Intelligent monitoring gas sensor and filtering alarm platform system

Abstract

This invention discloses an intelligent gas monitoring sensor and filter alarm platform system, belonging to the field of industrial equipment monitoring and predictive maintenance technology. The system uses multi-point sensors to collect real-time data on gas concentration, temperature, humidity, pressure difference, airflow, and equipment operating status of air filters and inlet / outlet air ducts. After data preprocessing and array sensor inversion algorithms, a two-dimensional gas flow field distribution image is reconstructed. Based on flow field feature extraction and weighted calculation, the filter blockage index (FCI), flow field offset index (LOI), and turbulence humidification index (TWI) are obtained and compared with corresponding thresholds to determine filter blockage, installation misalignment, and filter media moisture status, triggering corresponding strategies and establishing event records. A convolutional neural network is used to train the event data to construct a health degradation prediction model, enabling filter media life prediction and replacement recommendations. This system can achieve intelligent monitoring of air filter operating status, risk warning, and lifespan management.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment monitoring and predictive maintenance technology, specifically to an intelligent gas monitoring sensor and filtration alarm platform system. Background Technology

[0002] Air filters are crucial components of air purification systems in industrial workshops, and their operational status directly impacts the quality of the production environment and the reliability of equipment. Current air filter management methods primarily rely on single-parameter lifespan prediction models or periodic manual inspections, typically using simple assessments based on indicators such as pressure differential and airflow. These methods suffer from overly simplistic model structures that fail to reflect the complex degradation patterns of filter media under different operating conditions, making it impossible to accurately predict future filter media lifespan and promptly identify potential risks.

[0003] Meanwhile, existing alarms and maintenance processes are often disconnected. Even if the system detects anomalies such as blockage, humidification, or flow field deviation, it often only triggers a single alarm, which cannot be directly linked to subsequent maintenance actions. The lack of event tracing and operational data recording makes it difficult to provide a basis for subsequent analysis.

[0004] Furthermore, traditional systems lack predictive and proactive maintenance capabilities. Monitoring the health status of filter media is mostly limited to a reactive response phase, failing to identify potential problems in advance through multi-parameter data fusion and intelligent algorithms, and also struggling to propose proactive maintenance or replacement recommendations by combining historical events and real-time operational data. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent gas monitoring sensor and filtration alarm platform system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent gas monitoring sensor and filtration alarm platform system, comprising:

[0007] The data acquisition module is used to comprehensively monitor the gas diffusion, thermal and humidity environment, flow resistance, pressure difference changes, and operating status of the air filters and their inlet and outlet air ducts in the factory workshop; it collects the spatial gas concentration sequence Ci(t), temperature distribution sequence Ti(t), humidity distribution sequence Hi(t), spatial pressure difference distribution sequence ΔPi(t), air volume sequence Fa(t), and speed sequence Sa(t).

[0008] The flow field imaging module is used to unify the format, synchronize the time, align the space and remove anomalies from the data collected from multiple sources, and generate a standardized multi-parameter dataset; then, the two-dimensional gas flow field distribution image F(x,y) of the filter working area is jointly reconstructed through array sensor inversion algorithm.

[0009] The flow field blockage monitoring module is used to extract the main channel width Wc, local average velocity Vc, and pressure difference balance index ΔPvar based on the two-dimensional gas flow field distribution image F(x,y), calculate the filter blockage index FCI, and compare it with the filter blockage threshold Fth to determine whether the filter is at risk of blockage. If it is, an appropriate strategy is given, and a blockage event record dataset is established.

[0010] The flow field offset monitoring module is used to obtain the global centroid coordinates (xc, yc) and local flow distribution imbalance coefficient Qvar of the flow field based on the two-dimensional gas flow field distribution image F(x, y). Combined with the reference calibration coordinates (x0, y0), the flow field offset index LOI is calculated and compared with the flow field offset threshold Lth to determine whether the current filter installation position is reasonable. If it is not reasonable, an appropriate strategy is given and an offset event record dataset is established.

[0011] The turbulence humidification monitoring module is used to obtain the turbulent kinetic energy Ek and the spatial humidity distribution gradient based on the two-dimensional gas flow field distribution image F(x,y). The turbulent humidification index TWI is calculated by combining the local concentration fluctuation RMS value Crms with the humidification threshold Tth to determine whether the filter material of the air filter is damp in the current operating state. If it is damp, an appropriate strategy is given and a humidification event record dataset is established.

[0012] The health prediction module is used to build an initial model of filter health degradation using a convolutional neural network (CNN). It trains the model on data of blockage, offset, and humidification events to build a health degradation prediction model, predict the lifespan of the filter media, and output a recommended replacement date range.

[0013] Preferably, the data acquisition module includes a gas concentration acquisition unit, an environmental parameter acquisition unit, a pressure difference distribution acquisition unit, and a pneumatic equipment operating status acquisition unit;

[0014] The gas concentration acquisition unit is used to monitor the gas diffusion status within air filters and their inlet and outlet air ducts installed in the factory workshop environment. By deploying a multi-point miniature gas sensor array, a sampling synchronization triggering device, and a spatial node calibration device, it performs time-series gas concentration acquisition and spatial index calibration at each sampling location, acquiring the spatial gas concentration sequence Ci(t) for each sensor node. Combining a multi-point concentration difference algorithm and a spatial gradient reconstruction model, it performs local gradient calculation on the continuous sampling results to obtain the gas concentration gradient. ;

[0015] The environmental parameter acquisition unit is used to monitor the thermal and humidity state of the operating environment of the air filter and its impact on flow field imaging. By deploying temperature sensing nodes, humidity sensing nodes, and thermal-humidity coupling sampling devices, it monitors the inlet temperature field, filter material surface temperature distribution, and temperature changes in the clean area after filtration in real time, and acquires the temperature distribution sequence Ti(t). Through environmental humidity probes, miniature dew point sensors, and filter material moisture content change analysis equipment, it monitors the front-end humidity field of the filter, the moisture absorption state of the filter material, and downstream water vapor residue in real time, and acquires the humidity distribution sequence Hi(t).

[0016] The differential pressure distribution acquisition unit is used to monitor the flow resistance characteristics and differential pressure change behavior of the air filter under different operating conditions. By installing differential pressure sensors and miniature differential pressure probes at the air inlet, filter media cavity and air outlet, and combining differential pressure fluctuation analysis algorithms, the pressure changes before and after the filter are monitored in real time to obtain the spatial differential pressure distribution sequence ΔPi(t).

