An air purifier and its control method

Abstract

The present invention relates to an air purifier and a control method thereof. The air purifier includes: a housing having an air inlet and an air outlet; a blower disposed within the housing and connected to the air outlet; a multi-stage filter arranged in sequence on the gas flow path between the air inlet and the air outlet; and is characterized in that: it further includes a detection module disposed on each stage of the filter for detecting air data; a controller connected to the detection module and the control mechanism of the blower, and the controller is configured to: according to the detection results of the detection module, predict the remaining life prediction results of each stage of the filter and / or control the blower to perform corresponding actions. This air purifier can automatically detect and judge the filter status, and remind the user when replacement is needed; when the air quality is good, it reduces the power of the blower; when the air quality is poor, it increases the power of the blower to ensure the air purification effect and extend the service life of the blower and the filter.

Description

Technical Field

[0001] The present invention relates to the technical field of air purification, and particularly to an air purifier and a control method thereof. Background Art

[0002] At present, with the improvement of living standards, people's demand for healthy air in relatively enclosed spaces such as indoors and cars is becoming stronger and stronger. In the prior art, air purifiers are often used to replace indoor air to improve indoor air quality.

[0003] There are many air purifiers in the prior art. For example, a Chinese utility model patent with the patent number ZL 201621440691.8 (the authorized publication number is CN 206444369U) discloses an air purifier, which includes a housing, a housing cover provided at the upper end of the housing, and a blower provided inside the housing. An air outlet opposite to the air outlet of the blower is provided on the housing, and an air inlet is provided on the housing cover. A filter element is provided between the air inlet and the blower. Although the above air purifier can adsorb various harmful substances such as dust and bacteria in the air through the filter element to purify the air at the air inlet and blow out the purified air. However, since the filter element has an expected service life during manufacturing and needs to be replaced regularly, and in the prior art, the remaining service life of the filter element is usually estimated manually by people, so there are the following limitations in use: First, if the filter element is replaced before its actual service life has expired, it will cause great waste; Second, if the actual service life of the filter net has expired while the expected service life has not expired and it is not replaced, because the filter net loses its filtering ability and cannot further purify the air, it may even cause secondary pollution of the air.

[0004] To solve this technical problem, a Chinese invention patent "Air Purification Equipment Filter Life Detection Method, Device, System and Equipment" with the application number CN201910777465.0 (publication number CN110529976A) discloses a filter life detection method, which includes: determining the periodic dust collection amount of the filter in the air purification equipment in each sampling period according to the inlet air quality parameters and the outlet air speed corresponding to each sampling period; obtaining the cumulative dust collection amount of the filter according to the periodic dust collection amount of each sampling period; calculating the remaining life of the filter according to the cumulative dust collection amount and the target dust collection amount. Since the periodic dust collection amount of the filter is collected in this method, and the remaining life of the filter is calculated according to the cumulative dust collection amount and the target dust collection amount, and the filter in actual use is not completely affected by the dust collection amount. Most of the filters affected by the dust collection amount are only used to filter physical pollution (dust, pollen, animal dander, etc.), while many filters in the prior art can also filter out biological pollution (bacteria, viruses, etc.) and chemical pollution (NOx, SOx, TVOC, etc.), etc. Therefore, the filters adsorbing these pollutants are not affected by the dust collection amount, so the remaining life of this method is not accurate. Therefore, it is necessary to further improve the prior art. Summary of the Invention

[0005] The first technical problem to be solved by the present invention is directed to the above-mentioned prior art, and to provide an air purifier that can automatically monitor the life of the filter and improve the filtering ability.

[0006] The second technical problem to be solved by the present invention is to provide a control method for the above-mentioned air purifier to make the prediction of the filter life more accurate.

[0007] The technical solution adopted by the present invention to solve the above-mentioned first technical problem is: an air purifier, including:

[0008] A housing, on which an air inlet and an air outlet are provided;

[0009] A blower, arranged in the housing and opposite to the air outlet;

[0010] A multi-stage filter, arranged in sequence on the gas flow path between the air inlet and the air outlet;

[0011] Characterized in that: it further includes

[0012] A detection module, arranged on each stage of the filter, for detecting air data;

[0013] A controller, connected to the detection module and the control mechanism of the blower, and the controller is configured to: according to the detection results of the detection module, predict the remaining life prediction results of each stage of the filter and / or control the blower to perform corresponding actions.

