Air filtering method, system, equipment and medium

Abstract

The invention is applicable to the technical field of air filtration, and discloses an air filtration method, system, equipment and medium, the method comprises the following steps: acquiring pollutants and operating parameters, and constructing pollutant state vectors; performing time window processing on the pollutant state vector to obtain a pollution trend structure vector; based on the pollution trend structure vector, a KNN classifier is adopted to construct a pollution trend identification model, and a trend label is obtained; an optimal filtering path combination and a current recommended wind speed are obtained by combining the pollutant state vector and the operation parameter through the trend label and are used for path switching and wind speed adjustment; after path switching and wind speed adjustment are completed, the blocking trend is judged according to the operation parameters; and based on the existence of the blockage trend, a blockage duration recording mechanism is adopted to evaluate the risk level, and a maintenance prompt is generated and executed, so that air filtration is realized, and the safety, stability and intelligent level of operation of the air filtration method are remarkably enhanced.

Description

Technical Field

[0001] This invention relates to the field of air filtration technology, and more particularly to an air filtration method, system, device, and medium. Background Technology

[0002] As urbanization progresses, air quality problems are becoming increasingly apparent, making air filtration technology crucial for ensuring environmental health. Currently, air purification systems in residences, factories, and smart buildings mostly use combined filter elements, such as coarse filters, HEPA high-efficiency filters, and activated carbon layers, relying on physical or chemical methods to remove pollutants such as particulate matter, volatile organic compounds (VOCs), and carbon dioxide from the air. System operation primarily relies on fans to drive airflow and control the filtration path; some devices also include differential pressure sensors and VOC sensors to monitor filter status and environmental changes. However, in terms of control algorithms, most still use fixed thresholds or simple PID control, passively reacting only when pollution levels exceed limits, unable to proactively adjust.

[0003] Existing technologies have two core problems. First, most systems only focus on the current pollutant concentration and lack the ability to predict pollution trends. For example, if particulate matter concentration keeps rising but hasn't reached the set limit, the system continues to operate at a low fan speed, only adjusting when the concentration exceeds the limit, leading to pollution diffusion and a delayed response. Second, judging filter status mostly relies on pressure difference measurement or estimated usage time, failing to combine fan control with filter clogging trends for prediction. Prolonged operation under high load can cause excessive filter aging and even ventilation failures. Especially in complex pollution scenarios, if pollution trends, pressure differences, and filtration path switching cannot be smoothly coordinated, the fan speed strategy and path structure will be mismatched, reducing system stability and wasting energy.

[0004] Therefore, it is necessary to use machine learning algorithms to build dynamic prediction models, rely on digital twin technology to simulate the system's response under different strategies, and then use optimization algorithms to give control instructions, thereby fundamentally improving the system's intelligence level and ensuring environmental health and operational efficiency. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides an air filtration method, system, device, and medium to solve the problem that existing air purification systems generally use the instantaneous value of pollutant concentration as the only input signal, and the system cannot identify the change pattern and make pre-adjustments in time before the pollution changes. Most filtration systems rely on pressure difference or time prediction to judge the filter element status, and in complex pollution scenarios, they cannot dynamically link pollution trends, pressure difference status, and filtration path switching logic.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an air filtration method, comprising: Acquire pollutants and operating parameters, and construct a pollutant state vector; By applying a time window to the pollutant state vector, a pollution trend structure vector is obtained. Based on the pollution trend structure vector, a pollution trend recognition model is constructed using a KNN classifier to obtain trend labels; By combining trend tags with pollutant state vectors and operating parameters, the optimal filtration path combination and the current recommended wind speed are obtained, which are used for path switching and wind speed adjustment. After completing the path switching and wind speed adjustment, the blockage trend is determined based on the operating parameters; Based on the presence of a blockage trend, a blockage duration recording mechanism is used to assess the risk level, generate and execute maintenance reminders, and complete air filtration.

[0008] As a preferred embodiment of the air filtration method described in this invention, the method includes: acquiring pollutants and operating parameters and constructing a pollutant state vector; and obtaining a pollution trend structure vector by performing time window processing on the pollutant state vector, including: The pollutant parameters, including particulate matter concentration, total volatile organic compounds and carbon dioxide concentration, are obtained and constituted as a pollutant state vector. The operating parameters, including the pressure difference across the filter element and the gas flow rate, are also obtained. A sliding time window is constructed, and differential coding is performed on each type of pollutant within the time window to obtain the trend change vectors of particulate matter concentration, total volatile organic compounds, and carbon dioxide concentration; The trend change vectors of each type of pollutant together constitute the pollution trend structure vector.