[0017] The pneumatic equipment operation status acquisition unit is used to monitor the operation status of the blower and air pump that are matched with the filtration system and evaluate their impact on the flow field structure and filter material load. Through the air volume meter, speed sensor and operating power acquisition device, the output air volume and speed of the blower and air pump are monitored in real time, and the air volume sequence Fa(t) and speed sequence Sa(t) are acquired.

[0018] Preferably, the flow field imaging module includes a data preprocessing unit and a flow field imaging reconstruction unit;

[0019] The data preprocessing unit is used to process the gas concentration sequence Ci(t) and gas concentration gradient using a multi-source data format unified algorithm. The temperature distribution sequence Ti(t), humidity distribution sequence Hi(t), pressure difference distribution sequence ΔPi(t), air volume sequence Fa(t), and speed sequence Sa(t) are formatted uniformly; time synchronization correction is performed on each sequence using node timestamp calibration technology; spatial alignment and interpolation are performed on the data of each sampling point using spatial coordinate interpolation method; anomaly detection and rejection algorithms are adopted to automatically reject instantaneous impulse noise, sensor drift points, and abnormal data that do not conform to physical laws, and a standardized multi-parameter dataset is established.

[0020] The flow field imaging reconstruction unit is used to construct a gas flow parameter matrix and a spatial node observation matrix, and then use the array inverse solver algorithm based on array sensors to jointly invert the spatial concentration field, temperature and humidity field and pressure difference field in the standardized multi-parameter dataset to reconstruct the two-dimensional gas flow field distribution image F(x, y) of the filter working area.

[0021] Preferably, the flow field blockage monitoring module includes a flow field feature extraction unit, a first calculation unit, and a first analysis unit;

[0022] The flow field feature extraction unit is used to perform structured identification of the dominant flow path based on the two-dimensional gas flow field distribution image F(x, y) using a streamline connected domain search algorithm and channel skeleton extraction technology. Then, a cross-sectional flow distribution analysis method is used to calculate the cross-sectional flow contour lines of the streamline connected domains and skeleton centerlines, quantifying the flow distribution characteristics within the steady-state region, calculating the average effective width of the main channel, and obtaining the main channel width Wc. Based on the concentration distribution change rate in the two-dimensional gas flow field distribution image F(x, y), combined with the air volume sequence Fa(t) and the rotational speed sequence Sa(t), a flow field concentration gradient is used... The Gradient-FlowrateVelocityInference algorithm is used to calculate the instantaneous velocity field of the main channel and its surrounding local area. Then, the local average velocity Vc is obtained through regional averaging filtering and time sliding window smoothing. Based on the two-dimensional gas flow field distribution image F(x,y), the flow field is divided into spatial grids using a uniform grid division method, and the pressure difference distribution data ΔPi(t) in each spatial grid is mapped. Then, the regional pressure difference statistical analysis method and variance normalization calculation technique are used to perform statistical and variance normalization processing on the pressure difference in each spatial grid to obtain the pressure difference balance index ΔPvar.

[0023] Preferably, the first calculation unit is used to calculate the filter blockage index FCI by obtaining the main channel width Wc, the local average flow velocity Vc, and the pressure difference balance index ΔPvar, and after dimensionless processing.

[0024] The first analysis unit is used to obtain a first evaluation result by comparing the filter blocking index Fth with a preset filter blocking threshold Fth, and then comparing the filter blocking degree index Fth with the filter blocking threshold Fth.

[0025] When the filter clogging index FCI is less than the filter clogging threshold Fth, it is determined that there is no risk of filter clogging, and the filter is maintained in normal operation and continuously monitored.

[0026] When the filter clogging index FCI is greater than or equal to the filter clogging threshold Fth, the filter is determined to be at risk of clogging, triggering the first warning command and generating the first strategy: automatically increasing the fan output power by 20% to increase the average flow velocity in the main channel by 3%-10% to maintain the target air supply volume; and establishing a clogging event record dataset, linking the filter clogging index FCI, fan adjustment parameters and current operating status into the database.

[0027] Preferably, the flow field migration monitoring module includes a migration feature extraction unit, a second calculation unit, and a second analysis unit;

[0028] The offset feature extraction unit is used to obtain the global centroid coordinates (xc, yc) of the flow field based on the two-dimensional gas flow field distribution image F(x, y) by weighted integration of the flow intensity distribution of each spatial node in F(x, y) using the global centroid calculation method; it also obtains the benchmark calibration coordinates (x0, y0) through equipment installation and calibration records; it establishes a fan performance characteristic curve model and constructs a flow-speed mapping relationship based on the air volume sequence Fa(t) and the speed sequence Sa(t); it estimates the instantaneous air intake using a multi-parameter fitting inversion method and obtains the local flow distribution results through time-series sliding window smoothing; and it combines the flow distribution gradient change reflected in the two-dimensional gas flow field distribution image F(x, y) with the local flow fluctuation statistical algorithm to jointly quantify the flow dispersion and fluctuation amplitude in the spatial region to obtain the local flow distribution unevenness coefficient Qvar.

[0029] Preferably, the second calculation unit is used to calculate the flow field offset index LOI after dimensionless processing of the obtained global centroid coordinates (xc, yc), reference calibration coordinates (x0, y0) and local flow distribution imbalance coefficient Qvar.

[0030] The second analysis unit is used to obtain a second evaluation result by comparing the flow field offset index (LOI) with the flow field offset threshold (Lth) using a preset flow field offset threshold (Lth):

[0031] When the flow field offset index LOI is less than the flow field offset threshold Lth, it indicates that the current filter installation position is reasonable and there is no risk of gas leakage due to structural offset. Continuous monitoring is required.

[0032] When the flow field offset index LOI is greater than or equal to the flow field offset threshold Lth, it indicates that the current filter installation position is unreasonable and there is a risk of gas leakage due to structural offset. This triggers a second warning instruction and generates a second strategy: generating a filter installation position calibration suggestion, prompting the inspection of the filter frame and sealing strip; analyzing the offset direction, automatically marking the leaking spatial area, and initiating a secondary local imaging process to reconstruct the local area in the offset direction, improving the accuracy of leak point identification; and establishing an offset event record dataset, linking the flow field offset index LOI, local flow imbalance Qvar, and calibration suggestion into the database.

[0033] Preferably, the turbulence humidification monitoring module includes a turbulence humidification feature extraction unit, a third calculation unit, and a third analysis unit;

[0034] The turbulent humidification feature extraction unit is used to obtain the turbulent kinetic energy Ek of the flow field by weighting the velocity change rate of each spatial node based on the two-dimensional gas flow field distribution image F(x, y) and the flow field velocity gradient calculation method; based on the humidity distribution sequence Hi(t), it uses a gradient reconstruction algorithm to perform spatial difference and interpolation processing on the humidity data of each sampling point, calculates the humidity change rate of each spatial node, and obtains the spatial humidity distribution gradient. Based on the gas concentration sequence Ci(t), the time-series fluctuation analysis method is used to calculate the root mean square of the concentration time series at each sampling location and obtain the local concentration fluctuation RMS value Crms.

[0035] Preferably, the third calculation unit is used to calculate the turbulent kinetic energy Ek of the flow field and the spatial humidity distribution gradient. The RMS value of local concentration fluctuation (Crms) was used as the basis for the turbulent humidification index (TWI) after dimensionless processing.

[0036] The third analysis unit is used to obtain a third evaluation result by comparing the turbulent humidification index TWI with the humidification threshold Tth using a preset humidification threshold Tth.

[0037] When the turbulent humidification index TWI ≤ humidification threshold Tth, it indicates that the filter media of the air filter is not damp and is maintaining normal operation, requiring continuous monitoring.

[0038] When the turbulent humidification index TWI > humidification threshold Tth, it indicates that the filter media of the air filter is damp in its current operating state, triggering the third warning command and generating the third strategy: start the filter media drying program, perform heating and air drying, increase the inlet air temperature and turn on the heating element to evaporate the moisture on the surface of the filter media; start the low humidity circulation air, introduce low humidity air circulation to assist in the discharge of moisture; the time is controlled to last for 10-30 minutes; and establish a humidification event record dataset, linking the humidification status and corresponding operating data into the database.

[0039] Preferably, the health prediction module is used to construct an initial model of filter health degradation using a convolutional neural network (CNN), and to train and test the initial model of filter health degradation using a blockage event record dataset, an offset event record dataset, and a humidification event record dataset. The trained model is then used as a dynamic adjustment model for filter health. Simultaneously, the intermediate layer output of each event record data is extracted as a feature vector to identify key features of the operating status. The feature vector is then input again into the dynamic adjustment model of filter health for training and optimization, resulting in a trained health degradation prediction model. This model is used to predict the lifespan of the filter media and output a recommended replacement date range.

[0040] This invention provides an intelligent gas monitoring sensor and filtration alarm platform system. It has the following beneficial effects:

[0041] (1) This intelligent gas sensor and filter alarm platform system realizes the visualization monitoring of gas diffusion, thermal and humid environment, flow resistance and pressure difference distribution of air filter and air inlet and outlet channels by multi-source data acquisition and array sensor inversion reconstruction of two-dimensional gas flow field image, providing an intuitive and accurate data basis for subsequent blockage, displacement and humidification status assessment.

[0042] (2) This intelligent gas monitoring sensor and filter alarm platform system can quantify the blockage of the filter main channel, structural displacement and filter material moisture status by calculating the flow field blockage index FCI, flow field displacement index LOI and turbulence humidification index TWI, realize automatic early warning and strategy generation of abnormal operating status, and improve the system response speed and accuracy.

[0043] (3) This intelligent gas monitoring sensor and filter alarm platform system, combined with the fan performance characteristic curve and local flow analysis results, can automatically adjust the fan output or provide installation and calibration suggestions when blockage or offset risks are detected, thereby realizing proactive operation and maintenance and operation optimization of the air filtration system and reducing the cost of manual intervention.

[0044] (4) The intelligent gas monitoring sensor and filter alarm platform system, based on the health decay prediction model constructed by the convolutional neural network, can integrate blockage, offset and humidification event data, predict the future life of filter media and output the replacement suggestion date range, provide a scientific basis for operation and maintenance decision-making, and improve the reliability and safety of the filtration system. Attached Figure Description

[0045] Figure 1 This is a block diagram and flowchart of an intelligent gas monitoring sensor and filtration alarm platform system according to the present invention. Detailed Implementation

[0046] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example 1

[0048] Please see Figure 1 This invention provides an intelligent gas monitoring sensor and filtration alarm platform system, comprising:

[0049] The data acquisition module is used to comprehensively monitor the gas diffusion, thermal and humidity environment, flow resistance, pressure difference changes, and operating status of the air filters and their inlet and outlet air ducts in the factory workshop; it collects the spatial gas concentration sequence Ci(t), temperature distribution sequence Ti(t), humidity distribution sequence Hi(t), spatial pressure difference distribution sequence ΔPi(t), air volume sequence Fa(t), and speed sequence Sa(t).

[0050] The flow field imaging module is used to unify the format, synchronize the time, align the space and remove anomalies from the data collected from multiple sources, and generate a standardized multi-parameter dataset; then, the two-dimensional gas flow field distribution image F(x,y) of the filter working area is jointly reconstructed through array sensor inversion algorithm.

[0051] The flow field blockage monitoring module is used to extract the main channel width Wc, local average velocity Vc, and pressure difference balance index ΔPvar based on the two-dimensional gas flow field distribution image F(x,y), calculate the filter blockage index FCI, and compare it with the filter blockage threshold Fth to determine whether the filter is at risk of blockage. If it is, an appropriate strategy is given, and a blockage event record dataset is established.

[0052] The flow field offset monitoring module is used to obtain the global centroid coordinates (xc, yc) and local flow distribution imbalance coefficient Qvar of the flow field based on the two-dimensional gas flow field distribution image F(x, y). Combined with the reference calibration coordinates (x0, y0), the flow field offset index LOI is calculated and compared with the flow field offset threshold Lth to determine whether the current filter installation position is reasonable. If it is not reasonable, an appropriate strategy is given and an offset event record dataset is established.

[0053] The turbulence humidification monitoring module is used to obtain the turbulent kinetic energy Ek and the spatial humidity distribution gradient based on the two-dimensional gas flow field distribution image F(x,y). The turbulent humidification index TWI is calculated by combining the local concentration fluctuation RMS value Crms with the humidification threshold Tth to determine whether the filter material of the air filter is damp in the current operating state. If it is damp, an appropriate strategy is given and a humidification event record dataset is established.

[0054] The health prediction module is used to build an initial model of filter health degradation using a convolutional neural network (CNN). It trains the model on data of blockage, offset, and humidification events to build a health degradation prediction model, predict the lifespan of the filter media, and output a recommended replacement date range.

[0055] In this embodiment, through multi-module collaborative monitoring and analysis, a comprehensive perception of the gas diffusion, thermal and humid environment, flow resistance, pressure difference changes, and operating status of the air filter and inlet / outlet air ducts is achieved. Combined with the quantitative assessment of flow field blockage, deviation, and turbulence humidification index and the health prediction of convolutional neural networks, the operating status of the filter can be accurately determined, potential risks can be identified in advance, and filter media replacement suggestions can be provided, enabling proactive operation and maintenance and scientific decision-making, thereby improving the reliability and safety of the filtration system.

