Multi-scene-oriented volatile organic compound real-time monitoring and early warning method

Abstract

The invention discloses a multi-scene-oriented volatile organic compound real-time monitoring and early warning method, and relates to the technical field of intelligent sensor monitoring. The method comprises the following steps: dividing microenvironment types in a target scene, defining the health tolerance of a benchmarking object, selecting whether to send a marking signal or not, synchronously constructing a microenvironment health tolerance model, and outputting microenvironment health tolerance indexes corresponding to different microenvironment types; the method is technically characterized in that a dual-condition trigger logic of a mark signal and an anomaly residual error is utilized, so that a hierarchical response strategy is associated with spatial semantics and reflects pollution causes; and on the other hand, fine control of grading and enhanced response measures is also realized, balance between excessive intervention and insufficient response is sought, and the extensive problem existing in a switching mode in the traditional early warning is solved, so that the intervention measures are accurately matched with the risk grades, and the effects of efficiency and user experience are considered to a certain extent.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensor monitoring technology, specifically to a method for real-time monitoring and early warning of volatile organic compounds (VOCs) in multiple scenarios. Background Technology

[0002] Intelligent sensor monitoring refers to the technical process of using intelligent sensors that integrate sensing elements, microprocessors, signal conditioning circuits, and communication modules to perceive, process, and analyze physical, chemical, or biological parameters in the environment or equipment in real time, and transmit the results to a monitoring platform via wired or wireless means. Unlike traditional sensors that only output raw analog signals, intelligent sensors have features extraction and edge computing capabilities, enabling continuous monitoring. Existing technologies are widely used in workplace environments, industrial equipment status monitoring, smart building environmental management, public health protection, and Internet of Things (IoT) systems. Typical examples include photoionization detectors and electrochemical VOCs intelligent sensors. Combined with Bluetooth or NB-IoT communication, distributed real-time monitoring networks can be built for the real-time monitoring of volatile organic compounds.

[0003] In some public places, existing VOCs monitoring technologies generally suffer from the problem of emphasizing detection while neglecting the context. Traditional fixed sensor networks rely on a single threshold alarm, which cannot distinguish between temporary increases in TVOC caused by normal human activities and real pollution events, leading to frequent false alarms and disrupting daily operations. Conventional methods lack differentiated consideration of health risk tolerance in different microenvironments, such as corridors, offices, and restrooms. Often, due to one-sided judgment strategies, invalid alarms are frequently triggered in high-traffic areas. Furthermore, the monitoring scheme does not take into account the linkage with human behavior or spatial semantics, thus failing to achieve dynamic source tracing and precise intervention. As a result, the reliability of traditional monitoring and early warning systems is insufficient in emergency response to public health emergencies. Summary of the Invention

[0004] To achieve the above objectives, this invention provides the following technical solution: a real-time monitoring and early warning method for volatile organic compounds (VOCs) in multiple scenarios. This method includes: classifying microenvironment types within a target scenario, defining target health tolerance levels, selecting whether to send a marker signal, and simultaneously constructing a microenvironment health tolerance model, outputting the microenvironment health tolerance index corresponding to different microenvironment types; acquiring a trajectory-related dataset of the target object, and constructing a human-air-VOCs coupling model, using the trajectory-related dataset as input and outputting the results as the theoretical concentration field at each point at the current moment; the trajectory-related dataset includes the target object's movement trajectory, the time the target object enters the area corresponding to the microenvironment type, and the prior-set target object carrying... The system calculates the preset release rate of VOCs; extracts the theoretical concentration field and compares it with a standard threshold. When the theoretical concentration field exceeds the standard threshold, it drives the inspection unit equipped with GC-MS / MS to standby. Based on the microenvironmental health tolerance index corresponding to the target microenvironment type, combined with the preset basic health reference threshold and fluctuation amplification factor, a dynamic alarm threshold is generated. A two-level judgment mechanism is run synchronously to obtain the anomaly residual and select whether to trigger a graded warning. Under the condition of triggering a graded warning, a feedback action is executed, and it is determined whether the result obtained from the feedback action is taken as the standard. When a graded warning is triggered, a marker signal is received synchronously, and a graded response strategy is executed to implement response measures in the area corresponding to the target microenvironment type.

[0005] Furthermore, the defined microenvironment types include at least: open office area, passageway area, private office area, and health service area. The criteria for selecting whether to send a labeling signal are as follows: when VOCs concentration exceeds the target health tolerance level in the open office area, a Class I labeling signal is sent; when VOCs concentration exceeds the target health tolerance level in the passageway area, a Class II labeling signal is sent; when VOCs concentration exceeds the target health tolerance level in the private office area, a Class III labeling signal is sent; and when VOCs concentration exceeds the target health tolerance level in the health service area, a Class IV labeling signal is sent.

[0006] Furthermore, the process of constructing the microenvironment health tolerance model is as follows: extract the health tolerance T_max corresponding to the i-th type of microenvironment. i Simultaneously acquire the indoor reference concentration of VOCs T_ref; and match the health tolerance T_max of the i-th microenvironment type. i As the numerator, the indoor reference concentration of VOCs, T_ref, is used as the denominator to generate the microenvironment health tolerance index Mehtl corresponding to the i-th microenvironment type. i Where i represents the number of the corresponding microenvironment type.