[0009] As a preferred embodiment of the air filtration method described in this invention, a pollution trend recognition model is constructed using a KNN classifier based on a pollution trend structure vector to obtain trend labels, including: By using historical monitoring data of different pollutants, a sliding window process is performed to extract differential coding trend features to form training samples and label them. A pollution trend recognition model is constructed using a lightweight K-nearest neighbor (KNN) classifier. Using the pollution trend structure vector as input to the pollution trend identification model, Euclidean distance and majority voting are used to obtain the trend classification confidence and trend label.

[0010] As a preferred embodiment of the air filtration method described in this invention, the optimal filtration path combination is obtained by combining trend tags with pollutant state vectors for path switching, including: Based on trend labels combined with pollutant state vectors, a path selection model based on rule trees is used to obtain the dominant pollutant type; Based on the trend label and the dominant pollutant type, combined with the trend classification confidence level, enter the path selection logic branch; By modeling the path selection operation as a path mapping function, the optimal combination of filtering paths is obtained for filter path switching.

[0011] As a preferred embodiment of the air filtration method described in this invention, the recommended wind speed is obtained by combining trend tags with operating parameters, and used to adjust the wind speed, including: By mapping trend labels to wind speed gain signals, introducing wind speed suppression factors, and combining them with the current filter pressure difference, the current recommended wind speed is calculated. The wind speed is adjusted by converting the current recommended wind speed into a control signal.

[0012] As a preferred embodiment of the air filtration method described in this invention, the following steps are taken: after completing path switching and adjusting the fan speed, the clogging trend is determined based on operating parameters, including: After completing path switching and wind speed adjustment, pressure difference and flow rate are continuously sampled, and the rate of change is constructed based on continuous time series. The blockage trend is determined by comparing the rate of change of pressure difference and flow rate with the blockage trend judgment threshold.

[0013] As a preferred embodiment of the air filtration method described in this invention, the method includes: assessing the risk level based on a clogging trend using a clogging duration recording mechanism, generating and executing a maintenance reminder, and completing the air filtration process, including: Based on the existence of a congestion trend, we enter the congestion trend growth zone and use a congestion duration recording mechanism to obtain the congestion duration. The risk level is assessed by combining the duration of blockage with preset thresholds, pressure differentials, and flow rates. Based on the assessed risk level, corresponding maintenance reminders are received and corresponding protective actions are executed. Record all operational data and status changes and build an operational log to complete air filtration.

[0014] In a second aspect, the present invention provides an air filtration system, comprising: The acquisition module is used to acquire pollutants and operating parameters and construct pollutant state vectors. The trend feature extraction module is used to obtain the pollution trend structure vector by performing time window processing on the pollutant state vector; The model building module is used to construct a pollution trend recognition model based on the pollution trend structure vector and a KNN classifier to obtain trend labels. The decision module is used to obtain the optimal filtration path combination and the current recommended wind speed by combining the pollutant state vector and operating parameters with trend labels, and is used for path switching and wind speed adjustment. The judgment module is used to determine the blockage trend based on the operating parameters after the path switching and wind speed adjustment are completed; The execution module is used to assess the risk level based on the presence of a blockage trend, using a blockage duration recording mechanism, generate maintenance reminders and execute them to complete air filtration.

[0015] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the air filtration method.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the air filtration method.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: By introducing pollution trend structure vector modeling and KNN trend recognition mechanism, accurate judgment of pollutant evolution trends is achieved, overcoming the lag problem of existing technologies relying on instantaneous concentration response. A filtration path optimization mechanism based on trend label and pollution type joint drive is constructed, linking fan speed adjustment and differential pressure monitoring, effectively improving adaptability and energy efficiency management capabilities for complex pollution scenarios. Through dual-parameter judgment of blockage trend and duration integration, proactive early warning and protection control of filter status are achieved, significantly enhancing the safety, stability and intelligence level of air filtration method operation. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the overall process of an air filtration method according to an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Example 1, referring to Figure 1 According to one embodiment of the present invention, an air filtration method is provided, comprising: S100: Acquire pollutants and operating parameters and construct a pollutant state vector; S200: By applying a time window to the pollutant state vector, a pollution trend structure vector is obtained; S300: Based on the pollution trend structure vector, a pollution trend recognition model is constructed using a KNN classifier to obtain trend labels; S400: By combining trend tags with pollutant state vectors and operating parameters, the optimal filtration path combination and the current recommended wind speed are obtained, which are used for path switching and wind speed adjustment. S500: Based on the completed path switching and wind speed adjustment, the blockage trend is determined according to the operating parameters; S600: Based on the presence of a clogging trend, a clogging duration recording mechanism is used to assess the risk level, generate and execute maintenance reminders, and complete air filtration.