[0056] Example 2

[0057] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, the data acquisition module includes a gas concentration acquisition unit, an environmental parameter acquisition unit, a pressure difference distribution acquisition unit, and a pneumatic equipment operating status acquisition unit;

[0058] The gas concentration acquisition unit is used to monitor the gas diffusion status within air filters and their inlet and outlet air ducts installed in the factory workshop environment. By deploying a multi-point miniature gas sensor array, a sampling synchronization triggering device, and a spatial node calibration device, it performs time-series gas concentration acquisition and spatial index calibration at each sampling location, acquiring the spatial gas concentration sequence Ci(t) for each sensor node. Combining a multi-point concentration difference algorithm and a spatial gradient reconstruction model, it performs local gradient calculation on the continuous sampling results to obtain the gas concentration gradient. ;

[0059] The environmental parameter acquisition unit is used to monitor the thermal and humidity state of the operating environment of the air filter and its impact on flow field imaging. By deploying temperature sensing nodes, humidity sensing nodes, and thermal-humidity coupling sampling devices, it monitors the inlet temperature field, filter material surface temperature distribution, and temperature changes in the clean area after filtration in real time, and acquires the temperature distribution sequence Ti(t). Through environmental humidity probes, miniature dew point sensors, and filter material moisture content change analysis equipment, it monitors the front-end humidity field of the filter, the moisture absorption state of the filter material, and downstream water vapor residue in real time, and acquires the humidity distribution sequence Hi(t).

[0060] The differential pressure distribution acquisition unit is used to monitor the flow resistance characteristics and differential pressure change behavior of the air filter under different operating conditions. By installing differential pressure sensors and miniature differential pressure probes at the air inlet, filter media cavity and air outlet, and combining differential pressure fluctuation analysis algorithms, the pressure changes before and after the filter are monitored in real time to obtain the spatial differential pressure distribution sequence ΔPi(t).

[0061] The pneumatic equipment operation status acquisition unit is used to monitor the operation status of the blower and air pump that are matched with the filtration system and evaluate their impact on the flow field structure and filter material load. Through the air volume meter, speed sensor and operating power acquisition device, the output air volume and speed of the blower and air pump are monitored in real time, and the air volume sequence Fa(t) and speed sequence Sa(t) are acquired.

[0062] In this embodiment, through multi-unit collaborative data acquisition, comprehensive and real-time monitoring of gas concentration, ambient temperature and humidity, pressure differential distribution, and operating status of supporting pneumatic equipment in air filters and their inlet and outlet air channels is achieved. This enables the acquisition of high-precision, multi-dimensional operating parameters, providing a reliable data foundation for subsequent flow field imaging, blockage offset analysis, and health prediction, thereby improving the accuracy and traceability of system monitoring.

[0063] Example 3

[0064] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, the flow field imaging module includes a data preprocessing unit and a flow field imaging reconstruction unit;

[0065] The data preprocessing unit is used to process the gas concentration sequence Ci(t) and gas concentration gradient using a multi-source data format unified algorithm. The temperature distribution sequence Ti(t), humidity distribution sequence Hi(t), pressure difference distribution sequence ΔPi(t), air volume sequence Fa(t), and speed sequence Sa(t) are formatted uniformly; time synchronization correction is performed on each sequence using node timestamp calibration technology; spatial alignment and interpolation are performed on the data of each sampling point using spatial coordinate interpolation method; anomaly detection and rejection algorithms are adopted to automatically reject instantaneous impulse noise, sensor drift points, and abnormal data that do not conform to physical laws, and a standardized multi-parameter dataset is established.

[0066] The flow field imaging reconstruction unit is used to construct a gas flow parameter matrix and a spatial node observation matrix, and then use the array inverse solver algorithm based on array sensors to jointly invert the spatial concentration field, temperature and humidity field and pressure difference field in the standardized multi-parameter dataset to reconstruct the two-dimensional gas flow field distribution image F(x, y) of the filter working area.

[0067] In this embodiment, the flow field imaging module performs standardized preprocessing and joint inversion reconstruction on the multi-source acquired data, realizing high-precision visualization of the two-dimensional gas flow field in the working area of ​​the air filter. This not only eliminates abnormal data interference and temporal and spatial deviations, but also provides an accurate flow field basis for subsequent analysis of blockage, offset and humidification status, thereby improving the reliability and accuracy of monitoring.

[0068] The flow field blockage monitoring module includes a flow field feature extraction unit, a first calculation unit, and a first analysis unit;

[0069] The flow field feature extraction unit is used to perform structured identification of the dominant flow path based on the two-dimensional gas flow field distribution image F(x, y) using a streamline connected domain search algorithm and channel skeleton extraction technology. Then, a cross-sectional flow distribution analysis method is used to calculate the cross-sectional flow contour lines of the streamline connected domains and skeleton centerlines, quantifying the flow distribution characteristics within the steady-state region, calculating the average effective width of the main channel, and obtaining the main channel width Wc. Based on the concentration distribution change rate in the two-dimensional gas flow field distribution image F(x, y), combined with the air volume sequence Fa(t) and the rotational speed sequence Sa(t), a flow field concentration gradient is used... The Gradient-FlowrateVelocityInference algorithm is used to calculate the instantaneous velocity field of the main channel and its surrounding local area. Then, the local average velocity Vc is obtained through regional averaging filtering and time sliding window smoothing. Based on the two-dimensional gas flow field distribution image F(x,y), the flow field is divided into spatial grids using a uniform grid division method, and the pressure difference distribution data ΔPi(t) in each spatial grid is mapped. Then, the regional pressure difference statistical analysis method and variance normalization calculation technique are used to perform statistical and variance normalization processing on the pressure difference in each spatial grid to obtain the pressure difference balance index ΔPvar.

[0070] In this embodiment, the flow field blockage monitoring module extracts structured features, calculates flow velocity, and analyzes pressure differential balance in the two-dimensional gas flow field. This enables accurate quantification of the filter's main channel width, local flow velocity, and pressure differential distribution characteristics, thereby achieving early identification and accurate assessment of filter blockage risks and improving the operational safety and maintenance efficiency of the filtration system.

[0071] Example 5

[0072] This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 1 Specifically, the first calculation unit is used to calculate the filter clogging index FCI after dimensionless processing of the obtained main channel width Wc, local average flow velocity Vc, and pressure difference balance index ΔPvar, as follows:

[0073]

[0074] In the formula, w1, w2, and w3 represent weighting coefficients, and W0 represents the calibration channel width, which is measured under normal, unobstructed conditions during equipment calibration. The reference value representing the pressure difference is derived from the pressure difference variance measured during equipment calibration under normal, unblocked conditions. V0 represents the reference flow velocity, which is measured during equipment calibration under normal operating conditions.