[0014] In the above solution, the detection module is at least one sensor provided on each stage of the filter, and is used to monitor at least the air flow rate, pressure fluctuation conditions, and air quality monitoring data.

[0015] Preferably, the detection module on each stage of the filter includes at least one or more of a laser dust sensor capable of tracking the number, size, and aggregation speed of particles, a pressure sensor for measuring air pressure, and an air sensor for obtaining air quality monitoring data.

[0016] Preferably, the filter includes a pre-filter, a HEPA filter, and a pickling-impregnated activated carbon filter arranged in sequence along the air flow direction on the air flow path. The pre-filter is composed of a micron mesh dust filter, which blocks large particulate matter and protects other filter meshes.

[0017] In order to reduce noise pollution, it further includes a muffler provided in the housing. The muffler is made of porous sound-absorbing material and is arranged in close fit with the inner peripheral wall of the housing.

[0018] The technical solution adopted by the present invention to solve the above second technical problem is as follows: An air purifier control method as described above, characterized in that it includes the following steps:

[0019] After the air purifier is turned on and operates, the data during the operation of the air purifier is monitored and collected in real time through the detection module on each stage of the filter;

[0020] Combining the data collected by all detection modules to predict the remaining life of the filter, obtaining the remaining life prediction results of each stage of the filter, and judging whether the remaining life prediction value of any stage of the filter is greater than the preset replacement life. If so, there is no need to replace the filter. If not, it is prompted that the filter needs to be replaced;

[0021] and / or

[0022] Performing comprehensive calculations on the data collected by all detection modules to obtain the air quality parameter Q of the current air, and judging whether Q is greater than the preset value Q0. If so, control the blower to turn on, and if Q gradually decreases within the set time after the blower is turned on, correspondingly control the blower speed to decrease. If Q gradually increases within the set time, correspondingly control the blower speed to increase; if not, stop the air purifier.

[0023] As an improvement, the remaining life prediction of each of the above filters is carried out separately. The specific steps for predicting the remaining life of any one filter are as follows:

[0024] Step 1: Collect the status data during the operation of the filter monitored by the detection module in real time and record it in a time series manner to obtain time series data;

[0025] Step 2: Preprocess the time series data to obtain the preprocessed time series data, screen out the features with strong correlation with the performance degradation of the filter from the preprocessed time series data, and perform normalization processing on the screened features to obtain a data set;

[0026] Step 3: Sort the data set in chronological order, use the first N data in the data set as the training set, and use the remaining other data in the data set as the test set; N is a preset positive integer;

[0027] Step 4: Extract the attribute values and the corresponding label values of all training samples in the training set and store them as the first feature matrix and the first target vector respectively;

[0028] Among them, the first feature matrix is a matrix of size N*n, where n corresponds to the total number of features screened in Step 2. All data in the k-th row of the first feature matrix represent the n attribute values of the k-th training sample in sequence; the first target vector is a matrix of size N*1, and the value in the k-th row of the first target vector corresponds to the performance degradation rate of the k-th training sample;

[0029] Step 5: Divide the first feature matrix into multiple time windows to obtain multiple time window matrices;

[0030] Step 6: Construct a performance degradation trajectory model and use the multiple time window matrices in Step 5 to train the constructed performance degradation trajectory model in batches to obtain a trained performance degradation trajectory model;

[0031] Step 7: Arbitrarily select at least one test sample in the test set, and in the same way as in Step 4, extract the attribute values and the corresponding label values of each test sample and store them as the second feature matrix and the second target vector respectively. Input each data in the second feature matrix into the trained performance degradation trajectory model to obtain the performance degradation rate predicted for the attribute value of each test sample;

[0032] Step 8: Calculate the error between the predicted value of the performance degradation rate corresponding to each test sample and the true value of the performance degradation rate in the second target vector, and determine whether the error is less than the preset error value. If so, go to Step 10; if not, go to Step 9;

[0033] Step 9: Go to Step 1 to collect the data monitored by the detection module in real time again;

[0034] Step 10: Collect the status data of the filter during operation monitored by the detection module at the current moment, extract the n attribute values corresponding to the filter at the current moment in the manner of Step 2, and then input the n attribute values corresponding to the filter at the current moment into the performance degradation trajectory model trained in Step 6 to obtain the performance degradation rate of the filter at the current moment;

[0035] Step 11: Calculate the predicted remaining life according to the performance degradation rate of the filter at the current moment.

