Air conditioning filter monitoring system and method

Abstract

The present invention discloses an air conditioning filter supervision system and method, which specifically relates to the technical field of Internet of Things. The specific steps include detecting air quality data outside and inside a vehicle through a sensor, collecting and preprocessing the air quality data, performing comparative analysis and evaluating the filtering effect of the air conditioning filter, analyzing the effect change of the air conditioning filter according to air volume data, performing timeliness evaluation on the collected data to verify the validity of the data, modeling the air conditioning filter in combination with a convolutional neural network and a long short-term memory network, formulating a supervision strategy according to the actual effect of the air conditioning filter and the actual working condition of the vehicle, and effectively combining the advantageous features of the convolutional neural network and the long short-term memory network to predict the usable life of the air conditioning filter, which can effectively improve the accuracy of the predicted life of the air conditioning filter and avoid unnecessary replacement or maintenance during the life of the air conditioning filter.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and more specifically, to an air conditioning filter monitoring system and method. Background Art

[0002] The function of the air conditioning filter is to filter the air entering the cabin from the outside, improve the cleanliness of the air, prevent impurities, tiny particles, pollen, bacteria, industrial waste gas and dust from entering the air conditioning system, thereby protecting the air conditioning system and providing a good air environment for passengers in the car. The air conditioning filter will be replaced periodically according to the replacement cycle recommended by the manufacturer or service provider. However, when the vehicle often travels in areas with a lot of dust and poor air quality, pollutants such as particles and dust in the air will quickly accumulate on the filter, causing the filter to be blocked or the filtering effect to decrease, thus failing prematurely. The humidity and moisture in the air may also affect the filter. For example, water droplets, fog, snow and other moisture particles will adhere to the filter, affecting its dust storage capacity and filtering effect. If it is in a humid environment for a long time, the filter may also become moldy or corroded, further reducing its efficiency. Therefore, the air conditioning filter should be monitored in real time to verify the filtering efficiency of the air conditioning filter, but the real-time monitoring data fluctuates with the actual working conditions, so a comprehensive evaluation of the monitoring data and the working condition of the air conditioning filter should be conducted to verify the monitoring effectiveness.

[0003] In order to solve the above defects, a technical solution is now proposed. Summary of the invention

[0004] The object of the present invention is to provide an air conditioning filter monitoring system and method to solve the problems in the above-mentioned background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: an air conditioning filter monitoring method, the specific steps of which include detecting air quality data outside and inside the vehicle through a sensor and collecting and preprocessing the air quality data;

[0006] Compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering effect of the air conditioning filter. Analyze the effect change of the air conditioning filter based on the air volume data, and evaluate the timeliness of the collected data to verify the validity of the data.

[0007] The air conditioning filter is modeled by combining convolutional neural networks and long short-term memory networks, and the actual effectiveness of the air conditioning filter is analyzed based on data validity;

[0008] Develop a regulatory strategy based on the actual performance of the cabin air filter and the actual operating conditions of the vehicle.

[0009] Preferably, the method for preprocessing the data is:

[0010] Remove invalid data, delete or fill in data rows with missing values. Methods for filling missing values ​​include using the mean, median, mode, and using cluster analysis methods to identify and process data points that deviate from the range;

[0011] Integrate data into a unified data warehouse or data set, deduplicate and match data, and normalize data dimensions.

[0012] Preferably, the method for normalizing the data dimension is:

[0013] