A filter adaptive adjustment method and system based on big data

Abstract

The present invention discloses a filter self-adaptive adjustment method and system based on big data, which relates to the field of filter adjustment. The filter self-adaptive adjustment method based on big data includes the following steps: S1, obtaining standard air parameters and calculating the perfect service life parameters of the filter; S2, extracting features of the standard air parameters and obtaining damage parameters; S3, calculating the primary prediction parameters of the filter life and comparing them, and updating and judging according to the comparison results; S4, obtaining regional air parameters and filter type parameters; S5, calculating the advanced prediction parameters of the filter life and comparing them; S6, analyzing the comparison results according to the normal threshold value and verifying them; S7, matching the adjustment methods and performing adaptive graded adjustment on the filter. The present invention obtains air parameters and filter parameters to perform preliminary and advanced predictions on the service life of the filter, thereby improving the accuracy and efficiency of the prediction.

Description

Technical Field

[0001] The present invention relates to the field of filter adjustment, and in particular to a filter adaptive adjustment method and system based on big data. Background Art

[0002] In modern society, air quality has become an important concern in people's lives, especially in urban environments. The use of air purifiers and their filters has become more and more common. However, how to determine the service life of air filters and when to replace filters is a complex issue. Replacing filters too early will lead to waste of resources, while replacing them too late will affect the air purification effect and may even cause harm to health. Among them, the traditional air filter replacement method is usually based on the manufacturer's recommendation, which is usually preset according to the theoretical life under average usage conditions, but cannot take into account the impact of the actual use environment, such as air quality, frequency of use, etc.

[0003] Big data generally refers to huge data sets that cannot be captured, managed, analyzed, and processed using conventional software tools. Big data analysis can be used to discover hidden patterns and relationships that cannot be discovered by manual analysis methods, and to gain deeper insights. Big data analysis can also help companies develop new products and services, and reduce corporate costs by optimizing operations and improving efficiency.

[0004] However, when the existing filter adaptive adjustment method and system based on big data are in use, they only predict the life of the filter through basic calculations, and are unable to predict the life according to the environmental conditions when the filter is in use and the characteristics of the filter's own material. As a result, the existing filter adaptive adjustment method and system based on big data cannot accurately predict the life of the filter, and the existing filter adaptive adjustment method and system based on big data are mostly static predictions, which cannot perform timely parameter updates according to environmental changes during use, resulting in the inability to perform timely updates and adjustments when setting the adaptive adjustment plan for the filter plate, which greatly affects the use efficiency of the filter adaptive adjustment method and system based on big data.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a filter adaptive adjustment method and system based on big data, which has the advantage of improving the accuracy of filter life prediction, thereby solving the problem of low accuracy in filter life prediction.

[0007] In order to achieve the above-mentioned advantage of improving the accuracy of filter life prediction, the specific technical solution adopted by the present invention is as follows:

[0008] According to one aspect of the present invention, a filter adaptive adjustment method based on big data includes the following steps:

[0009] S1. Obtain standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter;

[0010] S2. Preset damage characteristic values, and perform feature extraction on standard air parameters according to the damage characteristic values ​​to obtain damage parameters;

[0011] S3, calculating the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, and comparing them with the perfect service life parameters of the filter, and updating the comparison results according to the update threshold;

[0012] S4. Obtaining regional air parameters and filter type parameters according to the updated standard air parameters and filter parameters;

[0013] S5. Calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life;

[0014] S6. Preset a normal threshold, analyze the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the normal threshold, and verify the advanced prediction parameter of the filter life according to the analysis result;

[0015] S7. Preset an adaptive graded adjustment method, match the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and perform adaptive graded adjustment on the filter according to the matching result.

[0016] As a preferred solution, the primary prediction parameters of the filter life are calculated according to the damage parameters and the filter parameters, and compared with the perfect service life parameters of the filter. The comparison results are updated according to the update threshold, including the following steps:

[0017] S31, normalizing the damage parameter and the filter parameter;

[0018] S32, constructing a primary prediction parameter model using the processed damage parameters and filter parameters through a neural network model, and calculating primary prediction parameters according to the primary prediction parameter model;

[0019] S33, comparing the primary prediction parameters with the perfect service life parameters of the filter by calculating the percentage difference;

[0020] S34, presetting an update threshold and an update rule, and performing an update judgment based on the comparison result of the primary prediction parameter and the perfect service life parameter of the filter according to the update threshold;

[0021] S35. Match the update rules according to the update judgment result, and execute the matching update rules on the primary prediction parameters.

[0022] As a preferred solution, calculating the advanced prediction parameters of filter life according to regional air parameters and filter type parameters, and comparing the advanced prediction parameters of filter life with the primary prediction parameters of filter life includes the following steps:

[0023] S51, obtaining regional air parameters, and comparing the regional air parameters with standard air parameters to obtain regional air characteristics;

[0024] S52, obtaining filter type parameters, and obtaining filter distinguishing features according to the filter type parameters and filter parameters;

[0025] S53, calculating a damage change value according to regional air characteristics and filter distinguishing characteristics, and updating the damage parameter according to the damage change value;

[0026] S54, calculating the advanced prediction parameters of the filter life according to the updated damage parameters and filter type parameters, and comparing the advanced prediction parameters of the filter life with the primary prediction parameters of the filter life.

[0027] As a preferred solution, calculating the damage change value according to the regional air characteristics and the filter distinguishing characteristics, and updating the damage parameter according to the damage change value includes the following steps:

[0028] S531. Weights are allocated based on regional air characteristics and filter characteristics;

[0029] S532, calculating the regional air characteristic value and the filter distinguishing characteristic value according to the weight distribution results of the regional air characteristics and the filter distinguishing characteristics, and constructing a damage change model in combination with a fatigue analysis algorithm;

[0030] S533, training the damage change model, and verifying the trained damage change model;

[0031] S534, substituting the regional air characteristic value and the filter difference characteristic value into the verified damage change model to calculate the damage change value;

[0032] S535. Update the damage parameter according to the damage change value.