[0036] Preferably, the formula of the degradation trajectory model in Step 6 is:

[0037]

[0038] where f(t) represents the performance degradation rate of the filter at time t, β0, β i and β ij are all parameters of the degradation trajectory model. β0 represents the intercept term of the degradation trajectory model, β i represents the influence coefficient of the i-th attribute on the performance degradation of the filter, and β ij represents the interaction coefficient between the i-th attribute and the j-th attribute; x i (t) represents the value of the i-th attribute at time t; x j (t) represents the value of the j-th attribute at time t, and ∈ represents the noise term.

[0039] Preferably, the calculation formula of the air quality parameter Q is:

[0040]

[0041] where N is the concentration of nitrogen dioxide, with the unit of mg / m 3 ; S is the concentration of sulfur dioxide, with the unit of mg / m 3 ; C is the concentration of carbon monoxide, with the unit of mg / m 3 , O is the concentration of ozone, with the unit of mg / m 3 ; P1 is the content of PM10, with the unit of mg / m 3 ; P2 is the content of PM2.5, with the unit of μg / m 3 ; E is the concentration of formaldehyde, with the unit of mg / m 3 ; T is the content of TVOC, with the unit of mg / m 3 .

[0042] Furthermore, the value of the preset Q0 is 1.

[0043] Compared with the prior art, the advantages of the present invention are as follows: By providing a detection module for detecting air data on each stage of the filter, and connecting the controller to the detection module and the control mechanism of the blower, the controller can, according to the detection results of the detection module, predict the remaining life prediction results of each stage of the filter and / or control the blower to perform corresponding actions. Therefore, the air purifier can automatically detect and judge the filter state, remind the user when replacement is needed, ensure the air purification quality and improve the utilization rate of the filter; and when the air quality is good, reduce the power of the blower; when the air quality is poor, increase the power of the blower to ensure the air purification effect, so as to extend the service life of the blower and the filter and reduce the maintenance cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic structural diagram of the air purifier in an embodiment of the present invention;

[0045] Figure 2 is Figure 1 a partial structural diagram of (omitting the front side of the housing). DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The present invention will be further described in detail below with reference to the embodiments of the drawings.

[0047] As Figures 1-2 shown, the air purifier in this embodiment includes a housing 1, a blower 2, a multi-stage filter, a detection module and a controller. Among them, an air inlet 11 and an air outlet 12 are provided on the housing 1; the blower 2 is arranged in the housing 1 and is oppositely arranged with the air outlet 12; the multi-stage filter is sequentially arranged on the gas flow path between the air inlet 11 and the air outlet 12; the detection module is arranged on each stage of the filter for detecting air data; the controller is connected to the detection module and the control mechanism of the blower 2, and the controller is configured to: according to the detection results of the detection module, predict the remaining life prediction results of each stage of the filter and / or control the blower to perform corresponding actions.

[0048] The above detection module is at least one sensor provided on each stage of the filter, which is used to monitor at least the air flow rate, pressure fluctuation and air quality monitoring data. The detection module on each stage of the filter includes at least one or more of a laser dust sensor capable of tracking the number, size and agglomeration speed of particles, a pressure sensor for measuring air pressure, and an air sensor for obtaining air quality monitoring data. In this embodiment, each stage of the filter separately monitors the air flow rate and pressure fluctuation, and uses a laser dust sensor to track the number, size and agglomeration speed of particles; and uses an air sensor to monitor air quality data. The air sensor reads the data every 10 - 15 minutes, and the specific reading time interval is set by the user. All the set data and air quality monitoring data are stored in the air purifier controller and uploaded to the central server. The control system retains the data within 3 - 7 days to ensure that the controller has sufficient storage space to maintain the normal operation state. The air quality monitoring data includes temperature, humidity, air pressure, various volatile organic compounds, carbon dioxide, nitrogen dioxide, ozone, formaldehyde, air particles with diameters of 1.0, 2.5, 4.0 and 10.0 microns, inlet and outlet air flow rates, and all related data.