[0022] It should be noted that air filtration systems operate in environments with coupled treatment of multiple pollutants. During operation, parameters such as pollutant type, concentration distribution, airflow velocity, and filter load change dynamically and significantly, causing nonlinear fluctuations in the system's filtration efficiency and ventilation resistance. Influenced by pollutant interactions, the effective adsorption capacity and penetration characteristics of the filter element also change in real time. Furthermore, the filtration system operates under variable loads and multiple conditions, making it difficult to assess the actual filtration status and remaining lifespan, resulting in performance warnings often being delayed. Simultaneously, due to continuous pollutant deposition and filter media fatigue, the filter element porosity continuously decreases, affecting system health. The variable airflow conditions and complex pollutant composition interfere with accurate status monitoring and frequent adjustments can lead to wear on actuators. Therefore, modeling and intelligently controlling the pollution evolution trend of the filtration system is particularly important.

[0023] Therefore, addressing the aforementioned issues of insufficient modeling capability for pollution evolution trends and lack of operational status-driven linkage response mechanisms, this paper introduces pollution trend structure vector modeling and KNN trend recognition mechanisms through steps S100-S600 to obtain trend labels, enabling path and wind speed control. It also provides preliminary judgment conditions for subsequent filter clogging assessment and fan energy-saving operation, improving the overall linkage efficiency of the system. A filter path optimization mechanism based on trend labels and pollution type is constructed, linking fan speed adjustment and differential pressure monitoring to achieve efficient adaptive control in steady-state and sudden pollution scenarios. By introducing a joint judgment mechanism of "differential pressure growth rate" and "airflow attenuation rate," dynamic monitoring of the air filtration path's operational status is achieved. Through dual-parameter judgment of clogging trends and duration integration, proactive early warning and protective control of filter status are realized, significantly enhancing the safety, stability, and intelligence level of the air filtration method.

[0024] Example 2, refer to Figure 1 As an embodiment of the present invention, an air filtration method is provided based on the above embodiment.

[0025] In this embodiment of the application, step S100, which involves acquiring pollutants and operating parameters and constructing a pollutant state vector, includes: Specifically, composite sensor modules are fixedly deployed at the air inlet and outlet of the air filtration equipment to monitor air conditions and operating status in an integrated manner.

[0026] During a fixed sampling period, the concentration of particulate matter (PM) is collected by a gas sensor at the air inlet. 2.5 The total volatile organic compounds (TVOC) and carbon dioxide concentration (CO2) are calculated and a pollutant state vector Q(t) is constructed. At the outlet end, the pressure difference before and after the filter element is collected by a differential pressure sensor, and the gas flow rate is measured in real time by a thermal flow meter.

[0027] All data is encapsulated into a unified state vector, specifically by combining the parameters of pollutant concentration, pressure difference, and flow rate collected in each sampling period according to the timestamp to form a state vector; In this embodiment of the application, step S200 involves processing the pollutant state vector using a time window to obtain a pollution trend structure vector, including: Specifically, a sliding time window of length w is selected, such as w=10 seconds. Starting from the current time t, w sampling points are traced back to construct the sliding time window. Within the time window, differential coding is performed on each type of pollutant to obtain the trend change vectors of particulate matter concentration, total volatile organic compounds (TVOC), and carbon dioxide concentration. in, It refers to the concentration of pollutants at time point t, such as particulate matter concentration, total volatile organic compounds (TVOC), and carbon dioxide concentration. The sampling period is the time interval between adjacent sampling times. The pollutants constructed at time t The trend change vector, where w is the number of sampling points currently used to calculate the pollutant trend.

[0028] Furthermore, based on the vectors of particulate matter concentration, total volatile organic compounds (TVOC), and carbon dioxide concentration, the pollution trend structure vector T(t) at the current moment is constructed.

[0029] It should be noted that the pollution trend structure vector T(t) comprehensively describes the rate of change and trend direction of the three key pollutants in the most recent w seconds. It is a dynamic modeling result of the pollution state, reflecting the "pollution evolution trend" rather than "instantaneous anomaly". This trend structure is the basis for subsequent path control and wind speed adjustment decisions, and can realize early warning and response to pollution changes.

[0030] In one alternative implementation, the construction of the sliding time window and the execution of differential coding in step S200 can also be adaptively configured by the window length, which can be configured according to the environmental characteristics of the system deployment. w can be adaptively adjusted according to the typical cycle of pollutant changes, such as the industrial emission cycle, or a shorter window can be used in scenarios requiring high responsiveness, such as sudden leaks, and a longer window can be used in scenarios requiring smooth and stable trends.

[0031] In another optional implementation, the sliding time window and differential encoding operation in step S200 can also extract trend features other than the first-order difference, such as mean, variance, and slope, or concatenate the original sequence with the first-order difference sequence to simultaneously include information on absolute concentration levels and relative change trends, forming a richer mixed feature vector.