[0033] As a preferred solution, the calculation formula for constructing the damage change model in combination with the fatigue analysis algorithm is: ;

[0034] in, P is the filter damage change rate;

[0035] Y is the activation function;

[0036] G Regional air characteristic value K The weight value of

[0037] K is the regional air characteristic value;

[0038] Regional air characteristic value K The shape parameters of

[0039] L Distinguish the characteristic value for the filter;

[0040] m Distinguishing characteristic values ​​for the filter L The weight value of .

[0041] As a preferred solution, a normal threshold is preset, and the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life is analyzed according to the predicted normal threshold, and the advanced prediction parameter of the filter life is verified according to the analysis result, including the following steps:

[0042] S61, presetting a normal threshold value according to the updated standard air parameters and filter parameters;

[0043] S62, performing difference calculation on the advanced prediction parameter of filter life and the primary prediction parameter of filter life, and performing comparative analysis on the difference calculation result and the normal threshold value;

[0044] S63. Evaluate and verify the advanced prediction parameters of filter life according to the analysis results;

[0045] S64. Adjust the advanced prediction parameters of the filter life according to the evaluation and verification results.

[0046] As a preferred solution, an adaptive graded adjustment method is preset, the verified advanced prediction parameters of filter life are matched with the adaptive graded adjustment method, and the filter is adaptively graded adjusted according to the matching results, including the following steps:

[0047] S71, presetting filter parameter classification standards, and performing dynamic threshold allocation according to the filter classification standards;

[0048] S72, dynamically thresholding the advanced prediction parameters of filter life through a machine learning model;

[0049] S73, presetting an adjustment scheme, matching the adjustment scheme according to the dynamic threshold classification result, and adjusting the filter according to the matched adjustment scheme;

[0050] S74, collecting the parameters of the filter after adjustment for real-time feedback, and updating the adjustment plan based on the feedback results.

[0051] As a preferred solution, presetting an adjustment scheme, matching the adjustment scheme according to the dynamic threshold classification result, and adjusting the filter according to the matched adjustment scheme include the following steps:

[0052] S731, defining an adjustment target, and presetting an adjustment strategy according to the adjustment target;

[0053] S732, establishing adjustment rules according to the adjustment strategy, and presetting adjustment parameters;

[0054] S733, dynamically calculating the dynamic threshold of the filter according to the machine learning model, and matching the dynamic threshold of the filter with the adjustment strategy;

[0055] S734: Execute adjustment operations according to the matched adjustment strategy.

[0056] As a preferred solution, collecting the adjusted parameters of the filter screen for real-time feedback and updating the adjustment scheme according to the feedback results include the following steps:

[0057] S741, obtaining the adjusted parameters of the filter, and constructing a visualization interface according to the feedback results;

[0058] S742, performing data analysis based on the visualization interface constructed according to the feedback results, obtaining the change trend of the filter, and updating the adjustment strategy according to the change trend;

[0059] S743, output the updated adjustment strategy, and store the filter change trend and adjustment strategy to update parameters.

[0060] According to another aspect of the present invention, a filter adaptive adjustment system based on big data comprises:

[0061] Parameter acquisition module, used to obtain standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter;

[0062] A damage parameter module is used to preset damage characteristic values, and extract characteristics of standard air parameters according to the damage characteristic values ​​to obtain damage parameters;

[0063] The primary prediction module is used to calculate the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, and compare them with the perfect service life parameters of the filter, and update the comparison results according to the update threshold;

[0064] A parameter updating module, used to obtain regional air parameters and filter type parameters according to updated standard air parameters and filter parameters;

[0065] An advanced prediction module is used to calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life;

[0066] An analysis and verification module is used to preset a normal threshold, analyze the comparison results of the advanced prediction parameters of the filter life and the primary prediction parameters of the filter life according to the normal threshold, and verify the advanced prediction parameters of the filter life according to the analysis results;

[0067] The adaptive adjustment module presets an adaptive graded adjustment method, matches the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and performs adaptive graded adjustment on the filter according to the matching results.

[0068] Compared with the prior art, the present invention provides a filter self-adaptive adjustment method and system based on big data, which has the following beneficial effects:

[0069] (1) The present invention obtains standard air parameters and filter parameters and combines them with preset damage characteristic values ​​to make preliminary and advanced predictions on the service life of the filter, thereby improving the accuracy and efficiency of the prediction. At the same time, through the preset normal threshold and adaptive graded adjustment method, the filter is dynamically adjusted according to the real-time prediction results and actual conditions, thereby improving the use efficiency and service life of the filter.

[0070] (2) The present invention collects the parameters of the adjusted filter for real-time feedback, performs data analysis through a visual interface, and updates the adjustment plan based on the feedback results, thereby improving the system's adaptability and user experience. In addition, through intelligent prediction and dynamic adjustment, the filter is used reasonably and effectively, the service life of the filter is extended, the cost and time of replacing the filter are saved, and resources are saved. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0072] Figure 1 is a method flow chart of a filter adaptive adjustment method based on big data according to an embodiment of the present invention;

[0073] Figure 2 4 is a system block diagram of a filter adaptive adjustment system based on big data according to an embodiment of the present invention.

[0074] In the figure:

[0075] 1. Parameter acquisition module; 2. Damage parameter module; 3. Primary prediction module; 4. Parameter update module; 5. Advanced prediction module; 6. Analysis and verification module; 7. Adaptive adjustment module. DETAILED DESCRIPTION

[0076] The present invention is further described with reference to the accompanying drawings and specific embodiments. According to one embodiment of the present invention, Figure 1 As shown, the filter adaptive adjustment method based on big data according to an embodiment of the present invention includes the following steps:

[0077] S1. Obtain standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter;

[0078] Specifically, standard air parameters are obtained, and air quality monitors or sensors are used to obtain various parameters in the air, such as PM2.5, PM10, CO2, temperature, humidity, etc., to monitor air quality in real time, generate data for further analysis, and obtain filter parameters through the technical specifications provided by the manufacturer, such as filtration efficiency, filtration area, maximum air volume, etc. In addition, the real-time status of the filter is obtained through sensors installed on the filter, such as filter resistance, particle concentration, etc., and the filter service life parameters are calculated. Based on the obtained standard air parameters and filter parameters, according to the average service life specified by the filter manufacturer, and the current air quality and usage, a preliminary prediction is made, and then combined with the real-time status of the filter, the preliminary prediction is corrected. Based on the prediction results, the use of the filter is dynamically adjusted, such as adjusting the filtration speed, replacing the filter, etc., to extend the service life of the filter and improve efficiency.