[0049] In this embodiment, the filter includes a pre-filter 31, a HEPA filter 32 and a pickling-impregnated activated carbon filter 33 arranged in sequence along the air flow direction on the air flow path. The pre-filter 31 is composed of a micron mesh dust filter. The mesh area is very small, which can adsorb small dust and particles. Dust, hair, large particles, etc. can also be blocked outside, and it can also play a role in protecting other filter meshes, and can reach the purification standard of MERV 8; the HEPA filter 32 uses the PAO (poly-α-olefin) method to replace a known carcinogen DOP (dioctyl phthalate); the pickling-impregnated activated carbon filter 33 is mainly activated carbon treated by pickling. The activated carbon can remove soluble organic substances, synthetic detergents, microorganisms, viruses and a certain amount of heavy metals, and can also decolorize and deodorize. Through the catalytic action of artificial enzyme decomposition and chemical adsorbents, the odor becomes lighter until the odor is fully adsorbed. After impregnation, its purification ability is significantly improved, and its adsorption changes from physical adsorption to the simultaneous action of physical adsorption and chemical adsorption, and at the same time solves the problem of secondary pollution of activated carbon. Of course, the type of filter can also be changed according to actual needs.

[0050] In addition, a composite filter screen is provided in the filtering device, and an ultraviolet germicidal lamp is installed between the filter screens, which can prevent the reproduction of bacteria and viruses; a pressure sensor is installed between the filter screens, and the air pressure is measured as a parameter for controlling the operating state of the air purifier; a fan is provided above the filtering device to cooperate with the composite filter screen to achieve physical purification; a negative ion generator is installed on the fan fixing frame, which is beneficial to increasing the diffusion range of negative ions at the air outlet, achieving better air purification and disinfection effects. A filter box door is installed on the front of the filtering device for the replacement and cleaning of the filter screen. The filter box door and the filter box are sealed with neoprene to prevent dust, PM2.5 particles, bacteria, etc. on the filter screen from spreading into the indoor air again and ensuring the air purification effect.

[0051] The above-mentioned air purifier further includes a silencer 4 provided in the housing 1. The silencer 4 is made of porous sound-absorbing material and is arranged to fit the inner peripheral wall of the housing 1. The silencer 4 can effectively reduce the noise pollution generated when the air purifier is in use. In addition, the noise reduction system also combines holes and frequency absorbers. By adopting these measures, the noise of the blower and air movement can be significantly reduced without reducing the air flow. This device also has the function of enhancing air movement, making the air move farther. Therefore, its air purification range is larger than that of traditional air purifiers.

[0052] The control method of the above-mentioned air purifier includes the following steps:

[0053] After the air purifier is turned on and working, the detection modules on each stage of the filter are used to monitor and collect the data during the operation of the air purifier in real time;

[0054] Combining the data collected by all detection modules to predict the remaining life of the filter, obtaining the remaining life prediction results of each stage of the filter, and judging whether the remaining life prediction value of any stage of the filter is greater than the preset replacement life. If so, there is no need to replace the filter; if not, it is prompted that the filter needs to be replaced;

[0055] and / or

[0056] Performing comprehensive calculation on the data collected by all detection modules to obtain the air quality parameter Q of the current air, and judging whether Q is greater than the preset value Q0. If so, control the blower to turn on, and if Q gradually decreases within the set time after the blower is turned on, correspondingly control the blower speed to decrease; if Q gradually increases within the set time, correspondingly control the blower speed to increase; if not, stop the air purifier from working;

[0057] The calculation formula for the above-mentioned air quality parameter Q is:

[0058]

[0059] where N is the concentration of nitrogen dioxide, with the unit of mg / m3 ; S is the sulfur dioxide concentration, with the unit of mg / m 3 ; C is the carbon monoxide concentration, with the unit of mg / m 3 , O is the ozone concentration, with the unit of mg / m 3 ; P1 is the PM10 content, with the unit of mg / m 3 ; P2 is the PM2.5 content, with the unit of μg / m 3 ; E is the formaldehyde concentration, with the unit of mg / m 3 ; T is the TVOC content, with the unit of mg / m 3 ; In this embodiment, the preset value Q0 is taken as 1.