[0032] In this embodiment of the application, step S300 involves constructing a pollution trend recognition model based on the pollution trend structure vector using a KNN classifier to obtain trend labels, including the following steps A1-A2: A1: By using historical monitoring data of different pollutants, a sliding window process is performed to extract differential coding trend features to form training samples and label them. A pollution trend recognition model is constructed using a lightweight K-nearest neighbor (KNN) classifier. Specifically, a pollution trend identification model is constructed using a lightweight K-nearest neighbor (KNN) classifier, including an input layer that takes the pollutant trend structure vector as input, a historical trend sample library, a distance metric module, and a majority voting classification output module.

[0033] Furthermore, during the model training phase, historical monitoring data for different pollutants are first processed using a sliding window to extract differential coding trend features to form training samples. These samples are then automatically labeled as "Rising", "Declining", "Volatile", or "Stable" based on the rate and volatility of pollution concentration changes.

[0034] Among them, the rate of change of pollution concentration is obtained by calculating the difference between the pollutant concentrations of two adjacent time points within a sliding time window and dividing it by the time interval to obtain a series of change rates (i.e., difference series). The average value of these change rates is then taken to indicate whether the overall trend is rising, falling, or stabilizing. The volatility is obtained by calculating the average value of the squares of the deviations between the difference series and the average rate of change and taking the square root, which indicates the severity and stability of pollution changes.

[0035] A2: Using the pollution trend structure vector as the input to the pollution trend identification model, Euclidean distance and majority voting are used to obtain the trend classification confidence and trend label.

[0036] Specifically, a sample library is constructed and Euclidean distance is used as the metric. Each type of sample is classified and stored according to its label. The model completes offline initialization by constructing a representative sample set with clear labels. The pollution trend structure vector T(t) at the current moment is input into the model, and the Euclidean distance is calculated with all trend samples in the sample library. The k nearest samples are selected and a majority vote is performed based on their labels. The trend category to which the current pollution state belongs, i.e., the trend label, is output to guide the switching of filter paths and the adjustment of fan power.

[0037] Among them, trend label L trend It consists of Rising, Declining, Volatile, and Stable, and is defined as follows: Rising: If the mean rate of change of pollutant concentration within the sliding window is greater than the positive threshold ω (e.g., ω = 0.15 mg / m³·min) and the standard deviation is less than the stability threshold σs (e.g., σs = 0.05), it indicates that the pollutant concentration is rising rapidly and continuously. Declining: If the average rate of change is less than the negative threshold θ (e.g., θ = −0.1 mg / m³·min) and the fluctuation range is stable, σs is also used as the standard, indicating that the pollution gradually decreases and the filtration effect is good; Volatile: If the standard deviation is greater than the fluctuation threshold σv (e.g., σv=0.12), meaning that the pollution changes drastically and fluctuates irregularly, the current path should be maintained and switching should be avoided; Stable: If the absolute value of the rate of change is lower than the set threshold λ (e.g., λ=0.03) and the standard deviation of the fluctuation is less than σs, the pollutant concentration changes smoothly without significant fluctuations, and can maintain stable operation with low power consumption.

[0038] Furthermore, the proportion of samples m belonging to the final predicted label such as Rising to the total number of voting samples k is the trend classification confidence level.

[0039] In an optional implementation, the pollution trend identification model constructed in step S300 using a KNN classifier based on Euclidean distance and majority voting can also be constructed using a weighted voting mechanism, which means that the votes are not equal. In this case, the voting weight of each nearest neighbor sample is inversely proportional to its Euclidean distance to the current sample, thereby giving more similar samples a higher say in the decision-making process.

[0040] In another optional implementation, the pollution trend identification model constructed in step S300 using a KNN classifier with Euclidean distance and majority voting can also be implemented using other distance metrics, such as Manhattan distance, which is less sensitive to small changes in a single data point in a certain dimension than Euclidean distance, and is therefore more robust in some cases; or cosine similarity, which ignores the overall concentration of pollutants and focuses on the core pattern of judging pollution trends.

[0041] In this embodiment of the application, step S400 uses trend tags to combine the pollutant state vector and operating parameters to obtain the optimal filtration path combination and the current recommended wind speed, which are used for path switching and wind speed adjustment, including the following steps B1-B5: B1: Based on trend labels combined with pollutant state vectors, a path selection model based on rule trees is used to obtain the dominant pollutant type; Specifically, the trend label L trend Together with the pollutant state vector Q(t), it is fed into the filter path combination controller. The controller integrates a path selection model based on a rule tree, which can dynamically decide the optimal filter combination structure according to the dominant pollutant type and evolution trend. Furthermore, after the controller receives the input, it first performs the pollutant dominance determination. Based on the pollutant concentration value in Q(t), it calculates the pollution source corresponding to the maximum value and marks it as the dominant pollutant.