[0079] S2. Preset damage characteristic values, and perform feature extraction on standard air parameters according to the damage characteristic values ​​to obtain damage parameters;

[0080] Specifically, the preset damage characteristic values ​​reflect the air quality indicators that may cause damage to machines, human bodies or the environment under specific environmental conditions, such as excessive temperature, low humidity, excessive PM2.5 concentration, etc. These values ​​need to be clearly defined and understood. According to the preset damage characteristic values, relevant features are extracted from the standard air parameters. For example, if the PM2.5 concentration is a damage characteristic value, the value of the PM2.5 concentration is extracted from the standard air parameters, and the standard air parameters are feature extracted based on these characteristic values ​​to obtain the damage parameters.

[0081] S3, calculating the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, and comparing them with the perfect service life parameters of the filter, and updating the comparison results according to the update threshold;

[0082] In the embodiment of the present application, the primary prediction parameters of the filter life are calculated according to the damage parameters and the filter parameters, and compared with the perfect service life parameters of the filter, and the comparison result is updated according to the update threshold value, including the following steps:

[0083] S31, normalizing the damage parameter and the filter parameter;

[0084] Specifically, determine the range of the parameters. For each parameter, record the minimum and maximum values ​​that it may take, that is, the parameter range, and standardize the parameters. Standardization is to map the parameter value to the range of 0-1. The formula is:

[0085] The standardized parameter value = (original parameter value - minimum parameter value) / (maximum parameter value - minimum parameter value);

[0086] Normalize the parameters, and normalization is to map the parameter values ​​to the range of -1 to 1. The formula is:

[0087] Normalized parameter value = 2*(normalized parameter value)-1;

[0088] Save the minimum and maximum values ​​used for normalization so that they can be used later when the parameter values ​​need to be restored. Perform the above standardization and normalization on each sample in the data set to obtain a normalized sample data set. Use the normalized sample data set during model training, and restore the prediction results to the actual value range during model prediction. The restoration formula is:

[0089] The restored parameter value = minimum parameter value + (maximum parameter value - minimum parameter value) * [(normalized parameter value + 1) / 2];

[0090] S32, constructing a primary prediction parameter model using the processed damage parameters and filter parameters through a neural network model, and calculating primary prediction parameters according to the primary prediction parameter model;

[0091] Specifically, a neural network model structure is constructed, such as using a convolutional neural network, setting the input layer as a damage parameter and a filter parameter vector, multiple convolutional layers extract features, and the last layer is an output layer to predict primary parameters, collect damage parameters, filter parameters and corresponding primary parameters as training data, and pre-process the training data, such as normalizing the damage parameters and filter parameters;

[0092] Initialize the network model parameters, input training data, forward propagate to calculate the output, calculate the error between the actual primary parameters, back propagate to calculate the parameter update value, use the gradient descent method to optimize the network parameters, minimize the error, and then iterate the training. Repeat forward propagation-backward propagation in each round, and continuously update the network parameters until convergence. After the training is completed, save the trained network model;

[0093] When it is necessary to predict the primary parameters, new damage parameters and filter parameters are input, and the network output is calculated by forward propagation, i.e., the primary prediction parameters. The primary prediction parameters are post-processed, such as restoring the normalized range, and the primary prediction parameter model is continuously improved and optimized through repeated predictions.

[0094] S33, comparing the primary prediction parameters with the perfect service life parameters of the filter by calculating the percentage difference;

[0095] Specifically, the perfect service life parameter data of the filter is collected, and then the neural network model is used to predict the primary prediction parameters to determine whether the percentage difference is within an acceptable range, such as ±5%. If |percentage difference|≤5%, it indicates that the primary prediction parameters have a high matching degree and good prediction accuracy.

[0096] If |percentage difference|>5%, it indicates that the primary prediction parameters do not match well and the prediction model needs to be further optimized;

[0097] Count the percentage difference distribution of all samples, observe the prediction accuracy, and further analyze which types of samples have poor prediction results based on the difference distribution, improve the model in a targeted manner, repeat the above process, continuously collect new data, optimize the model, reduce the average percentage difference, and improve the accuracy of the primary prediction parameters.

[0098] S34, presetting an update threshold and an update rule, and performing an update judgment based on the comparison result of the primary prediction parameter and the perfect service life parameter of the filter according to the update threshold;

[0099] Specifically, an update threshold is preset, for example, a percentage difference threshold is set to 10%, and an update rule is preset. If the percentage difference is less than or equal to the threshold (≤10%), the primary prediction parameters do not need to be updated; if the percentage difference is greater than the threshold (>10%), the primary prediction parameters need to be updated;

[0100] Compare all samples and calculate the percentage difference. Determine whether an update is needed based on the rules. If the difference is ≤10%, the primary prediction parameters remain unchanged. If the difference is >10%, the primary prediction parameters need to be updated. The method for updating the primary prediction parameters is to directly use the actual life value of the filter to overwrite the predicted value, or use the average of the filter value and the predicted value as the new predicted value. Repeat the above process, continuously collect new data for update judgment and parameter update, and appropriately adjust the threshold so that more samples meet the update conditions, speed up the model optimization, and as the sample size and model optimization proceed, the threshold is gradually reduced to improve the prediction accuracy.

[0101] S35. Match the update rules according to the update judgment result, and execute the matching update rules on the primary prediction parameters.