[0007] Furthermore, the target object represents each active person; the process of running the person-space-VOCs coupled model is as follows: input the movement trajectory (x) of each active person k. k (t), y k (t)), the time t for entering the region corresponding to the microenvironment type k A priori setting of the preset release rate Q carrying VOCs. k ; Calculate the dwell time Δt for each active user after entering the area corresponding to any microenvironment type. k =tr-t k Where tr represents the current global time; substituting into the defined Gaussian diffusion sub-model, the concentration contribution C of each active person k at the spatial point (x, y) is generated. k (x, y, t); C represents the concentration contribution C of each active person k at the spatial point (x, y). k Summing (x, y, t) yields the total predicted concentration, which is the theoretical concentration field C_pred(x, y, t) at each point at the current time.

[0008] Furthermore, the process of obtaining the fluctuation amplification factor is as follows: multiply the population density of the area corresponding to the current target microenvironment type by the set calibration coefficient to obtain the first product result, and add the first product result to 1 to obtain the fluctuation amplification factor; the process of generating the dynamic alarm threshold is as follows: multiply the microenvironment health tolerance index, the basic health reference threshold and the fluctuation amplification factor corresponding to the i-th type of microenvironment in sequence, and the resulting second product result is the dynamic alarm threshold Ht_alert.

[0009] Furthermore, the process of running the two-level judgment mechanism is as follows: extract the theoretical concentration field C_pred(x, y, t) at each point at the current time; obtain the measured concentration field C_meas(x, y, t) according to the inspection unit configured in GC-MS / MS; perform residual calculation based on: anomaly residual ΔC(x, y, t) = C_meas(x, y, t) - C_pred(x, y, t); and perform two-level judgment. The first-level judgment is: if C_meas(x, y, t) does not exceed Ht_alert, no response action is taken; otherwise, the second-level judgment is performed: compare ΔC(x, y, t) with the preset residual tolerance threshold. When the anomaly residual ΔC(x, y, t) is lower than the residual tolerance threshold, only a record is made, and no graded warning is triggered; otherwise, a graded warning is triggered.

[0010] Furthermore, the process of performing the feedback action is as follows: within a preset period, the inspection unit configured with GC-MS / MS performs fixed-point sampling in the corresponding areas of different microenvironment types; the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) are retrieved and fitted, and the fitting is based on the fixed personnel trajectory: the movement trajectory (x, y, t) of each active personnel k is made to be... k (t), y k (t) and the time t for entering the region corresponding to the microenvironment type k Keep it unchanged; set the parameter to be optimized: change the original prior setting to include the preset release rate Q of VOCs. k As the sole regulating variable; calculate the predicted concentration sequence: for each time point t, calculate the theoretical concentration field C_pred(x, y, t) using the Gaussian diffusion sub-model; construct the error function: compare the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) point by point, and calculate the sum of the mean square errors; optimize the solution: adjust the sole regulating variable using the least squares method to minimize the sum of the mean square errors, and the obtained optimal value is the optimal preset release rate Q carrying VOCs. k * Based on a priori settings, a preset release rate Q carrying VOCs is determined. k and the optimal preset release rate Q carrying VOCs k * The two results obtained from running the human-air-VOCs coupling model are compared using absolute difference calculation. If the difference is not lower than a preset threshold, the result obtained from the feedback action is taken; otherwise, the result based on Q is used. k The results obtained from running the human-air-VOCs coupling model shall prevail.

[0011] Furthermore, the tiered response strategy is based on the following: When dealing with open office areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.02 or a Class I marker signal is received, a Class I response measure is implemented; if both conditions are met, a Class I enhanced response measure is implemented. When dealing with passageway areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.1 or a Class II marker signal is received, a Class II response measure is implemented; if both conditions are met, a Class II enhanced response measure is implemented. When dealing with private office areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.05 or a Class III marker signal is received, a Class III response measure is implemented; if both conditions are met, a Class III enhanced response measure is implemented. When dealing with health service areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.2 or a Class IV marker signal is received, a Class IV response measure is implemented; if both conditions are met, a Class IV enhanced response measure is implemented.

[0012] Furthermore, the response measures are as follows: Category 1: turn on the fresh air system and push a reminder to the nearest workstation; Category 2: record the event and mark it as suspicious lingering if it lasts for more than 10 seconds; Category 3: if the APP pushes a notification that an unexpected odor has been detected, confirm whether chemicals have been used within 1 hour; Category 4: start the exhaust system and notify the cleaning service to carry out the established inspection procedures.

[0013] Furthermore, the enhanced response measures are as follows: First, immediately turn on the highest level of fresh air supply, push a reminder to the nearest workstation, and simultaneously push reminders to the other workstations; Second, record the event and directly mark the suspicious situation; Third, if an unexpected odor is detected by the APP, immediately confirm whether chemicals have been used; Fourth, immediately turn on the highest level of exhaust ventilation and simultaneously notify the cleaning and maintenance personnel to carry out the established inspection procedures.