[0042] B2: Based on the trend label and the dominant pollutant type, combined with the trend classification confidence level, proceed to the path selection logic branch; Specifically, based on trend labels and dominant pollutant types, the path selection logic branch is entered: If the dominant pollutant is PM 2.5 (t): When L trend =Rising, then the "coarse filter + HEPA + electrostatic dust collection" path combination is selected to quickly intercept high concentration of particulate matter and reduce the load on the filter element. When L trend=Volatile, then the “coarse filtration + HEPA” path combination is selected to maintain the basic particulate matter filtration path, avoid filter element impact caused by frequent fluctuations and accidental switching, and adapt to particulate matter changes with high volatility and unstable frequency. When L trend =Declining, then the “coarse filtration + HEPA” energy-saving path is selected, which maintains basic filtration efficiency while reducing air resistance pressure and extending the service life of the filter element. When L trend =Stable, then the "coarse filtration" path is selected to maintain the minimum filtration resistance, achieve low power consumption and stable operation, and is suitable for long-term stable particulate matter concentration areas. If the dominant pollutant is TVOC(t): When L trend =Volatile, then select "coarse filtration + activated carbon + catalytic decomposition layer" to deal with complex and fluctuating organic gases; When L trend =Rising, then the combination of "coarse filtration + activated carbon + catalytic decomposition layer" is selected to enhance the multi-stage decomposition capability of high-concentration organic gases and improve the efficiency of gas phase pollution treatment. When L trend =Declining, then the "coarse filtration + activated carbon" path combination is selected, which reduces the activation frequency of the catalytic module while ensuring basic gas phase filtration and saving resource consumption; When L trend =Stable, then the "coarse filtration + activated carbon (low wind speed)" path combination is selected to operate stably with the lowest wind resistance rate, delay the saturation rate of activated carbon, and achieve energy saving and stability in parallel. If the dominant pollutant is CO2(t): When L trend =Rising, then the combination of "coarse filtration + high wind speed ventilation bypass" is selected. By increasing the ventilation rate, the dilution is accelerated to avoid excessive indoor CO2 concentration that may cause hypoxia or fatigue, while maintaining basic particulate matter filtration. When L trend =Volatile, then the combination of "coarse filtration + intelligent wind speed adjustment + dynamic ventilation module" is selected, and the air volume is dynamically adjusted in combination with real-time concentration fluctuations to achieve highly responsive dilution control and avoid wind resistance fluctuations caused by accidental path switching; When L trend =Declining, then the energy-saving path of "coarse filtration + medium and low wind speed ventilation" is selected to reduce the power consumption of the fan and maintain an appropriate fresh air input, ensuring air freshness while reducing energy consumption; When L trend=Stable, then the combination of "coarse filtration + low-speed circulation ventilation" is selected to maintain airflow and CO2 level stability with minimum wind speed, which is suitable for the stable operation stage of air conditioning / fresh air. If an overall decrease in pollution concentration is detected (i.e., L) trend =Declining or L trend If the filter is Stable, it will switch to the energy-saving mode path "coarse filter + HEPA" to extend the filter life and reduce energy consumption. The process involves several key components: Coarse filtration (FEPA) refers to the primary filtration layer installed before the air enters the main filtration unit, used to intercept larger particles such as dust, hair, and fibers, reducing the load on subsequent fine filtration layers; HEPA refers to a high-efficiency filter material layer capable of efficiently capturing particles of 0.3 microns and larger, such as PM2.5, pollen, and bacteria; electrostatic dust collection refers to a device that uses a high-voltage electrostatic field to charge fine particles in the air and adsorb them onto a dust collection plate, achieving the capture of fine dust without traditional filter materials; activated carbon refers to an adsorption material with a large number of micropores, used to adsorb volatile organic compounds (TVOC), odors, and harmful gases from the air; and the catalytic decomposition layer refers to a layer coated with a catalyst, such as a photocatalyst or a metal catalyst, which decomposes adsorbed harmful gases into harmless substances through chemical reactions, improving purification efficiency and extending the lifespan of activated carbon.

[0043] Furthermore, if the confidence level of the trend classification is lower than the preset threshold U, the previous path configuration is maintained to prevent disturbances caused by frequent switching, where U is set through experimental optimization.

[0044] In an optional implementation, the path selection model built based on the rule tree in step S400 can also adopt a hierarchical decision framework, so that the first-layer rule tree still serves as the basis for quickly judging the dominant pollutants and basic trends, and the second layer introduces a lightweight neural network or fuzzy logic controller to make more refined decisions on complex working conditions with fuzzy boundaries, making the model more accurate.