[0102] Specifically, different update rules are preset, such as directly using the actual filter value to cover the predicted value during direct update, using the average of the filter value and the predicted value as the new predicted value during average update, using the linear combination of the filter value and the predicted value as the new predicted value during linear update, and using the nonlinear function of the filter value and the predicted value as the new predicted value during nonlinear update;

[0103] Calculate the percentage difference between the primary prediction parameter and the true value of the filter, and determine whether an update is needed based on the difference. If an update is needed, obtain the corresponding update rule. If it is a direct update, directly overwrite the predicted value with the filter value. If it is an average update, calculate the average of the filter value and the predicted value. If it is a linear update, calculate the linear combination value. If it is a nonlinear update, calculate the nonlinear function value. Use the calculation result as the new primary prediction parameter of the sample. Repeat the above process to update all samples that need to be updated, record the parameters before and after the update, count the update effect, and adjust the update rules according to the effect to optimize the update strategy.

[0104] S4. Obtaining regional air parameters and filter type parameters according to the updated standard air parameters and filter parameters;

[0105] Specifically, collect sample data, record the region and filter type labels corresponding to each sample, preprocess the sample data, such as normalization, divide the sample data into a training data set and a test data set, build a classification model, such as a decision tree model, SVM model, etc., use the training data set to train the classification model, learn the mapping relationship between sample features and region / filter type, perform classification prediction on the test data set, obtain the predicted region and filter type labels, calculate the classification accuracy, and evaluate the model effect. If the accuracy is not satisfactory, adjust the model structure or optimize the algorithm parameters to continue training.

[0106] When there are new standard air parameter and filter parameter samples, the new sample data is preprocessed and input into the trained classification model. The model makes predictions based on the learned rules and outputs the predicted region and filter type labels. It also directly extracts the region and filter information from the parameters according to certain rules. The final obtained region and filter type labels are used for subsequent analysis, such as statistics on air quality in different regions, usage of different filters, etc.

[0107] S5. Calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life;

[0108] In the embodiment of the present application, calculating the advanced prediction parameter of filter life according to the regional air parameter and the filter type parameter, and comparing the advanced prediction parameter of filter life with the primary prediction parameter of filter life includes the following steps:

[0109] S51, obtaining regional air parameters, and comparing the regional air parameters with standard air parameters to obtain regional air characteristics;

[0110] Specifically, collect air quality data of different regions over a period of time, such as PM2.5 concentration, etc., collect statistics on the air quality data of each region, calculate the average value, standard deviation and other indicators as the air parameter characteristic value of the region, and when there is new sample air parameter data, the corresponding regional label has been obtained. According to the sample regional label, the air parameter characteristic value of the region is obtained from the pre-stored regional air parameter characteristic value set, and then the sample air parameter is compared with the regional air parameter characteristic value. If the sample value is within a certain range of the regional average value, it indicates that the sample meets the regional characteristics;

[0111] If the sample value exceeds the regional range, it may indicate that the sample is abnormal or the regional characteristic value needs to be updated. The degree of match between all samples and the corresponding regional characteristic values ​​is counted, the validity of the current regional characteristic values ​​is evaluated, and the air parameter characteristic values ​​of each region are continuously improved and updated based on new samples. The difference in air quality between the sample and the corresponding region is obtained by comparison, that is, the regional air characteristics are obtained.

[0112] S52, obtaining filter type parameters, and obtaining filter distinguishing features according to the filter type parameters and filter parameters;

[0113] Specifically, technical parameters of different types of filters, such as filtration efficiency, service life, etc., are collected as filter type parameter feature value sets. When there is new filter sample data, the corresponding filter type label has been obtained. According to the sample filter type label, the feature value of the type is obtained from the pre-stored filter type parameter feature value set, and then the actual parameters of the sample filter are compared with the feature value of the type of filter parameter. If the sample value is similar to the feature value, it indicates that the sample belongs to this type. If the difference is large, it may be a new type or the sample is abnormal and the feature value needs to be updated. The consistency of all samples with the feature value of the corresponding type is counted, and the effect of the current feature value is evaluated. According to the inconsistent samples, the parameter feature values ​​of various types of filters are improved and updated, and the parameter feature values ​​of different types of filters are compared to obtain their distinguishing features. For example, type Q has a short life, and type W has a medium efficiency but a long life. By analyzing the advantages and disadvantages of different types of filters in various technical parameters, their distinguishing features can be obtained.

[0114] S53, calculating a damage change value according to regional air characteristics and filter distinguishing characteristics, and updating the damage parameter according to the damage change value;

[0115] In the embodiment of the present application, calculating the damage change value according to the regional air characteristics and the filter distinguishing characteristics, and updating the damage parameter according to the damage change value includes the following steps:

[0116] S531. Weights are allocated based on regional air characteristics and filter characteristics;

[0117] Specifically, collect the characteristic values ​​of air quality in different regions and the characteristic values ​​of technical parameters of different filter types, determine the importance of each characteristic value, and judge through expert evaluation or historical data analysis that some characteristic values ​​have a greater impact on the results. For example, PM2.5 concentration has the greatest impact on air quality, so it has the largest weight, and the service life of the filter has an important impact on the selection, so it has a high weight. Give each feature a relative weight, give important features a larger weight, such as 0.3-0.5, give general features a medium weight, such as 0.1-0.3, and give features that have a smaller impact on the results a small weight, such as 0.01-0.1. When distributing, the sum of the weights is 1. At the same time, the more evenly the weight distribution is distributed, the better, and consider the correlation between features to avoid repeated calculations;

[0118] Adjust the weights based on the new data feedback results. When the influence of important features increases, increase the weights accordingly. Record the weight allocation results for subsequent matching analysis.

[0119] S532, calculating the regional air characteristic value and the filter distinguishing characteristic value according to the weight distribution results of the regional air characteristics and the filter distinguishing characteristics, and constructing a damage change model in combination with a fatigue analysis algorithm;

[0120] In the embodiment of the present application, the calculation formula for constructing the damage change model in combination with the fatigue analysis algorithm is:

[0121] ;

[0122] in, P is the filter damage change rate;

[0123] Y is the activation function;

[0124] G Regional air characteristic value K The weight value of

[0125] K is the regional air characteristic value;

[0126] Regional air characteristic value K The shape parameters of

[0127] L Distinguish the characteristic value for the filter;

[0128] m Distinguishing characteristic values ​​for the filter L The weight value of .