[0060] The remaining life prediction of each of the above filters is carried out separately, and the remaining life prediction method for each filter is the same. The specific steps of the remaining life prediction method are as follows:

[0061] Step 1: Collect the status data during the operation of the filter monitored by the acquisition and detection module in real time, and record it in the form of time series to obtain time series data;

[0062] In this embodiment, the status data during the operation of the filter includes status parameters, air quality data, wind speed, the number and size of particles, etc.;

[0063] Step 2: Preprocess the time series data to obtain the preprocessed time series data, screen out the features with strong correlation with the filter performance degradation from the preprocessed time series data, and perform normalization processing on the screened features to obtain a data set;

[0064] In this embodiment, the feature screening method is obtained through common historical data analysis or historical experience. Generally, the screened features are features such as temperature, humidity, differential pressure change rate, air volume, etc. However, the screened features are still related to different types of filters. The features with strong correlation with the filter performance degradation corresponding to different types of filters will be different and will not be elaborated here. It needs to be determined according to the actual filter type used;

[0065] In addition, the purpose of performing normalization processing on the screened features is to eliminate the dimensional difference between different features and make the weights between features more balanced;

[0066] Step 3: Sort the data set in chronological order, and take the first N data in the data set as the training set, and take the remaining other data in the data set as the test set; N is a preset positive integer;

[0067] In this embodiment, the training set is the first 80% of the data (i.e., N / A = 80%, where A is the total number of the data set), and the remaining 20% of the data is used as the test set;

[0068] Step 4: Extract the attribute values and corresponding label values of all training samples in the training set, and store them as the first feature matrix X and the first target vector y respectively;

[0069] Among them, the first feature matrix X is a matrix of size N*n, where n corresponds to the total number of features selected in Step 2. All the data in the k-th row of the first feature matrix represent the n attribute values of the k-th training sample in sequence; the first target vector is a matrix of size N*1, and the value in the k-th row of the first target vector corresponds to the performance degradation rate of the k-th training sample;

[0070]

[0071] y = (y1, y2,..., y N ) T

[0072] where p 1,1 , p 2,1 ..p n,1 are the first attribute value of the first training sample, the second attribute value of the first training sample, and the n-th attribute value of the first training sample respectively, and y1 is the performance degradation rate of the first training sample;

[0073] p 1,2 , p 2,2 ..p n,2 are the first attribute value of the second training sample, the second attribute value of the second training sample, and the n-th attribute value of the second training sample respectively, and y2 is the performance degradation rate of the second training sample;

[0074] p 1,N , p 2,N ..p n,N are the first attribute value of the N-th training sample, the second attribute value of the N-th training sample, and the n-th attribute value of the N-th training sample respectively, and y N is the performance degradation rate of the N-th training sample; each of the above training samples corresponds to 1 time point;

[0075] Step 5: Divide the first feature matrix into multiple time windows to obtain multiple time window matrices;

[0076] In this embodiment, the sliding window cutting method is used to segment the first feature matrix to divide the first feature matrix into multiple time windows with a length of w and a sliding step of s;

[0077] Sliding window cutting method:

[0078] X = [x1, x2, …, x T ​T , X 1+(k-1)s:w+(k-1)s = [x i , x i+1 , …, x i+w-1 T

[0079] Among them, T represents the total length of the time series data, and X 1+(k-1)s:w+(k-1)s represents the k-th time window, and x i represents the feature vector of the i-th time series data;

[0080] Step 6: Construct a performance degradation trajectory model, and use multiple time window matrices in Step 5 to train the constructed performance degradation trajectory model in batches to obtain a trained performance degradation trajectory model;

[0081] In this embodiment, the degradation trajectory model formula is:

[0082]

[0083] Among them, f(t) represents the performance degradation rate of the filter at time t, and β0, β i and β ij are all parameters of the degradation trajectory model. β0 represents the intercept term of the degradation trajectory model, and β i represents the influence coefficient of the i-th attribute on the filter performance degradation, and β ij represents the interaction coefficient between the i-th attribute and the j-th attribute; x i (t) represents the value of the i-th attribute at time t; x j (t) represents the value of the j-th attribute at time t, and ∈ represents the noise term; by fitting the existing historical data (attribute values and label values), the least squares method is used to determine the specific values of the model parameters β;