[0045] In another optional implementation, the path selection model built based on the rule tree in step S400 can also make the model's parameters configurable, such as the threshold U, which can be updated through cloud distribution, user customization, or through a self-learning mechanism, enabling the system to evolve.

[0046] B3: By modeling the path selection operation as a path mapping function, the optimal combination of filtering paths is obtained for filter path switching; Specifically, the path selection operation is modeled as a path mapping function: in, It is the optimal air filtration path number selected at the current time t based on the pollution type and trend. It is a path selection function built on rule tree logic, used to select a path based on the pollutant state vector Q(t) and trend label L. trend To determine the most suitable combination of filtering paths at the moment.

[0047] The obtained path selection result is the defined path number, such as Path-1, Path-2, etc., which are mapped to different filter element structure combinations. This number is sent as a control signal to the filter element switching control unit. The PLC control system outputs digital signals to drive multi-channel electric actuators, such as electric ball valves or solenoid valves, to realize the opening and closing control of different filter element channels.

[0048] It should be noted that by integrating the dominant pollution type and evolution trend to achieve dynamic path switching, compared with the traditional static filter configuration, it can respond quickly under conditions of severe pollution fluctuations or high load, improve filtration efficiency, and actively enter energy-saving mode during pollution mitigation phases, effectively extending filter life and reducing system energy consumption. This reflects the essential transformation of air filtration systems from "passive response" to "active adjustment".

[0049] B4: By mapping the trend label to the wind speed gain signal, introducing the wind speed suppression factor, and combining it with the current filter pressure difference, the current recommended wind speed is calculated. Specifically, the trend label L trend Mapped to wind speed gain signal α: If L trend =Rising indicates a rapid increase in pollution, requiring a faster ventilation rate to improve filtration throughput. Let α = α max >0; If L trend =Declining / Stable indicates that the pollution load is reduced or tends to stabilize, and the wind speed can maintain the default value. Let α=0. If L trend =Volatile indicates that pollution levels fluctuate significantly, and frequent speed adjustments are not recommended; a gradual response should be maintained. Let α = α mid ∈(0,α max ); Considering that an increase in pressure differential usually indicates an increase in filter load or a tendency to become clogged, a wind speed suppression factor β is introduced to adjust the wind speed downward based on the current pressure differential value. The current recommended wind speed v(t) is calculated in real time according to the following adjustment formula: In the formula, This is the default base wind speed, and α is the wind speed gain signal. It's a logical judgment: it takes 1 when the pollution trend is "Rising," and 0 otherwise. This is the current filter element pressure difference; It is a wind speed suppression factor, among which The settings are configured through experimental fitting or simulation parameter tuning to map the pressure difference into a wind speed compensation value. In an optional implementation, step S400 can further process the calculated current recommended wind speed into a dynamically coupled variable, whereby the wind speed gain signal or wind speed suppression factor can be processed into a dynamic coupling variable. For example, the wind speed gain signal can be related not only to the trend label but also to the pollutant concentration; for instance, a trend label with a high concentration of pollutants can trigger a stronger gain. Similarly, the wind speed suppression factor can be designed as a function of the filter cartridge's usage time, meaning that the longer the usage time, the more sensitive it is to pressure differential, thus enabling early protection.

[0050] In another optional implementation, the current recommended wind speed calculated in step S400 can also have a nonlinear compensation term added to the wind speed calculation formula to represent the response to the pressure difference change trend. If the pressure difference rises rapidly due to the acceleration of blockage, the compensation term provides an additional wind speed reduction to improve control accuracy.

[0051] B5: Adjusts wind speed by converting the current recommended wind speed into a control signal; Specifically, the calculated recommended wind speed v(t) is converted into a pulse width modulation or voltage control signal and transmitted to the fan control module. After receiving the signal, the fan module gradually adjusts the motor speed to ensure that the wind speed change is slow and stable, avoiding oscillations or filter structure disturbances caused by sudden air volume changes.