[0129] S533, training the damage change model, and verifying the trained damage change model;

[0130] Specifically, collect damage change data, including characteristic data such as damage cause, time, and damage degree, divide the data into training data sets and test data sets, select appropriate machine learning algorithms, such as decision trees, neural networks, etc. for model training, input features into the training data set, output the damage degree, train the model, and learn the mapping relationship between data features and damage degree. After training, calculate the training accuracy and evaluate the effect of the model on the training data. Input the test data set into the trained model for prediction, calculate the test accuracy, evaluate the generalization ability of the model on unseen data, compare the training accuracy and the test accuracy. If the difference is large, it may indicate overfitting. Adjust the model complexity to avoid overfitting, increase regularization terms or add dropout layers to optimize the model, continue to adjust the model on the test set, find the model with the strongest generalization ability, record the best model structure and parameters, and when there is new real data, use the recorded best model for prediction. Continuously collect new data for retraining to improve the model effect.

[0131] S534, substituting the regional air characteristic value and the filter difference characteristic value into the verified damage change model to calculate the damage change value;

[0132] Specifically, obtain the characteristic values ​​of air quality in a specific area, such as the average PM2.5 concentration, etc., obtain the characteristic values ​​of technical parameters of a certain type of filter, such as filtration efficiency, service life, etc., combine the characteristic values ​​into a characteristic vector, and use the characteristic vector as input to the trained and verified damage change model. The damage change model gives a predicted damage degree output based on the mapping relationship between the learned characteristic values ​​and the damage degree. The output damage degree is the predicted value of the filter damage considering the air quality of the area and the technical parameters of the filter. Repeat the above process for different regions and different filter models to obtain damage prediction values ​​for multiple region-filter combinations. According to the prediction results, evaluate the severity of filter damage under different conditions to provide a reference for selecting filters. When there is new air quality data or filter parameters, repeat the calculation to update the prediction results.

[0133] S535. Update the damage parameter according to the damage change value.

[0134] Specifically, continue to collect a new batch of regional air quality data and filter usage data, and repeatedly calculate the damage prediction values ​​for different region-filter combinations based on the new data, and compare the actual damage situation with the model prediction value. If the actual damage value is significantly different from the predicted value, the model parameters need to be adjusted and the damage parameters need to be updated. The new data may also reflect new damage influencing factors. Analyze the causes of the prediction errors, such as whether important influencing features are missing, etc. On the basis of the original training data, add the new influencing features to retrain the model, or directly use part of the new data to retrain the model. The new round of training learns new damage rules, updates the damage parameters, and uses the updated damage model to test the effect on a new batch of verification data. If the effect is improved, it indicates that the damage parameters have been successfully updated. If the effect is poor, it is necessary to continue to adjust the model structure or parameters, and continue to repeat the above process to continuously optimize the damage model with the support of growing new data.

[0135] S54, calculating the advanced prediction parameters of the filter life according to the updated damage parameters and filter type parameters, and comparing the advanced prediction parameters of the filter life with the primary prediction parameters of the filter life.

[0136] Specifically, updated damage parameters are obtained, and a damage model that considers more influencing factors is used, and then technical parameters of various types of filters are obtained, such as filtration efficiency, usage environment, etc. According to the damage model, the expected damage values ​​of different filters under different usage conditions are calculated. According to the filter technical parameters and the expected damage values, numerical simulation or machine learning methods are used to predict the life of each filter. These life prediction values ​​calculated based on the updated damage parameters and filter parameters are the advanced life prediction parameters, which are compared with the primary life prediction parameters previously trained based on the old parameters. If the advanced prediction values ​​are more consistent with the actual ones, it means that the updated parameters take into account more comprehensive factors and have better prediction effects. The advanced prediction parameters are used instead of the primary prediction parameters. If the effect difference is not obvious, it is necessary to continue to optimize the damage model to obtain better parameters. It is also possible that the primary prediction parameters are already very effective. New data should be continuously collected to repeat the above process to continuously improve the prediction accuracy.

[0137] S6. Preset a normal threshold, analyze the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the normal threshold, and verify the advanced prediction parameter of the filter life according to the analysis result;

[0138] In the embodiment of the present application, a normal threshold is preset, and the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life is analyzed according to the predicted normal threshold, and the advanced prediction parameter of the filter life is verified according to the analysis result, including the following steps:

[0139] S61, presetting a normal threshold value according to the updated air parameters and filter parameters;

[0140] Specifically, collect historical data on air quality parameters in different regions, such as PM2.5 concentration and ozone concentration, and collect historical data on technical parameters of different filter types, such as filtration efficiency and service life. Set a normal range for each parameter based on historical data, set a normal range for standard air parameters with reference to local air quality standards, and set a normal range for filter parameters with reference to filter product technical indicators. Continue to collect a new batch of air quality and filter usage data, and update the impact assessment model of each parameter based on the new data. On the basis of the old normal range, adjust the parameter range according to the degree of impact. The range of parameters with a large impact will be relatively narrowed, and the range of parameters with a small impact will be appropriately expanded. The newly set range will be used as the preset normal threshold range. Collect new data periodically and adjust the threshold range in time according to actual conditions. If the data exceeds the existing range, the threshold range can be appropriately expanded to keep the threshold range dynamically adjusted to be closer to the actual situation.

[0141] S62, performing difference calculation on the advanced prediction parameter of filter life and the primary prediction parameter of filter life, and performing comparative analysis on the difference calculation result and the normal threshold value;

[0142] Specifically, the advanced prediction parameters and primary prediction parameter values ​​of the life of each filter are obtained. For each filter, the difference between its advanced prediction parameter value and the primary prediction parameter value is calculated, and the positivity of the difference is analyzed. A positive value indicates that the advanced prediction value is greater than the primary prediction value, and a negative value indicates that the advanced prediction value is less than the primary prediction value. The differences of all filters are counted, and the distribution of the differences is observed. The difference range is compared with the preset normal threshold range. If the differences are mainly concentrated in the normal threshold range, it indicates that the prediction parameter update effect is ideal. If some differences exceed the normal threshold, it is necessary to further analyze the reasons. For filters with excessively large differences, analyze the differences in their technical parameters and usage conditions. Based on the comparison results, determine whether the parameter update has well reflected the actual situation and where further optimization is needed. Continuously collect new data and repeat the process to monitor whether the prediction accuracy is improving.