[0014] This invention provides a real-time monitoring and early warning method for volatile organic compounds in multiple scenarios, which has the following beneficial effects: (1) This solution not only realizes the transformation of personnel movement behavior from interference noise into an active driving factor for VOCs diffusion, but also dynamically predicts the pollution concentration field by constructing a human-air-VOCs coupling model. Furthermore, it realizes graded management of health tolerance based on differences in microenvironmental functions. By setting thresholds differently through the microenvironmental health tolerance index, it solves the problem of high false alarm rate of traditional monitoring systems for volatile organic compounds in office scenarios, and achieves the effect of accurately identifying real pollution risks even under high-traffic activity.

[0015] (1) This scheme performs linkage operation of GC-MS / MS high-precision data and edge sensor network. It selectively optimizes the preset release rate of VOCs by periodically feeding back and realizes a dual verification mechanism of model prediction and actual measurement deviation. The parameters are updated only when there is a significant deviation. To a certain extent, it solves the problem of prediction inaccuracy caused by model drift and sensor aging in long-term operation, and ensures that the system can achieve adaptive and self-correcting monitoring effect.

[0016] (2) This scheme uses the absolute difference to calculate and compare the original Q. k With feedback optimization Q k * If the difference between the obtained concentration field prediction results and the actual difference is not lower than the threshold, it indicates that the original model has a significant bias. In this case, Q is used. k It can more accurately reflect actual pollution behavior; otherwise, it retains the original Q. kThis avoids over-adjustment of the model due to sampling noise or occasional interference caused by high-frequency use of GC-MS / MS. It effectively solves the technical problem of balancing model parameter drift and the reliability of measured data. While ensuring prediction stability, it achieves dynamic calibration of people's VOCs release behavior, and improves the long-term accuracy and adaptability of the human-air-VOCs coupling model in complex office scenarios.

[0017] (3) This scheme utilizes the dual-condition triggering logic of the marker signal and the anomaly residual to make the graded response strategy both related to spatial semantics and reflect the causes of pollution; on the other hand, it also realizes the refined control of graded and enhanced response measures, seeks a balance between excessive intervention and insufficient response, solves the extensive problem of the traditional early warning mode of whether or not there is a switch, and makes the intervention measures and risk levels accurately matched, which to a certain extent takes into account the efficiency and user experience, and ensures the reliability of the overall scheme operation. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for real-time monitoring and early warning of volatile organic compounds (VOCs) across multiple scenarios. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: Please refer to Figure 1 This embodiment provides a real-time monitoring and early warning method for volatile organic compounds (VOCs) in multiple scenarios. The brief outline of this solution is as follows: In a public workplace air environment, a multi-layered architecture is constructed, specifically: a personnel trajectory perception layer: selectively identifying personnel density and movement paths through Wi-Fi probes, Bluetooth beacons, or infrared thermal imaging; a microenvironment semantic modeling layer: dividing the physical space into functional semantic units, such as open public areas, passageways / corridors, and sanitary areas; a smart sensor dynamic deployment layer: employing mobile / reconfigurable sensor nodes combined with fixed nodes to form a hybrid network; and a spatiotemporal coupling early warning layer: fusing location, time, concentration, and four-dimensional data of human flow to achieve precise mapping of where people are, where pollution is, and where the risk is. Its key innovation lies in: adopting a human-space coupling monitoring paradigm, treating personnel movement as a driving factor for VOCs diffusion rather than interference noise, and proposing a Microenvironment Health Tolerance Index (MEHTI) to differentiate and dynamically set corresponding early warning thresholds.

[0021] The specific steps of this method are as follows: S1. Divide the microenvironment types in the target scenario, define the target health tolerance, select whether to send a marker signal, and simultaneously construct a microenvironment health tolerance model to output the microenvironment health tolerance index corresponding to the i-th type of microenvironment.

[0022] The target scenario can be selected according to actual needs, such as the office space of the CDC or a public place, in order to respond to public health emergencies. In this embodiment, the target scenario is the public office space of the CDC.

[0023] The microenvironment types include at least: open office area, passageway area, private office area, and health service area, etc.; this embodiment divides these four areas, and the target health tolerance is: the target health tolerance for the open office area is 0.1, in mg / m³. 3 This is considered an extremely low level, representing the maximum allowable instantaneous VOCs concentration for the corresponding microenvironment type; the health tolerance level for the passageway area is 0.5, expressed in mg / m³. 3 The level is considered medium; the health tolerance level for private office areas is 0.2, measured in mg / m³. 3 This falls under the low-level category; the health tolerance level for the health service area is 1.0, measured in mg / m³. 3 It belongs to the high level.

[0024] The table for classifying microenvironment types is based on Table 1, which defines the environment types and explains feature extraction.