[0052] In this embodiment of the application, step S500, based on the completion of path switching and wind speed adjustment, determines the blockage trend according to operating parameters, including: Specifically, after completing the wind speed linkage adjustment and re-establishing the air circulation channel, the system enters the continuous tracking phase of the filtration path operation status. At this time, the switched filter structure is working, and the wind speed has been stabilized. It is necessary to monitor the key operating parameters of the current channel in real time to determine whether there is a tendency for filter blockage. Pressure difference and flow rate are continuously sampled, and the rate of change is constructed based on continuous time series to determine whether there is a joint trend. in, It is the rate of increase of pressure difference per unit time, reflecting the increase in filter element load; It is the rate of airflow attenuation per unit time, reflecting the increase in air resistance; and To determine the threshold for judging clogging trends, the following settings were implemented: In a controlled experimental environment, by adjusting the pollutant concentration and fan operating status, the growth curves of filter cartridge pressure differential over time and the flow rate decay trajectory were recorded under multiple typical clogging processes. The derivative features of the continuous time series were extracted using methods such as least squares or sliding window fitting to determine the typical value ranges of pressure differential growth rate and flow rate decay rate at different stages. Critical change rates that can significantly distinguish between "normal dust accumulation" and "abnormal clogging" were selected as the initial threshold candidate set. Cross-validation was performed based on large-sample measured data. Taking into account both false alarm rate and false negative rate, the final judgment threshold was optimized and selected using indicators such as ROC curves or F1-scores. and This ensures that it has sufficient versatility and engineering adaptability.

[0053] When the joint conditions are met consecutively for more than two adoption cycles, it is considered that there is a congestion trend and the region enters the congestion trend growth zone.

[0054] In this embodiment of the application, step S600 assesses the risk level based on the existence of a blockage trend using a blockage duration recording mechanism, generates and executes a maintenance reminder, and completes air filtration, including the following steps C1-C3: C1: If a congestion trend is observed and the congestion trend is increasing, the congestion duration recording mechanism will be activated to accumulate the duration of this state in real time. C2: By combining the duration of blockage with preset thresholds, differential pressure, and flow rate, the risk level is assessed. Based on the assessed risk level, corresponding maintenance reminders are obtained and corresponding protection actions are executed. Specifically, if the duration of blockage exceeds the threshold T, where T is set through experimental optimization, it indicates that the blockage trend has been maintained beyond the initial tolerance threshold but has not reached a severe level. Immediately, a filter maintenance reminder will be issued through the human-machine interface or remote port, prompting the operator to check or prepare to replace the filter. Furthermore, if the differential pressure exceeds the safety upper limit, the flow rate is below the safety lower limit, and the duration of clogging exceeds the threshold T, the filter element is determined to be severely clogged, and the following three-level protection actions are immediately executed to ensure uninterrupted system operation: Activate the bypass channel electric valve to switch the airflow to the backup channel, bypassing the blockage structure; Control the fan speed to reduce the preset safe power to reduce the load; Send protection signals to downstream controlled equipment to cause it to enter self-protection state or shutdown standby mode; The bypass channel is used to switch the airflow to the backup channel when the main channel is blocked. This backup channel is an independently designed complete filtration path.

[0055] C3: Record the pollutant concentration, trend label, selected path, wind speed, pressure difference and flow rate changes collected in each sampling period in chronological order to form a traceable operation log, and finally complete the air filtration.

[0056] In summary, this invention generates a pollution trend structure vector by constructing a pollutant state vector and performing time window processing. A KNN classifier is used to identify pollution trend labels, and the optimal filtration path combination and recommended wind speed are dynamically determined based on these trend labels, achieving intelligent linkage between path switching and wind speed adjustment. Furthermore, operating parameters are used to determine the filter clogging trend, and a duration recording mechanism is used to assess the risk level and issue maintenance reminders. This method overcomes the lag problem of traditional systems relying on instantaneous concentration response, achieving proactive perception of pollution evolution trends and multi-parameter collaborative control, significantly improving the adaptability, energy efficiency management capabilities, and operational safety of air filtration systems.

[0057] Example 3 illustrates an air filtration method. It should be noted that the technical solution of this air filtration system belongs to the same concept as the air filtration method described above. Details not described in detail in this embodiment can be found in the description of the air filtration method described above.

[0058] This embodiment also provides an air filtration system, including: The acquisition module is used to acquire pollutants and operating parameters and construct pollutant state vectors. The trend feature extraction module is used to obtain the pollution trend structure vector by performing time window processing on the pollutant state vector; The model building module is used to construct a pollution trend recognition model based on the pollution trend structure vector and a KNN classifier to obtain trend labels. The decision module is used to obtain the optimal filtration path combination and the current recommended wind speed by combining the pollutant state vector and operating parameters with trend labels, and is used for path switching and wind speed adjustment. The judgment module is used to determine the blockage trend based on the operating parameters after the path switching and wind speed adjustment are completed; The execution module is used to assess the risk level based on the presence of a blockage trend, using a blockage duration recording mechanism, generate maintenance reminders and execute them to complete air filtration.

[0059] This embodiment also provides an electronic device suitable for air filtration, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the air filtration method proposed in the above embodiment.

[0060] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the air filtration method as described in the above embodiments.