[0143] S63. Evaluate and verify the advanced prediction parameters of filter life according to the analysis results;

[0144] Specifically, collect a batch of actual usage data of filters, including usage time and scrap time, etc., take the actual scrap time as the standard, compare it with the advanced life prediction parameters, calculate the prediction error rate: the percentage of the error between the actual scrap time and the predicted value to the predicted value, count the prediction error rates of all filters, observe the distribution, and according to the error rate distribution, if the error rate of most filters is within a certain range, it indicates that the prediction effect is good. If the error rate of some filters is too large, it is necessary to further analyze the reasons. For filters with too large error rates, analyze the particularity of their technical parameters or usage conditions;

[0145] Based on the comparison results, determine whether the prediction parameters need further adjustment. Repeat the above process to compare the effects of prediction parameters at different stages, observe whether the accuracy is improving, continue to collect new actual data for verification, and dynamically optimize the prediction model.

[0146] S64. Adjust the advanced prediction parameters of the filter life according to the evaluation and verification results.

[0147] S7. Preset an adaptive graded adjustment method, match the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and perform adaptive graded adjustment on the filter according to the matching result.

[0148] In the embodiment of the present application, an adaptive graded adjustment method is preset, the verified advanced prediction parameters of the filter life are matched with the adaptive graded adjustment method, and the adaptive graded adjustment of the filter is performed according to the matching result, including the following steps:

[0149] S71, presetting filter parameter classification standards, and performing dynamic threshold allocation according to the filter classification standards;

[0150] Specifically, based on the main technical parameters of the filter, such as filtration efficiency and service life, a preliminary filter grade standard is designed, and then each filter is initially graded according to the grade standard. For example, actual usage data of the filter is continuously collected, and the parameter values ​​and usage are recorded. According to historical data, statistics are conducted on each technical parameter for each grade group to obtain the parameter distribution characteristics of the filter at that grade. On this basis, the threshold range for each grade is dynamically set. The high-level threshold range is relatively loose, and the low-level range is relatively strict. The threshold range is continuously fine-tuned according to actual data. When a new filter is online, the grade is automatically determined according to its parameters and the threshold range is recorded. The parameters are monitored in real time according to the threshold range to see if they are abnormal and need repair or replacement. The grade standards and thresholds are continuously improved to make them more in line with actual conditions.

[0151] S72, dynamically thresholding the advanced prediction parameters of filter life through a machine learning model;

[0152] Specifically, historical filter usage data is collected, including various technical parameters and actual service life data, and the data is hierarchically clustered using a machine learning algorithm. Based on the clustering results, the threshold range for each level is preliminarily determined. New data is continuously collected, and the model is retrained regularly to update parameters. After each training, the new and old cluster centers and ranges are compared to observe whether they have converged stably. When the clustering results are basically stable, the cluster centers and ranges of this result are used as dynamic thresholds for each level. For new filter data, the level is automatically determined according to the trained model. After continuously collecting new data, training and threshold updates are performed again. At the same time, the actual scrapping time of each filter is recorded and compared with the predicted life. Based on the error distribution, it is determined whether the threshold range needs to be adjusted or the model needs to be retrained. Through continuous iterative training and verification, the threshold range can truly reflect the characteristics of filters of each level.

[0153] S73, presetting an adjustment scheme, matching the adjustment scheme according to the dynamic threshold classification result, and adjusting the filter according to the matched adjustment scheme;

[0154] In the embodiment of the present application, a preset adjustment scheme is used, a matching adjustment scheme is performed according to the dynamic threshold classification result, and the filter adjustment is performed according to the matching adjustment scheme, including the following steps:

[0155] S731, defining an adjustment target, and presetting an adjustment strategy according to the adjustment target;

[0156] Specifically, define the adjustment target. Ensure that the filter efficiency is within a reasonable range, extend the service life of the filter, and reduce the cost of filter replacement. According to the target, design the corresponding adjustment strategy. Clean the filter regularly according to the use time of the filter, remove the surface dirt, maintain the filtration efficiency, monitor the health status of the filter according to the filter parameters, and determine whether it is necessary to replace the parts in advance to extend the service life. Dynamically adjust the replacement cycle according to the filter level. The high-level filter cycle is longer. Set quantitative indicators for each strategy, cleaning cycle, efficiency drop threshold, component replacement standard, and replacement cycle of each level of filter. Continuously collect data to evaluate the effectiveness of each strategy, whether the filtration efficiency is within the target range, whether the service life is extended, and whether the replacement cost is reduced. According to the evaluation results, timely optimize the adjustment strategy and indicator settings, and regularly evaluate whether the target needs to be adjusted.

[0157] S732, establishing adjustment rules according to the adjustment strategy, and presetting adjustment parameters;

[0158] Specifically, establish clear adjustment rules for each adjustment strategy, set cleaning rules, specify cleaning cycles and efficiency drop thresholds, set replacement rules, specify component replacement standards and replacement cycles for filters of various levels, define necessary parameters for each rule, such as cleaning cycle: 1 month / 2 months, efficiency drop threshold: 5% / 10%, etc., component replacement standards: usage time / running hours, etc., replacement cycle: 6 months for high level, 3 months for medium level, 1 month for low level, etc., set initial values ​​for each parameter based on historical data, and set them with reference to the actual usage of most filters, clarify the adjustment frequency of rules and parameters, such as evaluation every 3 / 6 months, set a decision-making process for rule and parameter adjustment, data collection and analysis, evaluation results, improvement suggestions, decision confirmation, add rules and parameters to the management system and use continuous data collection to optimize rules and parameters.