[0075] The main channel width Wc represents the impact of the filter clogging risk. It has a high weight and is a key indicator, reflecting the contribution of the channel structure integrity to the gas flow capacity.

[0076] The pressure differential uniformity index ΔPvar has a medium weighting and represents the impact of the uniformity of the overall filter resistance distribution on the flow field stability.

[0077] : Characterizes the impact of the local average flow velocity Vc on the filter clogging risk, with the second highest weight, reflecting the effect of changes in the flow velocity in the main channel and surrounding areas on clogging sensitivity;

[0078] FCI quantifies flow field obstruction by weighted fusion of main channel width, local flow velocity, and pressure differential uniformity. Narrowing of the main channel indicates airflow obstruction, uneven pressure differential distribution indicates increased local resistance, and decreased flow velocity indicates local blockage. The three factors combined reflect the overall risk of filter blockage.

[0079] The first analysis unit is used to obtain a first evaluation result by comparing the filter blocking index Fth with a preset filter blocking threshold Fth, and then comparing the filter blocking degree index Fth with the filter blocking threshold Fth.

[0080] When the filter clogging index FCI is less than the filter clogging threshold Fth, it is determined that there is no risk of filter clogging, and the filter is maintained in normal operation and continuously monitored.

[0081] When the filter clogging index FCI is greater than or equal to the filter clogging threshold Fth, the filter is determined to be at risk of clogging, triggering the first warning command and generating the first strategy: automatically increasing the fan output power by 20% to increase the average flow velocity in the main channel by 3%-10% to maintain the target air supply volume; and establishing a clogging event record dataset, linking the filter clogging index FCI, fan adjustment parameters and current operating status into the database.

[0082] The filter clogging threshold Fth is obtained through statistical analysis of a large amount of air filter flow field characteristic data under normal operating conditions and different degrees of clogging. Specifically, this includes statistically analyzing the distribution range of the main channel width Wc, the local average flow velocity Vc, and the pressure differential balance index ΔPvar. Combined with equipment calibration data and historical maintenance records, the clogging coefficient variation range between unclogging and initial clogging states is extracted. Through the experience of professional technicians and with reference to performance parameters provided by air filter manufacturers and industry standards, a reasonable threshold Fth is ultimately determined to effectively distinguish between normal filter operation and clogging risk states, ensuring the safe and stable operation of the air filtration system.

[0083] In this embodiment, by calculating the filter clogging index (FCI) and comparing it with a preset threshold, a quantitative assessment and intelligent early warning of the filter clogging risk can be achieved. The fan output can be automatically adjusted to maintain the target air supply, thereby ensuring the stable operation of the filtration system, reducing manual intervention, and improving operating efficiency and reliability.

[0084] Example 6

[0085] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, the flow field migration monitoring module includes a migration feature extraction unit, a second calculation unit, and a second analysis unit;

[0086] The offset feature extraction unit is used to obtain the global centroid coordinates (xc, yc) of the flow field based on the two-dimensional gas flow field distribution image F(x, y) by weighted integration of the flow intensity distribution of each spatial node in F(x, y) using the global centroid calculation method; it also obtains the benchmark calibration coordinates (x0, y0) through equipment installation and calibration records; it establishes a fan performance characteristic curve model and constructs a flow-speed mapping relationship based on the air volume sequence Fa(t) and the speed sequence Sa(t); it estimates the instantaneous air intake using a multi-parameter fitting inversion method and obtains the local flow distribution results through time-series sliding window smoothing; and it combines the flow distribution gradient change reflected in the two-dimensional gas flow field distribution image F(x, y) with the local flow fluctuation statistical algorithm to jointly quantify the flow dispersion and fluctuation amplitude in the spatial region to obtain the local flow distribution unevenness coefficient Qvar.

[0087] In this embodiment, by calculating the global centroid coordinates of the flow field and the local flow imbalance coefficient Qvar, accurate monitoring of filter installation offset is achieved. This enables timely identification of structural offset risks and provides calibration strategies, thereby effectively preventing gas leakage and improving the safety and reliability of system operation.

[0088] Example 7

[0089] This embodiment is an explanation based on Embodiment 6. Please refer to it. Figure 1 Specifically, the second calculation unit is used to calculate the flow field offset index LOI after dimensionless processing of the obtained global centroid coordinates (xc, yc), reference calibration coordinates (x0, y0), and local flow distribution unevenness coefficient Qvar. The formula is as follows:

[0090]

[0091] In the formula, a1 and a2 represent weighting coefficients; xc represents the global centroid coordinates of the inverted flow field image in the x direction, yc represents the global centroid coordinates of the inverted flow field image in the y direction, x0 represents the x-direction coordinates of the flow field reference center under calibration conditions, and y0 represents the y-direction coordinates of the flow field reference center under calibration conditions.

[0092] The centroid offset distance of the flow field has a high weighting and is a key indicator, reflecting the effect of the filter installation location on the flow field uniformity.

[0093] Qvar, which characterizes the impact of local flow distribution unevenness on gas leakage risk, has the second highest weight and reflects the effect of local flow fluctuations on the overall offset sensitivity.

[0094] LOI quantifies the rationality of filter installation location by weighted fusion of global centroid offset and local flow imbalance. Centroid offset represents the overall flow field shift, and Qvar represents local flow fluctuations. Joint evaluation can determine potential gas leakage risks.

[0095] The second analysis unit is used to obtain a second evaluation result by comparing the flow field offset index (LOI) with the flow field offset threshold (Lth) using a preset flow field offset threshold (Lth):

[0096] When the flow field offset index LOI is less than the flow field offset threshold Lth, it indicates that the current filter installation position is reasonable and there is no risk of gas leakage due to structural offset. Continuous monitoring is required.

[0097] When the flow field offset index LOI is greater than or equal to the flow field offset threshold Lth, it indicates that the current filter installation position is unreasonable and there is a risk of gas leakage due to structural offset. This triggers a second warning instruction and generates a second strategy: generating a filter installation position calibration suggestion, prompting the inspection of the filter frame and sealing strip; analyzing the offset direction, automatically marking the leaking spatial area, and initiating a secondary local imaging process to reconstruct the local area in the offset direction, improving the accuracy of leak point identification; and establishing an offset event record dataset, linking the flow field offset index LOI, local flow imbalance Qvar, and calibration suggestion into the database.