[0061] The storage medium proposed in this embodiment and the air filtration method proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0062] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0063] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An air filtration method, characterized in that, include: Acquire pollutants and operating parameters, and construct a pollutant state vector; By applying a time window to the pollutant state vector, a pollution trend structure vector is obtained. Based on the pollution trend structure vector, a pollution trend recognition model is constructed using a KNN classifier to obtain trend labels; By combining trend tags with pollutant state vectors and operating parameters, the optimal filtration path combination and the current recommended wind speed are obtained, which are used for path switching and wind speed adjustment. Based on the completion of path switching and wind speed adjustment, the blockage trend is judged according to the operating parameters; Based on the presence of a blockage trend, a blockage duration recording mechanism is used to assess the risk level, generate and execute maintenance reminders, and complete air filtration.

2. The air filtration method as described in claim 1, characterized in that, Pollutants and operating parameters are acquired, and a pollutant state vector is constructed. By applying a time window to the pollutant state vector, a pollution trend structure vector is obtained, including: The pollutant parameters, including particulate matter concentration, total volatile organic compounds and carbon dioxide concentration, are obtained and constituted as a pollutant state vector. The operating parameters, including the pressure difference across the filter element and the gas flow rate, are also obtained. A sliding time window is constructed, and differential coding is performed on each type of pollutant within the time window to obtain the trend change vectors of particulate matter concentration, total volatile organic compounds, and carbon dioxide concentration; The trend change vectors of each type of pollutant together constitute the pollution trend structure vector.

3. The air filtration method as described in claim 2, characterized in that, Based on the pollution trend structure vector, a pollution trend recognition model is constructed using a KNN classifier to obtain trend labels, including: By using historical monitoring data of different pollutants, a sliding window process is performed to extract differential coding trend features to form training samples and label them. A pollution trend recognition model is constructed using a lightweight K-nearest neighbor (KNN) classifier. Using the pollution trend structure vector as input to the pollution trend identification model, Euclidean distance and majority voting are used to obtain the trend classification confidence and trend label.

4. The air filtration method as described in claim 3, characterized in that, The optimal filtering path combination is obtained by combining trend labels with pollutant state vectors, and is used for path switching, including: Based on trend labels combined with pollutant state vectors, a path selection model based on rule trees is used to obtain the dominant pollutant type; Based on the trend label and the dominant pollutant type, combined with the trend classification confidence level, enter the path selection logic branch; By modeling the path selection operation as a path mapping function, the optimal combination of filtering paths is obtained for filter path switching.

5. The air filtration method as described in claim 4, characterized in that, The recommended wind speed is obtained by combining trend tags with operating parameters, and is used to adjust the wind speed, including: By mapping trend labels to wind speed gain signals, introducing wind speed suppression factors, and combining them with the current filter pressure difference, the current recommended wind speed is calculated. The wind speed is adjusted by converting the current recommended wind speed into a control signal.

6. The air filtration method as described in claim 5, characterized in that, Based on the completed path switching and wind speed adjustment, the blockage trend is determined according to the operating parameters, including: After completing path switching and wind speed adjustment, pressure difference and flow rate are continuously sampled, and the rate of change is constructed based on continuous time series. The blockage trend is determined by comparing the rate of change of pressure difference and flow rate with the blockage trend judgment threshold.

7. The air filtration method as described in claim 6, characterized in that, Based on the existing blockage trend, a blockage duration recording mechanism is used to assess the risk level, generate and execute maintenance reminders, and complete air filtration, including: Based on the existence of a congestion trend, we enter the congestion trend growth zone and use a congestion duration recording mechanism to obtain the congestion duration. The risk level is assessed by combining the duration of blockage with preset thresholds, pressure differentials, and flow rates. Based on the assessed risk level, corresponding maintenance reminders are received and corresponding protective actions are executed. Record all operational data and status changes and build an operational log to complete air filtration.

8. An air filtration system, employing the method as described in any one of claims 1-7, characterized in that, include: The acquisition module is used to acquire pollutants and operating parameters and construct pollutant state vectors. The trend feature extraction module is used to obtain the pollution trend structure vector by performing time window processing on the pollutant state vector; The model building module is used to construct a pollution trend recognition model based on the pollution trend structure vector and a KNN classifier to obtain trend labels. The decision module is used to obtain the optimal filtration path combination and the current recommended wind speed by combining the pollutant state vector and operating parameters with trend labels, and is used for path switching and wind speed adjustment. The judgment module is used to determine the blockage trend based on the completed path switching and wind speed adjustment, and according to the operating parameters. The execution module is used to assess the risk level based on the presence of a blockage trend, using a blockage duration recording mechanism, generate maintenance reminders and execute them to complete air filtration.

9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the air filtration method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It includes storing computer-executable instructions that, when executed by a processor, implement the steps of the air filtration method according to any one of claims 1 to 7.

Images