[0159] S733, dynamically calculating the dynamic threshold of the filter according to the machine learning model, and matching the dynamic threshold of the filter with the adjustment strategy;

[0160] Specifically, a prediction model is trained using a machine learning algorithm to predict the usage status of the filter based on the technical parameters. The model is continuously verified during training, and the model with the highest accuracy is selected. The trained model is used to predict the new filter to obtain the dynamic threshold range of each technical parameter. The dynamic threshold ranges of various technical parameters are integrated into the overall dynamic threshold level of the filter. The dynamic threshold level is matched with the preset adjustment strategy, such as a maintenance strategy for a high level, a partial replacement strategy for a medium level, and a full replacement strategy for a low level. New data is continuously collected to retrain the model, the dynamic threshold calculation is optimized, and the matching effect is evaluated based on the actual filter usage. If necessary, the strategy settings are adjusted, and the calculation and matching process is written into the management system for automatic execution.

[0161] S734: Execute adjustment operations according to the matched adjustment strategy.

[0162] S74, collecting the parameters of the filter after adjustment for real-time feedback, and updating the adjustment plan based on the feedback results.

[0163] In the embodiment of the present application, collecting the parameters after the filter adjustment for real-time feedback, and updating the adjustment scheme according to the feedback results include the following steps:

[0164] S741, obtaining the adjusted parameters of the filter, and constructing a visualization interface according to the feedback results;

[0165] Specifically, after the filter is adjusted, the technical parameter data is obtained in real time through the monitoring equipment, the parameter data is uploaded to the background database for storage, a data acquisition interface is developed, the interface is called regularly to obtain the latest parameter data, and the parameter data is classified and summarized according to different adjustment strategies. The changes in various parameters before and after adjustment are compared, and the adjustment effect indicators are calculated. The interface framework is designed using a visual development tool, and dynamic binding and interaction of different parts of the interface are realized. A tree structure is used to select the filter to drive the parameter chart and data update. The time range selection drives the chart data update. It is integrated into the background management system to realize online monitoring and query functions.

[0166] S742, performing data analysis based on the visualization interface constructed according to the feedback results, obtaining the change trend of the filter, and updating the adjustment strategy according to the change trend;

[0167] Specifically, extract the long-term recorded data of each filter parameter from the interface database, pre-process the data, such as outlier processing, missing value filling, etc., apply time series analysis methods, perform trend analysis on the historical data of each parameter, identify its changing pattern, correspond the trends of different parameters, analyze the changing pattern of the overall performance of the filter over time, and evaluate the pros and cons of the current adjustment strategies based on the trend analysis results, whether the strategy can well match the actual changing pattern of the filter, and whether the strategy can effectively delay the decline of the filter performance. According to the evaluation results, put forward strategy optimization suggestions, such as adjusting the cleaning cycle or standard, adding a certain maintenance measure, and feedback the recommended results to the decision maker for deliberation and determination. After determination, update the adjustment strategy and management system, continue to collect new data, and repeat the analysis and optimization process regularly;

[0168] S743, output the updated adjustment strategy, and store the filter change trend and adjustment strategy to update parameters.

[0169] According to another embodiment of the present invention, Figure 2 As shown, a filter adaptive adjustment system based on big data is provided, and the system includes:

[0170] Parameter acquisition module 1, used to obtain standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter;

[0171] Damage parameter module 2, used to preset damage characteristic values, and extract characteristics of standard air parameters according to the damage characteristic values ​​to obtain damage parameters;

[0172] Primary prediction module 3, used to calculate the primary prediction parameters of filter life according to the damage parameters and filter parameters, and compare them with the perfect service life parameters of the filter, and update the comparison results according to the update threshold;

[0173] Parameter updating module 4, used for obtaining regional air parameters and filter type parameters according to the updated standard air parameters and filter parameters;

[0174] Advanced prediction module 5, used to calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life;

[0175] An analysis and verification module 6 is used to preset a normal threshold, analyze the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the normal threshold, and verify the advanced prediction parameter of the filter life according to the analysis result;

[0176] The adaptive adjustment module 7 presets an adaptive graded adjustment method, matches the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and performs adaptive graded adjustment on the filter according to the matching result.

[0177] To summarize, with the aid of the above-mentioned technical scheme of the present invention, the present invention obtains standard air parameters and filter parameters, and combines preset damage characteristic values ​​to make preliminary predictions and advanced predictions on the service life of the filter, thereby improving the accuracy and efficiency of the prediction. At the same time, through the preset normal threshold and adaptive graded adjustment method, the filter is dynamically adjusted according to the real-time prediction results and actual conditions, thereby improving the use efficiency and service life of the filter.

[0178] In addition, the present invention improves the system's adaptability and user experience by collecting adjusted filter parameters for real-time feedback, performing data analysis through a visual interface, and updating adjustment plans based on feedback results. It also uses intelligent prediction and dynamic adjustment to rationally and effectively use the filter, extend the filter's service life, save the cost and time of replacing the filter, and thus save resources.

[0179] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A filter adaptive adjustment method based on big data, characterized in that: The following steps are involved: S1. Obtain standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter; S2. Preset damage characteristic values, and perform feature extraction on standard air parameters according to the damage characteristic values ​​to obtain damage parameters; The preset damage characteristic value represents an air quality index reflecting damage to a machine, human body or environment under specific environmental conditions; S3, calculating the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, and comparing them with the perfect service life parameters of the filter, and updating the comparison results according to the update threshold; S4. Obtaining regional air parameters and filter type parameters according to the updated standard air parameters and filter parameters; S5. Calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life; S6. Preset a normal threshold, analyze the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the normal threshold, and verify the advanced prediction parameter of the filter life according to the analysis result; S7. Preset an adaptive graded adjustment method, match the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and perform adaptive graded adjustment on the filter according to the matching result.

2. The filter adaptive adjustment method based on big data according to claim 1 is characterized in that: The method of calculating the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, comparing them with the perfect service life parameters of the filter, and updating the comparison results according to the update threshold comprises the following steps: S31, normalizing the damage parameter and the filter parameter; S32, constructing a primary prediction parameter model using the processed damage parameters and filter parameters through a neural network model, and calculating primary prediction parameters according to the primary prediction parameter model; S33, comparing the primary prediction parameters with the perfect service life parameters of the filter by calculating the percentage difference; S34, presetting an update threshold and an update rule, and performing an update judgment based on the comparison result of the primary prediction parameter and the perfect service life parameter of the filter according to the update threshold; S35. Match the update rules according to the update judgment result, and execute the matching update rules on the primary prediction parameters.