[0098] The flow field offset threshold Lth is obtained through statistical analysis of a large amount of filter installation and operation data. Specifically, this includes analyzing the distribution range of the global centroid coordinates (xc, yc) and the local flow distribution imbalance coefficient Qvar under normal installation and structural offset conditions. Combining the baseline calibration coordinates (x0, y0) under equipment calibration conditions and historical installation calibration records, and through the experience of professional technicians, and referring to the filter manufacturer's installation specifications and industry standards, a reasonable flow field offset judgment threshold Lth is determined. This threshold effectively distinguishes the rationality of filter installation locations and reduces the risk of gas leakage due to structural offset.

[0099] In this embodiment, by calculating the flow field offset index (LOI) and comparing it with a preset threshold, it is possible to accurately determine whether the filter installation position is reasonable. When the risk of offset exists, an early warning is triggered in time and calibration suggestions are provided, thereby achieving proactive prevention and control of gas leakage risks and improving the safety and maintenance efficiency of system operation.

[0100] Example 8

[0101] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, the turbulence humidification monitoring module includes a turbulence humidification feature extraction unit, a third calculation unit, and a third analysis unit;

[0102] The turbulent humidification feature extraction unit is used to obtain the turbulent kinetic energy Ek of the flow field by weighting the velocity change rate of each spatial node based on the two-dimensional gas flow field distribution image F(x, y) and the flow field velocity gradient calculation method; based on the humidity distribution sequence Hi(t), it uses a gradient reconstruction algorithm to perform spatial difference and interpolation processing on the humidity data of each sampling point, calculates the humidity change rate of each spatial node, and obtains the spatial humidity distribution gradient. Based on the gas concentration sequence Ci(t), the time-series fluctuation analysis method is used to calculate the root mean square of the concentration time series at each sampling location and obtain the local concentration fluctuation RMS value Crms.

[0103] In this embodiment, by comprehensively extracting the turbulent kinetic energy, spatial humidity gradient, and local concentration fluctuations of the two-dimensional gas flow field, the moisture status of the filter material during filter operation can be monitored in real time, enabling early identification and proactive intervention of humidification risks, thereby effectively extending the service life of the filter material and ensuring air purification efficiency.

[0104] Example 9

[0105] This embodiment is an explanation based on Embodiment 8. Please refer to it. Figure 1Specifically, the third calculation unit is used to obtain the turbulent kinetic energy Ek of the flow field and the spatial humidity distribution gradient. The turbulent humidification index (TWI) is calculated by dimensionless processing of the local concentration fluctuation RMS value (Crms) and the turbulent humidification index (TWI), as shown in the following formula:

[0106]

[0107] In the formula, s1, s2 and s3 represent weighting coefficients, Eref represents the reference value of turbulent kinetic energy of the flow field, which is measured under normal drying conditions during equipment calibration, Href represents the reference value of humidity gradient, which is measured during equipment calibration, and Cref represents the reference value of concentration fluctuation, which is measured during equipment calibration.

[0108] The turbulent kinetic energy Ek of the flow field has a high weighting and is a key indicator that reflects the contribution of turbulence intensity to moisture mixing and local humidification.

[0109] Characterizing the gradient of spatial humidity distribution The impact on the risk of moisture absorption of filter media accounts for the second highest weight, reflecting the effect of uneven distribution of moisture on the surface of filter media on moisture sensitivity;

[0110] The RMS value Crms, which characterizes the impact of local concentration fluctuations on the risk of moisture absorption of filter media, has a medium weight and reflects the gas mixing and local moisture accumulation.

[0111] TWI (Total Moisture Intake) quantifies the risk of filter media becoming damp by weighted fusion of turbulent kinetic energy, humidity gradient, and concentration fluctuation. High turbulent kinetic energy indicates strong air disturbance, high humidity gradient indicates concentrated moisture, and high local concentration fluctuation indicates significant local humidification. The three factors are combined to assess the moisture status of the filter media.

[0112] The third analysis unit is used to obtain a third evaluation result by comparing the turbulent humidification index TWI with the humidification threshold Tth using a preset humidification threshold Tth.

[0113] When the turbulent humidification index TWI ≤ humidification threshold Tth, it indicates that the filter media of the air filter is not damp and is maintaining normal operation, requiring continuous monitoring.

[0114] When the turbulent humidification index TWI > humidification threshold Tth, it indicates that the filter media of the air filter is damp in its current operating state, triggering the third warning command and generating the third strategy: start the filter media drying program, perform heating and air drying, increase the inlet air temperature and turn on the heating element to evaporate the moisture on the surface of the filter media; start the low humidity circulation air, introduce low humidity air circulation to assist in the discharge of moisture; the time is controlled to last for 10-30 minutes; and establish a humidification event record dataset, linking the humidification status and corresponding operating data into the database.

[0115] The humidification threshold Tth is obtained by analyzing the turbulent kinetic energy Ek and the spatial humidity distribution gradient of a large number of air filters under different ambient humidity and operating conditions. Statistical analysis of local concentration fluctuation RMS values ​​(Crms) was conducted. By extracting the distribution range of the turbulent humidification index (TWI) under wet and dry filter media conditions, combined with the experience judgment of professional technicians, and referring to the moisture absorption performance parameters of filter media provided by air filter manufacturers and industry operating standards, a reasonable humidification judgment threshold (Tth) was finally determined to accurately identify the moisture status of the filter media and ensure the operating efficiency and service life of the filter.

[0116] In this embodiment, the turbulent humidification index is calculated by integrating turbulent kinetic energy, spatial humidity gradient, and local concentration fluctuations, so as to achieve accurate assessment of the moisture status of the filter material. It can automatically trigger drying and low humidity circulation strategies in the early stage of humidification, effectively prevent the performance of the filter material from deteriorating, ensure air filtration efficiency, and extend the service life of the filter material.

[0117] Example 10

[0118] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, the health prediction module uses a convolutional neural network (CNN) to construct an initial model for filter health degradation. This initial model is then trained and tested using a clogging event record dataset, an offset event record dataset, and a humidification event record dataset. The trained model serves as the filter health dynamic adjustment model. Simultaneously, the intermediate layer output of each event record data is extracted as a feature vector to identify key features of the operating state. This feature vector is then input again into the filter health dynamic adjustment model for training and optimization, resulting in a trained health degradation prediction model. This model predicts filter media lifespan and outputs a recommended replacement date range.

[0119] In this embodiment, a convolutional neural network is used to jointly train and optimize the data on blockage, offset, and humidification events, thereby enabling dynamic prediction of the health status of the air filter. This allows for accurate prediction of the filter material degradation trend and remaining lifespan, providing scientific replacement recommendations, improving maintenance efficiency, and reducing operating costs.

[0120] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0121] The above formulas are all derived from software simulation using a large amount of data and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the protection scope of the present invention.