[0084] Of course, this performance degradation trajectory model can also adopt neural networks in the prior art, such as: mainstream networks such as convolutional neural networks and BP neural networks;

[0085] Step 7: Arbitrarily select at least one test sample in the test set, and in the same manner as in Step 4, extract the attribute values and the corresponding label values of each test sample respectively, and store them as the second feature matrix and the second target vector respectively. Input each data in the second feature matrix into the trained performance degradation trajectory model to obtain the performance degradation rate predicted by the attribute values of each test sample;

[0086] Step 8: Calculate the error between the predicted value of the performance degradation rate corresponding to each test sample and the true value of the performance degradation rate in the second target vector, and determine whether the error is less than the preset error value. If so, go to Step 10; if not, go to Step 9;​

[0087] In this embodiment, the root mean square error (RMSE) is used as the error. If the RMSE is less than the preset error value, it indicates that the model has good generalization ability. If the RMSE is large, the model needs to be optimized by collecting more monitoring data, reselecting features, adjusting model parameters, etc. to improve the model's generalization ability.

[0088] The calculation formula of the root mean square error (RMSE) is as follows:

[0089]

[0090] where y i represents the true value, represents the predicted value, and n represents the number of samples;

[0091] Step 9: Go to Step 1 and re-collect the data real-time monitored by the detection module;

[0092] Step 10: Collect the status data of the filter during operation monitored by the detection module at the current moment, and extract the n attribute values corresponding to the filter at the current moment in the manner of Step 2. Then, input the n attribute values corresponding to the filter at the current moment into the performance degradation trajectory model trained in Step 6 to obtain the performance degradation rate of the filter at the current moment;

[0093] Step 11: Calculate the predicted remaining life according to the performance degradation rate of the filter at the current moment.

[0094] The calculation formula of the remaining life R(t) is as follows:

[0095]

[0096] where R0 represents the initial life of the filter, and f(t) is the performance degradation rate of the filter in the current state.

[0097] In this embodiment, with the accumulation of air quality monitoring data, this control system will be able to accurately predict the remaining life of each filter with the help of an artificial intelligence model. A maintenance and servicing plan can be formulated based on the predicted remaining life of the filter, and the filter can be replaced or cleaned in a timely manner, which not only ensures the air purification quality, but also improves the utilization rate of the filter, avoids premature replacement of the filter in a relatively clean environment or late replacement in a relatively dirty environment, and improves the working efficiency of the air purifier.