[0025] Microenvironment types Typical areas Personnel behavioral characteristics Main sources of VOCs Health tolerance Open office area Workstation cluster Prolonged stay (>4 hours), slow movement Printer, carpet, and screen cleaner Extremely low, 0.1 mg / m³ Passage corridor area Connecting areas Frequent short stops (<1 minute), quick passage Shoe soles can introduce contaminants and wall paint. Moderate, 0.5 mg / m³ Private office area Private room Medium stay (1 to 4 hours), occasional entry and exit Equipment release, personal items Low, 0.2 mg / m³ Health service area Toilet Short stay (2 to 5 minutes), intermittent use Cleaning agents, excrement volatiles High, 1.0 mg / m³ .

[0026] As shown in Table 1 above, the CDC's office space is divided into these four microenvironment types. The table also provides the behavioral characteristics of the personnel and the main sources of VOCs within each microenvironment, which will be helpful for subsequent strategy configuration.

[0027] The process for selecting whether to send a labeling signal is based on the following criteria: when VOCs concentrations exceed the target health tolerance level in an open office area, a Class I labeling signal is sent; when VOCs concentrations exceed the target health tolerance level in a passageway area, a Class II labeling signal is sent; when VOCs concentrations exceed the target health tolerance level in a private office area, a Class III labeling signal is sent; and when VOCs concentrations exceed the target health tolerance level in a health service area, a Class IV labeling signal is sent.

[0028] The process of synchronously constructing the microenvironment health tolerance model is as follows: extract the health tolerance T_max of the i-th microenvironment type. iSimultaneously acquire the indoor reference concentration of VOCs T_ref; and match the health tolerance T_max of the i-th microenvironment type. i As the numerator, the indoor reference concentration of VOCs, T_ref, is used as the denominator to generate the microenvironment health tolerance index Mehtl corresponding to the i-th microenvironment type. i Where i represents the number of the corresponding microenvironment type. For example, the number corresponding to the open office area is 1, i.e., i=1. Then the numbers corresponding to the passageway area, private office area and health service area are 2, 3 and 4 respectively, which will not be elaborated here. The indoor reference concentration of VOCs T_ref can be obtained by referring to the WHO Indoor Air Quality Guidelines. In this embodiment, the specific value set can be selected as T_ref=0.1mg / m³.

[0029] S2. Obtain the trajectory-related dataset of the target object and build a human-space-VOCs coupling model. Take the trajectory-related dataset as input and output the results as the theoretical concentration field C_pred(x, y, t) at each point at the current time. Here, the target object represents each active person. The trajectory-related dataset includes at least: the movement trajectory of the target object, the time when the target object enters the area corresponding to the microenvironment type, and the preset release rate of VOCs carried by the target object.

[0030] In this context, each active person, represented by the target object, is defined by both spatial and behavioral conditions. The spatial condition is: the person is located within the area corresponding to the currently monitored microenvironment type; if the person has left the area (e.g., from an open office area to a corridor), they are not counted in the total number of active people in that area. The behavioral condition satisfies any of the following sub-conditions: Sub-condition 1: Currently engaging in a predefined high-emission behavior; Sub-condition 2: Having engaged in high-emission behavior within a preset period; Sub-condition 3: Being in a state of dense human interaction. Specifically, high-emission behaviors include at least using a printer and cleaning a desktop, which can be identified through camera behavior recognition; details are omitted here. Regarding "having engaged in high-emission behavior within a preset period," the preset period can be the past 5 minutes or a period set according to actual needs. Because VOCs release is continuous, sub-condition 2 needs to be set. A state of dense human interaction indicates an event that triggers a meeting or group discussion.

[0031] In trajectory-related datasets, the movement trajectory of a target object can be achieved through two complementary methods; Method 1: Bluetooth beacon-based positioning, for example: positioning the target object at approximately 10m above the ceiling in an open office area. 2The system employs two main methods: Method 1: Spacing-based deployment of low-power Bluetooth beacons supporting AoA or broadcast modes. Personnel wear name tags or authorize mobile apps with built-in BLE receiver modules. The device continuously scans the RSSI (Radio Signal Strength Index) and broadcast packet content of surrounding beacons. The system uses multi-beacon triangulation or fingerprint matching algorithms to calculate the personnel's two-dimensional coordinates at 1 Hz in the edge gateway and outputs a movement path sequence, such as (x1, y1, t1) → (x2, y2, t2)... Method 2: Infrared thermal imaging-assisted perception. Infrared thermal imagers are installed at key nodes such as passageways and entrances / exits to detect the distribution of human body heat sources in real time in a non-contact manner. The device's built-in embedded algorithm only extracts the number of people and centroid movement vectors, without saving the original images, ensuring privacy compliance. This also operates at 1 Hz. The system outputs pedestrian density and direction data at Hz frequency. After fusing the two methods, it can accurately reconstruct the dynamic trajectory of people in the corresponding areas of various microenvironment types, providing spatiotemporal driving input for the subsequent human-air-VOCs coupled model that reflects VOCs diffusion. The preset release rate of VOCs carried by the target object is set to the prior value obtained by previous GC-MS / MS, i.e., gas chromatography-tandem mass spectrometry.