Claims

1. An intelligent monitoring gas sensor and filtration alert platform system, characterized by, include: The data acquisition module is used to comprehensively monitor the gas diffusion, thermal and humidity environment, flow resistance, pressure difference changes, and operating status of the air filters and their inlet and outlet air ducts in the factory workshop; it collects the spatial gas concentration sequence Ci(t), temperature distribution sequence Ti(t), humidity distribution sequence Hi(t), spatial pressure difference distribution sequence ΔPi(t), air volume sequence Fa(t), and speed sequence Sa(t). The flow field imaging module is used to unify the format, synchronize the time, align the space and remove anomalies from the data collected from multiple sources, and generate a standardized multi-parameter dataset; then, the two-dimensional gas flow field distribution image F(x,y) of the filter working area is jointly reconstructed through array sensor inversion algorithm. The flow field blockage monitoring module is used to extract the main channel width Wc, local average velocity Vc, and pressure difference balance index ΔPvar based on the two-dimensional gas flow field distribution image F(x,y), calculate the filter blockage index FCI, and compare it with the filter blockage threshold Fth to determine whether the filter is at risk of blockage. If it is, an appropriate strategy is given, and a blockage event record dataset is established. The flow field offset monitoring module is used to obtain the global centroid coordinates (xc, yc) and local flow distribution imbalance coefficient Qvar of the flow field based on the two-dimensional gas flow field distribution image F(x, y). Combined with the reference calibration coordinates (x0, y0), the flow field offset index LOI is calculated and compared with the flow field offset threshold Lth to determine whether the current filter installation position is reasonable. If it is not reasonable, an appropriate strategy is given and an offset event record dataset is established. A turbulent humidification monitoring module is used to obtain a flow field turbulent kinetic energy Ek, a spatial humidity distribution gradient and a local concentration fluctuation RMS value Crms based on a two-dimensional gas flow field distribution image F(x, y), calculate a turbulent humidification index TWI, and compare it with a humidification threshold Tth to determine whether the filter material of the current operating state of the air filter is humidified, and if so, give corresponding strategies and establish a humidification event record data set. The health prediction module is used to build an initial model of filter health degradation using a convolutional neural network (CNN). It trains the model on data of blockage, offset, and humidification events to build a health degradation prediction model, predict the lifespan of the filter media, and output a recommended replacement date range.

2. The intelligent monitoring gas sensor and filtering alarm platform system according to claim 1, characterized in that, The data acquisition module includes a gas concentration acquisition unit, an environmental parameter acquisition unit, a pressure difference distribution acquisition unit, and a pneumatic equipment operating status acquisition unit. The gas concentration acquisition unit is used for monitoring the gas diffusion condition in the air filter installed in the factory workshop environment and the air inlet and outlet channels thereof; through the arrangement of a multi-point micro gas sensor array, a sampling synchronous trigger device and a space node calibration device, the gas concentration time sequence acquisition and space index calibration are performed on each sampling position, and the space gas concentration sequence Ci(t) of each sensing node is acquired; in combination with a multi-point concentration difference algorithm and a space gradient reconstruction model, the local gradient is calculated based on the continuous sampling result, and the gas concentration gradient is obtained. The environmental parameter acquisition unit is used to monitor the thermal and humidity state of the operating environment of the air filter and its impact on flow field imaging. By deploying temperature sensing nodes, humidity sensing nodes, and thermal-humidity coupling sampling devices, it monitors the inlet temperature field, filter material surface temperature distribution, and temperature changes in the clean area after filtration in real time, and acquires the temperature distribution sequence Ti(t). Through environmental humidity probes, miniature dew point sensors, and filter material moisture content change analysis equipment, it monitors the front-end humidity field of the filter, the moisture absorption state of the filter material, and downstream water vapor residue in real time, and acquires the humidity distribution sequence Hi(t). The differential pressure distribution acquisition unit is used to monitor the flow resistance characteristics and differential pressure change behavior of the air filter under different operating conditions. By installing differential pressure sensors and miniature differential pressure probes at the air inlet, filter media cavity and air outlet, and combining differential pressure fluctuation analysis algorithms, the pressure changes before and after the filter are monitored in real time to obtain the spatial differential pressure distribution sequence ΔPi(t). The pneumatic equipment operation status acquisition unit is used to monitor the operation status of the blower and air pump that are matched with the filtration system and evaluate their impact on the flow field structure and filter material load. Through the air volume meter, speed sensor and operating power acquisition device, the output air volume and speed of the blower and air pump are monitored in real time, and the air volume sequence Fa(t) and speed sequence Sa(t) are acquired. 3.The intelligent monitoring gas sensor and filtering alarm platform system of claim 1, wherein, The flow field imaging module includes a data preprocessing unit and a flow field imaging reconstruction unit. The data preprocessing unit is configured to perform format unification processing on the gas concentration sequence Ci(t), the gas concentration gradient sequence ΔPi(t) and the air volume sequence Fa(t) and the rotation speed sequence Sa(t) by using a multi-source data format unification algorithm; perform time synchronization correction on each sequence by using a node timestamp calibration technique; and perform spatial alignment and interpolation processing on data of each sampling point by using a spatial coordinate interpolation method. Anomaly detection and elimination algorithms are used to automatically eliminate transient impulse noise, sensor drift points, and abnormal data that do not conform to physical laws, and to establish a standardized multi-parameter dataset. The flow field imaging reconstruction unit is used to construct a gas flow parameter matrix and a spatial node observation matrix, and then use the array inverse solver algorithm based on array sensors to jointly invert the spatial concentration field, temperature and humidity field and pressure difference field in the standardized multi-parameter dataset to reconstruct the two-dimensional gas flow field distribution image F(x, y) of the filter working area. 4.The intelligent monitoring gas sensor and filtering alarm platform system of claim 1, wherein, The flow field blockage monitoring module includes a flow field feature extraction unit, a first calculation unit, and a first analysis unit; The flow field feature extraction unit is used to perform structured identification of the dominant flow path based on the two-dimensional gas flow field distribution image F(x, y) using a streamline connected domain search algorithm and channel skeleton extraction technology. Then, a cross-sectional flow distribution analysis method is used to calculate the cross-sectional flow contour lines of the streamline connected domains and skeleton centerlines, quantifying the flow distribution characteristics within the steady-state region, calculating the average effective width of the main channel, and obtaining the main channel width Wc. Based on the concentration distribution change rate in the two-dimensional gas flow field distribution image F(x, y), combined with the air volume sequence Fa(t) and the rotational speed sequence Sa(t), a flow field concentration gradient is used... The Gradient-FlowrateVelocityInference algorithm is used to calculate the instantaneous velocity field of the main channel and its surrounding local area. Then, the local average velocity Vc is obtained through regional averaging filtering and time sliding window smoothing. Based on the two-dimensional gas flow field distribution image F(x,y), the flow field is divided into spatial grids using a uniform grid division method, and the pressure difference distribution data ΔPi(t) in each spatial grid is mapped. Then, the regional pressure difference statistical analysis method and variance normalization calculation technique are used to perform statistical and variance normalization processing on the pressure difference in each spatial grid to obtain the pressure difference balance index ΔPvar.