[0098] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A control method for an air purifier, characterized in that: The air purifier includes: A housing (1) with an air inlet (11) and an air outlet (12) provided thereon; A blower (2) disposed inside the housing (1) and opposite to the air outlet (12); A multi-stage filter arranged successively on the gas flow path between the air inlet (11) and the air outlet (12); A detection module disposed on each stage of the filter for detecting air data; A controller connected to the detection module and the control mechanism of the blower (2), the controller being configured to: predict the remaining life prediction result of each stage of the filter and / or control the blower to perform corresponding actions according to the detection result of the detection module; The control method of the air purifier includes the following steps: After the air purifier is turned on and operates, the detection module on each stage of the filter is used to continuously monitor and collect data during the operation of the air purifier; Combining the data collected by all detection modules to predict the remaining life of the filter, obtaining the remaining life prediction result of each stage of the filter, and judging whether the remaining life prediction value of any stage of the filter is greater than the preset replacement life. If so, there is no need to replace the filter. If not, it is prompted that the filter needs to be replaced; And / or Performing comprehensive calculation on the data collected by all detection modules to obtain the air quality parameter Q of the current air, and judging whether Q is greater than the preset value Q0. If so, control the blower to turn on, and if Q gradually decreases within the set time after the blower is turned on, correspondingly control the blower speed to decrease. If Q gradually increases within the set time, correspondingly control the blower speed to increase; if not, stop the air purifier from working; The remaining life prediction of each of the above filters is carried out separately. The specific steps for predicting the remaining life of any one filter are as follows: Step 1: Collect the state data of the filter during operation monitored by the detection module in real time and record it in a time series manner to obtain time series data; Step 2: Preprocess the time series data to obtain the preprocessed time series data, and screen out the features strongly correlated with the performance degradation of the filter from the preprocessed time series data, and perform normalization processing on the screened features to obtain a data set; Step 3: Sort the data set in chronological order, and use the first N data in the data set as the training set, and the remaining other data in the data set as the test set; N is a preset positive integer; Step 4: Extract the attribute values and the corresponding label values of all training samples in the training set and store them as the first feature matrix and the first target vector respectively; Where the first feature matrix is a matrix of size N*n, n corresponding to the total number of features screened in Step 2. All the data in the k-th row of the first feature matrix represent the n attribute values of the k-th training sample in sequence; the first target vector is a matrix of size N*1, and the value of each k-th row in the first target vector corresponds to the performance degradation rate of the k-th training sample; Step 5: Divide the first feature matrix into multiple time windows to obtain multiple time window matrices; Step 6: Construct a performance degradation trajectory model, and use multiple time window matrices in Step 5 to train the constructed performance degradation trajectory model in batches to obtain a trained performance degradation trajectory model; Step 7: Arbitrarily select at least one test sample from the test set, and in the same manner as in Step 4, extract the attribute values and the corresponding label values of each test sample respectively, and store them as a second feature matrix and a second target vector. Input each data in the second feature matrix into the trained performance degradation trajectory model to obtain the performance degradation rate predicted by the attribute value of each test sample; Step 8: Calculate the error between the predicted performance degradation rate value corresponding to each test sample and the true value of the performance degradation rate in the second target vector, and determine whether the error is less than a preset error value. If so, go to Step 10; if not, go to Step 9; Step 9: Go to Step 1 to collect the data real-time monitored by the detection module again; Step 10: Collect the state data during the operation of the filter monitored by the detection module at the current moment, and extract the n attribute values corresponding to the filter at the current moment in the manner of Step 2. Then input the n attribute values corresponding to the filter at the current moment into the performance degradation trajectory model trained in Step 6 to obtain the performance degradation rate of the filter at the current moment; Step 11: Calculate the predicted remaining life value according to the performance degradation rate of the filter at the current moment.

2. The control method according to claim 1, wherein: The detection module is at least one sensor provided on each stage of the filter, and is used to monitor at least the air flow rate, the pressure fluctuation condition, and the air quality monitoring data.

3. The control method according to claim 2, wherein: The detection module on each stage of the filter includes at least one or more of a laser dust sensor capable of tracking the number, size, and aggregation speed of particles, a pressure sensor for measuring air pressure, and an air sensor for obtaining air quality monitoring data.

4. The control method according to claim 1, characterized in that: The filter includes a pre-filter (31), a HEPA filter (32), and a pickling-impregnated activated carbon filter (33) arranged in sequence along the air flow direction on the air flow path.

5. The control method according to any one of claims 1 to 4, characterized in that: It further includes a silencer (4) provided in the housing (1). The silencer (4) is made of porous sound-absorbing material and is arranged in close contact with the inner peripheral wall of the housing (1).

6. The control method according to claim 1, wherein: The degradation trajectory model formula in Step 6 is: Among them, f(t) represents the performance degradation rate of the filter at time t, β0, β i and β ij are all parameters of the degradation trajectory model. β0 represents the intercept term of the degradation trajectory model, and β i represents the influence coefficient of the i-th attribute on the performance degradation of the filter, and β ij represents the interaction coefficient between the i-th attribute and the j-th attribute; x i (t) represents the value of the i-th attribute at time t; x j (t) represents the value of the j-th attribute at time t, and ∈ represents the noise term.

7. The control method according to claim 1, wherein: The calculation formula for the air quality parameter Q is: Among them, N is the concentration of nitrogen dioxide, with the unit of mg / m 3 ; S is the concentration of sulfur dioxide, with the unit of mg / m 3 ; C is the concentration of carbon monoxide, with the unit of mg / m 3 , O is the concentration of ozone, with the unit of mg / m 3 ; P1 is the content of PM10, with the unit of mg / m 3 ; P2 is the content of PM2.5, with the unit of μg / m 3 ; E is the concentration of formaldehyde, with the unit of mg / m 3 ; T is the content of TVOC, with the unit of mg / m 3 .

8. The control method according to claim 7, wherein: The value of the preset value Q0 is 1.

Images