[0032] By adopting the above technical solutions, on the one hand, it realizes the transformation of personnel movement behavior from interference noise into an active driving factor for VOCs diffusion, and dynamically predicts the pollution concentration field by constructing a human-air-VOCs coupling model; on the other hand, it realizes the hierarchical management of health tolerance based on the differences in microenvironmental functions, and solves the problem of high false alarm rate of traditional monitoring systems for volatile organic compounds in office scenarios by setting thresholds differently through the microenvironmental health tolerance index, and achieves the effect of accurately identifying the real pollution risk even under high traffic conditions.

[0033] The process of running the human-space-VOCs coupled model is as follows: S2.1, input the movement trajectory (x) of each active person k. k (t), y k (t)), the time t for entering the region corresponding to the microenvironment type k A priori setting of the preset release rate Q carrying VOCs. k S2.2 Calculate the dwell time Δt of each active user after entering the area corresponding to any microenvironment type. k =tr-t k Where tr represents the current global time, which serves as the time reference for model calculation; S2.3, Substitute the set Gaussian diffusion sub-model to generate the concentration contribution C of each active person k at the spatial point (x, y). k (x, y, t); where, the process of running the Gaussian diffusion sub-model is as follows: S2.3.1, preset the air turbulence diffusion coefficient D, and combine it with 4π, residence time Δt kThe cumulative multiplication yields the diffusion area scale factor Dt, which serves as the denominator; where the air turbulence diffusion coefficient D is set to a typical value for an office environment, D = 0.1m. 2 / s; S2.3.2, Set the preset release rate Q carrying VOCs according to prior settings. k As the numerator; S2.3.3, for any spatial point (x, y), calculate its movement trajectory (x, y) relative to the corresponding active person k. k (t), y k The squared Euclidean distance r of (t) 2 Square the Euclidean distance r 2 Substituting into the exponential function A: A = exp(-r 2 / Dt); where, as r increases, the exponential term decays rapidly, indicating that the concentration decreases in a Gaussian manner with distance; S2.3.4, multiply the ratio of the numerator to the denominator by the exponential function A to obtain the concentration contribution C of each active person k at the spatial point (x, y). k (x, y, t); S2.4, the concentration contribution C of each active person k at the spatial point (x, y). k Summing (x, y, t) yields the total predicted concentration, which is the theoretical concentration field C_pred(x, y, t) at each point at the current time.

[0034] S3. Extract the theoretical concentration field C_pred(x, y, t) and compare it with the standard threshold. If the theoretical concentration field C_pred(x, y, t) exceeds the standard threshold, the area corresponding to the target microenvironment type is determined to be a predicted high-pollution area, and the inspection unit configured with GC-MS / MS is driven to stand by; otherwise, no response action is taken.

[0035] The standard threshold is a pre-set definition indicator used to initially determine whether the area corresponding to the target microenvironment type is a predicted high-pollution area, thereby completing the subsequent drive adjustment operation, that is, driving the inspection unit equipped with GC-MS / MS to stand by; and GC-MS / MS refers to gas chromatography-tandem mass spectrometry, and the inspection unit can be an inspection drone or a human inspection team; its purpose is to obtain accurate VOCs type and concentration spectrum, which can be used to correct sensor drift and cross-interference errors to a certain extent.

[0036] S4. Based on the microenvironment health tolerance index corresponding to the target microenvironment type, combined with the preset basic health reference threshold and fluctuation amplification factor, a dynamic alarm threshold is generated, and a two-level judgment mechanism is run simultaneously to obtain the abnormality residual ΔC(x, y, t), and to select whether to trigger a graded warning.

[0037] The basic health reference threshold is taken from established standards, such as the WHO's recommended long-term indoor TVOC exposure limit of 0.1 mg / m³. 3 , representing the maximum acceptable concentration in an ideal silent environment; the fluctuation amplification factor reflects the reasonable fluctuation amplification factor brought about by the current human activity, based on: multiplying the human density of the area corresponding to the current target microenvironment type with the set calibration coefficient to obtain the first product result, and adding the first product result to 1 to obtain the fluctuation amplification factor; it should be noted that the calibration coefficient is experimentally calibrated, and the value range is from 0 to 1. In this embodiment, the value can be 0.2, which means that for every additional 1 person / m², the tolerance increases by 20%; the basis for generating the dynamic alarm threshold is: multiplying the microenvironment health tolerance index, the basic health reference threshold and the fluctuation amplification factor corresponding to the i-th type of microenvironment in sequence, and the resulting second product result is the dynamic alarm threshold Ht_alert.