5. The intelligent monitoring gas sensor and filtering alarm platform system according to claim 4, wherein, The first calculation unit is used to calculate the filter blockage index FCI by obtaining the main channel width Wc, the local average flow velocity Vc, and the pressure difference balance index ΔPvar, and after dimensionless processing. The first analysis unit is used to obtain a first evaluation result by comparing the filter blocking index Fth with a preset filter blocking threshold Fth, and then comparing the filter blocking degree index Fth with the filter blocking threshold Fth. When the filter clogging index FCI is less than the filter clogging threshold Fth, it is determined that there is no risk of filter clogging, and the filter is maintained in normal operation and continuously monitored. When the filter clogging index FCI is greater than or equal to the filter clogging threshold Fth, the filter is determined to be at risk of clogging, triggering the first warning command and generating the first strategy: automatically increasing the fan output power by 20% to increase the average flow velocity in the main channel by 3%-10% to maintain the target air supply volume; and establishing a clogging event record dataset, linking the filter clogging index FCI, fan adjustment parameters and current operating status into the database.

6. The intelligent gas monitoring sensor and filtration alarm platform system according to claim 1, characterized in that, The flow field migration monitoring module includes a migration feature extraction unit, a second calculation unit, and a second analysis unit; The offset feature extraction unit is used to obtain the global centroid coordinates (xc, yc) of the flow field based on the two-dimensional gas flow field distribution image F(x, y) by weighted integration of the flow intensity distribution of each spatial node in F(x, y) using the global centroid calculation method; it also obtains the benchmark calibration coordinates (x0, y0) through equipment installation and calibration records; it establishes a fan performance characteristic curve model and constructs a flow-speed mapping relationship based on the air volume sequence Fa(t) and the speed sequence Sa(t); it estimates the instantaneous air intake using a multi-parameter fitting inversion method and obtains the local flow distribution results through time-series sliding window smoothing; and it combines the flow distribution gradient change reflected in the two-dimensional gas flow field distribution image F(x, y) with the local flow fluctuation statistical algorithm to jointly quantify the flow dispersion and fluctuation amplitude in the spatial region to obtain the local flow distribution unevenness coefficient Qvar.

7. The intelligent monitoring gas sensor and filtering alarm platform system according to claim 6, wherein, The second calculation unit is used to calculate the flow field offset index LOI after dimensionless processing of the obtained global centroid coordinates (xc, yc), reference calibration coordinates (x0, y0) and local flow distribution imbalance coefficient Qvar. The second analysis unit is used to obtain a second evaluation result by comparing the flow field offset index (LOI) with the flow field offset threshold (Lth) using a preset flow field offset threshold (Lth): When the flow field offset index LOI < the flow field offset threshold Lth, it indicates that the current filter installation position is reasonable and there is no risk of gas leakage due to structural offset. Continuous monitoring is required. When the flow field offset index LOI is greater than or equal to the flow field offset threshold Lth, it indicates that the current filter installation position is unreasonable and there is a risk of gas leakage due to structural offset. This triggers a second warning instruction and generates a second strategy: generating a filter installation position calibration suggestion, prompting the inspection of the filter frame and sealing strip; analyzing the offset direction, automatically marking the leaking spatial area, and initiating a secondary local imaging process to reconstruct the local area in the offset direction, improving the accuracy of leak point identification; and establishing an offset event record dataset, linking the flow field offset index LOI, local flow imbalance Qvar, and calibration suggestion into the database. 8.The intelligent monitoring gas sensor and filtering alarm platform system of claim 1, wherein, The turbulence humidification monitoring module includes a turbulence humidification feature extraction unit, a third calculation unit, and a third analysis unit; The turbulent humidification feature extraction unit is used to obtain the turbulent kinetic energy Ek of the flow field by weighting the velocity change rate of each spatial node based on the two-dimensional gas flow field distribution image F(x, y) and the flow field velocity gradient calculation method; based on the humidity distribution sequence Hi(t), it uses a gradient reconstruction algorithm to perform spatial difference and interpolation processing on the humidity data of each sampling point, calculates the humidity change rate of each spatial node, and obtains the spatial humidity distribution gradient. Based on the gas concentration sequence Ci(t), the time-series fluctuation analysis method is used to calculate the root mean square of the concentration time series at each sampling location and obtain the local concentration fluctuation RMS value Crms.

9. The intelligent gas monitoring sensor and filtration alarm platform system according to claim 8, characterized in that, The third calculation unit is configured to calculate a turbulent wetting index TWI by processing the obtained flow field turbulent kinetic energy Ek, the spatial humidity distribution gradient and the local concentration fluctuation RMS value Crms, which are dimensionless. The third analysis unit is used to obtain a third evaluation result by comparing the turbulent humidification index TWI with the humidification threshold Tth using a preset humidification threshold Tth. When the turbulent humidification index TWI ≤ humidification threshold Tth, it indicates that the filter media of the air filter is not damp and is maintaining normal operation, requiring continuous monitoring. When the turbulent humidification index TWI > humidification threshold Tth, it indicates that the filter media of the air filter is damp in its current operating state, triggering the third warning command and generating the third strategy: start the filter media drying program, perform heating and air drying, increase the inlet air temperature and turn on the heating element to evaporate the moisture on the surface of the filter media; start the low humidity circulation air, introduce low humidity air circulation to assist in the discharge of moisture; the time is controlled to last for 10-30 minutes; and establish a humidification event record dataset, linking the humidification status and corresponding operating data into the database. 10.The intelligent monitoring gas sensor and filtering alarm platform system of claim 1, wherein, The health prediction module uses a convolutional neural network (CNN) to construct an initial model for filter health degradation. This initial model is then trained and tested using a clogging event record dataset, an offset event record dataset, and a humidification event record dataset. The trained model serves as the filter health dynamic adjustment model. Simultaneously, the intermediate layer outputs of each event record dataset are extracted as feature vectors to identify key features of the operating status. These feature vectors are then input again into the filter health dynamic adjustment model for training and optimization, resulting in a trained health degradation prediction model. This model predicts filter media lifespan and outputs a recommended replacement date range.

Images