index. The wind speed correction factor is used to characterize the weakening effect of increased wind speed on local pollutant retention. The wind speed correction factor is determined based on a preset wind speed coefficient, the measured wind speed, and a preset wind speed compensation constant. The temperature and humidity correction factor is equal to the preset temperature and humidity coefficient when the measured temperature is greater than the preset high temperature threshold and the measured humidity is greater than the preset high humidity threshold; otherwise, the temperature and humidity correction factor is equal to zero.

9. A real-time air pollutant monitoring system based on a gas sensor network according to claim 8, characterized in that, The synchronous acquisition module also has a built-in data anomaly detection mechanism: After acquiring the current sampled value of any sensing unit in the sensor module in the current sampling period, calculate the change between the current sampled value and the sampled value in the previous period; If the change is greater than a preset change threshold, the current sampled value is marked as suspicious data and cached for interception. Otherwise, mark the current sampled value as normal data and clear the continuous count of suspicious data in the current sensing unit; If the sampled values ​​for three consecutive sampling periods are marked as suspicious data, the current output of the corresponding sensing unit is determined to be abnormal, an abnormal alarm is triggered, and data interpolation compensation is performed on the data during the abnormal period using a polynomial interpolation method based on the historical normal data of the sensing unit.

10. A real-time air pollutant monitoring system based on a gas sensor network according to claim 9, characterized in that, The risk index assessment and feedback module performs feedback control to switch to a low-power intermittent operating mode through the following logic: The comprehensive air quality index generated in real time is monitored. When the comprehensive air quality index value is less than the preset good or excellent level threshold for a preset number of consecutive basic sampling cycles, the basic sampling cycle is extended to the dormant sampling cycle. During the dormant sampling period, an output linkage command is used to cut off the power supply circuits of the gaseous pollutant sensing unit and the laser scattering particulate matter sensing unit, leaving only the capacitive temperature and humidity sensing unit to maintain environmental background monitoring; when the change of any environmental background parameter within a preset time window exceeds the preset environmental background parameter change threshold, the sampling period is immediately restored.

Images