[0038] The two-level judgment mechanism operates as follows: S4.1, extract the theoretical concentration field C_pred(x, y, t) for each point at the current time; S4.2, obtain the measured concentration field C_meas(x, y, t) based on the configured GC-MS / MS inspection unit; S4.3, perform residual calculation, based on: anomaly residual ΔC(x, y, t) = C_meas(x, y, t) - C_pred(x, y, t); S4.4, Perform a two-level judgment; the first-level judgment is: if C_meas(x, y, t) does not exceed Ht_alert, no response action is taken, i.e., no alarm is triggered; if C_meas(x, y, t) exceeds Ht_alert, the second-level judgment is performed: the anomaly residual ΔC(x, y, t) is compared with the preset residual tolerance threshold. When the anomaly residual ΔC(x, y, t) is lower than the residual tolerance threshold, it indicates that the exceedance is mainly caused by human behavior, and only a record is made, without triggering a graded warning; when the anomaly residual ΔC(x, y, t) is not lower than the residual tolerance threshold, a graded warning is triggered; it should be noted that the residual tolerance threshold in this embodiment is 0.01 mg / m 3 It is used to distinguish between reasonable concentration fluctuations caused by known human behavior and real risks caused by unknown or abnormal pollution sources; Level 1 judgment is mainly used for judging exceedances, while Level 2 judgment is mainly used for judging anomalies.

[0039] The effects of the two-level judgment mechanism described above are as follows: The application of this scheme can avoid over-reliance on the model. Even if the model prediction is inaccurate, as long as the concentration exceeds the standard, it is still considered safe. At the same time, it can prevent misjudging normal activities. When dealing with the situation where the measured concentration field C_meas(x,y,t) slightly exceeds the dynamic alarm threshold Ht_alert due to high traffic, the anomaly residual ΔC(x,y,t) is very small, indicating that it is still within the expected behavior. Intervention is only initiated when the concentration exceeds the standard and cannot be explained by human behavior, that is, a graded early warning is triggered, which takes into account both safety and robustness and achieves accurate identification of true anomalies.

[0040] S5. Under the condition of triggering a graded warning, perform a feedback action within a preset period and solve for the optimal preset release rate Q of VOCs. k * According to Q k and Q k * The two results obtained from running the human-air-VOCs coupling model are compared using absolute difference calculation. If the difference is not lower than a preset threshold, the result obtained from the feedback action is taken; otherwise, the result based on Q is used. k The results obtained from running the human-air-VOCs coupling model shall prevail.

[0041] This solution not only enables the coordinated operation of high-precision GC-MS / MS data and edge sensor networks, and selectively optimizes the preset release rate of VOCs by periodically feeding back the data, but also implements a dual verification mechanism for model prediction and measured deviation. Parameters are updated only when there is a significant deviation, which to some extent solves the problem of prediction inaccuracy caused by model drift and sensor aging during long-term operation, ensuring that the system can achieve adaptive and self-correcting monitoring effects.

[0042] The process of performing the feedback action is as follows: S5.1, within a preset period, use the inspection unit configured with GC-MS / MS to perform fixed-point sampling in the corresponding areas of different microenvironment types; wherein, the preset period can be selected as 1 week in this embodiment; S5.2, retrieve the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) for fitting, the fitting basis being: S5.2.1, fixed personnel trajectory: using the movement trajectory (x, y, t) of each active personnel k obtained in S2.1. k (t), y k (t)), the time t for entering the region corresponding to the microenvironment type k The parameters remain unchanged; S5.2.2, Set the parameters to be optimized: The original prior settings will carry the preset release rate Q of VOCs. kAs the sole regulating variable; S5.2.3, Calculate the predicted concentration sequence: For each time point t, calculate the theoretical concentration field C_pred(x, y, t) according to the Gaussian diffusion sub-model recorded in S2.3; S5.2.4, Construct the error function: Compare the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) point by point, and calculate the sum of mean square errors; S5.2.5, Optimize the solution: Adjust the sole regulating variable using the least squares method, such as gradient descent or grid search, to minimize the sum of mean square errors. The optimal value obtained is the optimal preset release rate Q carrying VOCs. k * .

[0043] The absolute difference calculation process involves calculating the difference between two results and then performing absolute value processing. This way, regardless of which result is subtracted from which, the result obtained after obtaining the absolute value will be the same. The threshold value is set in advance according to actual needs to indicate whether it is within a reasonable error range.

[0044] By adopting the above technical solution, the absolute difference is used to calculate and compare the original Q. k With feedback optimization Q k * If the difference between the obtained concentration field prediction results and the actual difference is not lower than the threshold, it indicates that the original model has a significant bias. In this case, Q is used. k It can more accurately reflect actual pollution behavior; otherwise, it retains the original Q. k This avoids over-adjustment of the model due to sampling noise or occasional interference caused by high-frequency use of GC-MS / MS. It effectively solves the technical problem of balancing model parameter drift and the reliability of measured data. While ensuring prediction stability, it achieves dynamic calibration of people's VOCs release behavior, and improves the long-term accuracy and adaptability of the human-air-VOCs coupling model in complex office scenarios.

[0045] S6. When a graded early warning is triggered, a marker signal is received synchronously, and a graded response strategy is executed to implement response measures in the area corresponding to the target microenvironment type.