3. The filter adaptive adjustment method based on big data according to claim 2 is characterized in that: The step of calculating the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and comparing the advanced prediction parameters of filter life with the primary prediction parameters of filter life comprises the following steps: S51, obtaining regional air parameters, and comparing the regional air parameters with standard air parameters to obtain regional air characteristics; S52, obtaining filter type parameters, and obtaining filter distinguishing features according to the filter type parameters and filter parameters; S53, calculating a damage change value according to regional air characteristics and filter distinguishing characteristics, and updating the damage parameter according to the damage change value; S54, calculating the advanced prediction parameters of the filter life according to the updated damage parameters and filter type parameters, and comparing the advanced prediction parameters of the filter life with the primary prediction parameters of the filter life.

4. The filter adaptive adjustment method based on big data according to claim 3 is characterized in that: The step of calculating the damage change value according to the regional air characteristics and the filter distinguishing characteristics, and updating the damage parameter according to the damage change value comprises the following steps: S531. Weights are allocated based on regional air characteristics and filter characteristics; S532, calculating the regional air characteristic value and the filter distinguishing characteristic value according to the weight distribution results of the regional air characteristics and the filter distinguishing characteristics, and constructing a damage change model in combination with a fatigue analysis algorithm; S533, training the damage change model, and verifying the trained damage change model; S534, substituting the regional air characteristic value and the filter difference characteristic value into the verified damage change model to calculate the damage change value; S535. Update the damage parameter according to the damage change value.

5. The filter adaptive adjustment method based on big data according to claim 4 is characterized in that: The calculation formula for constructing the damage change model in combination with the fatigue analysis algorithm is: P=1-Y[-(G / K) a Lm]; Wherein, P is the filter damage change rate; Y is the activation function; G is the weight value of the regional air characteristic value K; K is the regional air characteristic value; a is the shape parameter of the regional air characteristic value K; L is the filter distinguishing characteristic value; m is the weight value of the filter distinguishing eigenvalue L.

6. The filter adaptive adjustment method based on big data according to claim 1 is characterized in that: The preset normal threshold, analyzing the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the predicted normal threshold, and verifying the advanced prediction parameter of the filter life according to the analysis result includes the following steps: S61, presetting a normal threshold value according to the updated standard air parameters and filter parameters; S62, performing difference calculation on the advanced prediction parameter of filter life and the primary prediction parameter of filter life, and performing comparative analysis on the difference calculation result and the normal threshold value; S63. Evaluate and verify the advanced prediction parameters of filter life according to the analysis results; S64. Adjust the advanced prediction parameters of the filter life according to the evaluation and verification results.

7. The filter adaptive adjustment method based on big data according to claim 1 is characterized in that: The preset adaptive graded adjustment method matches the verified advanced prediction parameters of filter life with the adaptive graded adjustment method, and adaptively adjusts the filter according to the matching result, including the following steps: S71, presetting filter parameter classification standards, and performing dynamic threshold allocation according to the filter classification standards; S72, dynamically thresholding the advanced prediction parameters of filter life through a machine learning model; S73, presetting an adjustment scheme, matching the adjustment scheme according to the dynamic threshold classification result, and adjusting the filter according to the matched adjustment scheme; S74, collecting the parameters of the filter after adjustment for real-time feedback, and updating the adjustment plan based on the feedback results.

8. The filter adaptive adjustment method based on big data according to claim 7 is characterized in that: The preset adjustment scheme, matching the adjustment scheme according to the dynamic threshold classification result, and adjusting the filter according to the matching adjustment scheme include the following steps: S731, defining an adjustment target, and presetting an adjustment strategy according to the adjustment target; S732, establishing adjustment rules according to the adjustment strategy, and presetting adjustment parameters; S733, dynamically calculating the dynamic threshold of the filter according to the machine learning model, and matching the dynamic threshold of the filter with the adjustment strategy; S734: Execute adjustment operations according to the matched adjustment strategy.

9. The filter adaptive adjustment method based on big data according to claim 7 is characterized in that: The collecting filter adjusted parameters for real-time feedback and updating the adjustment scheme according to the feedback results include the following steps: S741, obtaining the adjusted parameters of the filter, and constructing a visualization interface according to the feedback results; S742, performing data analysis based on the visualization interface constructed according to the feedback results, obtaining the change trend of the filter, and updating the adjustment strategy according to the change trend; S743, output the updated adjustment strategy, and store the filter change trend and adjustment strategy to update parameters.

10. A filter self-adaptive adjustment system based on big data, used to implement the steps of the filter self-adaptive adjustment method based on big data described in any one of claims 1 to 9, characterized in that: The system includes: A parameter acquisition module (1), used to acquire standard air parameters and filter parameters, and calculate the perfect service life parameters of the filter; A damage parameter module (2) is used to preset a damage characteristic value and extract characteristics of a standard air parameter according to the damage characteristic value to obtain a damage parameter; A primary prediction module (3) is used to calculate the primary prediction parameters of the filter life according to the damage parameters and the filter parameters, and compare them with the perfect service life parameters of the filter, and update the comparison results according to the update threshold; A parameter updating module (4), used for obtaining regional air parameters and filter type parameters according to the updated standard air parameters and filter parameters; An advanced prediction module (5), used to calculate the advanced prediction parameters of filter life according to the regional air parameters and the filter type parameters, and compare the advanced prediction parameters of filter life with the primary prediction parameters of filter life; An analysis and verification module (6) is used to preset a normal threshold, analyze the comparison result of the advanced prediction parameter of the filter life and the primary prediction parameter of the filter life according to the normal threshold, and verify the advanced prediction parameter of the filter life according to the analysis result; The adaptive adjustment module (7) presets an adaptive graded adjustment method, matches the verified advanced prediction parameters of the filter life with the adaptive graded adjustment method, and performs adaptive graded adjustment on the filter according to the matching result.

Images