[0046] Specifically, the received marker signals mentioned in S6 correspond to the Class I to Class IV marker signals given in S1 above. The hierarchical response strategy is based on the following: When dealing with open office areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.02 or a Class I marker signal is received, a Class I response measure is implemented; if both the anomaly residual ΔC(x, y, t) exceeds 0.02 and a Class I marker signal is received, a Class I enhanced response measure is implemented. When dealing with passageway areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.1 or a Class II marker signal is received, a Class II response measure is implemented; if both the anomaly residual ΔC(x, y, t) exceeds 0.1, a Class II enhanced response measure is implemented. If a Class II marker signal is received, a Class II enhanced response measure is implemented. When dealing with private office areas, if the anomaly residual ΔC(x, y, t) exceeds 0.05 and any one of the Class III marker signals is received, a Class III response measure is implemented; if both the anomaly residual ΔC(x, y, t) and a Class III marker signal are received, a Class III enhanced response measure is implemented. When dealing with health service areas, if the anomaly residual ΔC(x, y, t) exceeds 0.2 and any one of the Class IV marker signals is received, a Class IV response measure is implemented; if both the anomaly residual ΔC(x, y, t) and a Class IV marker signal are received, a Class IV enhanced response measure is implemented.

[0047] The response measures are categorized as follows: Category 1: Activate fresh air supply and send a notification to the nearest workstation; Category 1 Enhanced Response: Immediately activate the highest fresh air supply setting, send a notification to the nearest workstation, and simultaneously send notifications to other workstations; Category 2: Record the event and mark it as suspicious if it lasts longer than 10 seconds; Category 2 Enhanced Response: Record the event and directly mark it as suspicious; Category 3: If an unexpected odor is detected via the app, confirm whether chemicals have been used within 1 hour; Category 3 Enhanced Response: If an unexpected odor is detected via the app, immediately confirm whether chemicals have been used; Category 4: Activate exhaust ventilation and notify the cleaning service to perform the established inspection procedures; Category 4 Enhanced Response: Immediately activate the highest exhaust ventilation setting and simultaneously notify the cleaning and maintenance services to perform the established inspection procedures. It should be noted that the established inspection procedures can be determined based on the actual procedures developed.

[0048] By adopting the above technical solution, and utilizing the dual-condition triggering logic of the marker signal and the anomaly residual, the graded response strategy is made to both relate to spatial semantics and reflect the causes of pollution. On the other hand, it also realizes the refined control of graded and enhanced response measures, seeks a balance between excessive intervention and insufficient response, and solves the extensive problem of the traditional early warning mode of "on or off". This makes the intervention measures and risk levels accurately matched, and to a certain extent takes into account both efficiency and user experience.

[0049] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0050] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0051] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for real-time monitoring and early warning of volatile organic compounds (VOCs) in multiple scenarios, characterized in that, The method includes: Under the target scenario, the microenvironment types are divided, the health tolerance of the target is defined, the choice of whether to send a marker signal is selected, and the microenvironment health tolerance model is constructed simultaneously to output the microenvironment health tolerance index corresponding to different microenvironment types. The trajectory-related dataset of the target object is obtained, and a human-space-VOCs coupling model is built. The trajectory-related dataset is used as input, and the output result is used as the theoretical concentration field of each point at the current time. The trajectory-related dataset includes the movement trajectory of the target object, the time when the target object enters the area corresponding to the microenvironment type, and the preset release rate of VOCs carried by the target object. Extract the theoretical concentration field and compare it with the standard threshold. When the theoretical concentration field exceeds the standard threshold, drive the inspection unit configured with GC-MS / MS to standby. Based on the microenvironment health tolerance index corresponding to the target microenvironment type, combined with the preset basic health reference threshold and fluctuation amplification factor, a dynamic alarm threshold is generated; a two-level judgment mechanism is run simultaneously to obtain the abnormality residual and select whether to trigger a graded warning. Under the condition of triggering a graded warning, a feedback action is executed, and it is decided whether to take the result of the feedback action as the standard. When a graded early warning is triggered, a marker signal is received simultaneously, and a graded response strategy is executed to implement response measures in the area corresponding to the target microenvironment type.

2. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 1, characterized in that, The microenvironment types are divided into at least the following: open office area, passageway area, private office area, and health service area. The criteria for selecting whether to send a labeling signal are as follows: when the VOCs concentration in the open office area exceeds the target health tolerance level, a Class I labeling signal is sent; when the VOCs concentration in the passageway area exceeds the target health tolerance level, a Class II labeling signal is sent; when the VOCs concentration in the private office area exceeds the target health tolerance level, a Class III labeling signal is sent; and when the VOCs concentration in the health service area exceeds the target health tolerance level, a Class IV labeling signal is sent.

3. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 1, characterized in that, The process of constructing a microenvironment health tolerance model is as follows: extract the health tolerance T_max corresponding to the i-th type of microenvironment. i Simultaneously acquire indoor VOCs reference concentration T_ref; The health tolerance T_max is used to categorize the i-th type of microenvironment. i As the numerator, the indoor reference concentration of VOCs, T_ref, is used as the denominator to generate the microenvironment health tolerance index Mehtl corresponding to the i-th microenvironment type. i Where i represents the number of the corresponding microenvironment type.

4. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 2, characterized in that, The target object represents each active person; the process of running the human-space-VOCs coupled model is as follows: input the movement trajectory (x) of each active person k. k (t), y k (t)), the time t for entering the region corresponding to the microenvironment type k A priori setting of the preset release rate Q carrying VOCs. k ; Calculate the dwell time Δt for each active user after entering the area corresponding to any microenvironment type. k =tr-t k Where tr represents the current global time; substituting into the defined Gaussian diffusion sub-model, the concentration contribution C of each active person k at the spatial point (x, y) is generated. k (x, y, t); C represents the concentration contribution C of each active person k at the spatial point (x, y). k Summing (x, y, t) yields the total predicted concentration, which is the theoretical concentration field C_pred(x, y, t) at each point at the current time.

5. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 4, characterized in that, The process of obtaining the fluctuation amplification factor is as follows: multiply the population density of the area corresponding to the current target microenvironment type by the set calibration coefficient to obtain the first product result, and add the first product result to 1 to obtain the fluctuation amplification factor; The process of generating the dynamic alarm threshold is as follows: the microenvironment health tolerance index, the basic health reference threshold, and the fluctuation amplification factor corresponding to the i-th microenvironment type are multiplied in sequence, and the result of the second product is the dynamic alarm threshold Ht_alert.

6. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 5, characterized in that, The process of running the two-level judgment mechanism is as follows: extract the theoretical concentration field C_pred(x, y, t) at each point at the current time; obtain the measured concentration field C_meas(x, y, t) according to the inspection unit configured in GC-MS / MS; perform residual calculation based on: anomaly residual ΔC(x, y, t) = C_meas(x, y, t) - C_pred(x, y, t); and perform two-level judgment. The first-level judgment is: if C_meas(x, y, t) does not exceed Ht_alert, no response action is taken; otherwise, the second-level judgment is performed: compare ΔC(x, y, t) with the preset residual tolerance threshold. When the anomaly residual ΔC(x, y, t) is lower than the residual tolerance threshold, only a record is made, and no graded warning is triggered; otherwise, a graded warning is triggered.

7. The method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 6, characterized in that, The process of performing the feedback action is as follows: Within a preset period, the inspection unit configured with GC-MS / MS performs fixed-point sampling in the corresponding areas of different microenvironment types; the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) are retrieved and fitted, and the fitting is based on the fixed personnel trajectory: the movement trajectory (x, y, t) of each active personnel k is made to be... k (t), y k (t) and the time t for entering the region corresponding to the microenvironment type k Remain unchanged; Set the parameters to be optimized: Replace the original prior settings with the preset release rate Q of VOCs. k As the sole regulating variable; calculate the predicted concentration sequence: for each time point t, calculate the theoretical concentration field C_pred(x, y, t) using the Gaussian diffusion sub-model; construct the error function: compare the measured concentration field C_meas(x, y, t) and the theoretical concentration field C_pred(x, y, t) point by point, and calculate the sum of the mean square errors; optimize the solution: adjust the sole regulating variable using the least squares method to minimize the sum of the mean square errors, and the obtained optimal value is the optimal preset release rate Q carrying VOCs. k * Based on a priori settings, a preset release rate Q carrying VOCs is determined. k and the optimal preset release rate Q carrying VOCs k * The two results obtained from running the human-air-VOCs coupling model are compared using absolute difference calculation. If the difference is not lower than a preset threshold, the result obtained from the feedback action is taken; otherwise, the result based on Q is used. k The results obtained from running the human-air-VOCs coupling model shall prevail.

8. A method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 6, characterized in that, The tiered response strategy is based on the following: When dealing with open office areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.02 or a Class I marker signal is received, a Class I response measure is implemented; if both conditions are met, a Class I enhanced response measure is implemented. When dealing with passageway areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.1 or a Class II marker signal is received, a Class II response measure is implemented; if both conditions are met, a Class II enhanced response measure is implemented. When dealing with private office areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.05 or a Class III marker signal is received, a Class III response measure is implemented; if both conditions are met, a Class III enhanced response measure is implemented. When dealing with health service areas, if either the anomaly residual ΔC(x, y, t) exceeds 0.2 or a Class IV marker signal is received, a Class IV response measure is implemented; if both conditions are met, a Class IV enhanced response measure is implemented.

9. A method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 8, characterized in that, The first type of response measure is to turn on the fresh air and push a reminder to the nearest workstation; the second type of response measure is to record the event and mark it as a suspicious stay if it lasts for more than 10 seconds; the third type of response measure is to push an APP notification that an unexpected odor has been detected and confirm whether chemicals have been used within 1 hour; the fourth type of response measure is to start the exhaust ventilation and notify the cleaning staff to carry out the established inspection procedures.

10. A method for real-time monitoring and early warning of volatile organic compounds for multiple scenarios according to claim 8, characterized in that, The first type of enhanced response measure is to immediately turn on the highest level of fresh air supply, send a reminder to the nearest workstation, and simultaneously send reminders to the other workstations; the second type of enhanced response measure is to record the event and directly mark the suspicious situation; the third type of enhanced response measure is to immediately confirm whether any chemicals have been used if an unexpected odor is detected by the APP; the fourth type of enhanced response measure is to immediately turn on the highest level of exhaust ventilation and simultaneously notify the cleaning and maintenance personnel to carry out the established inspection procedures.

Images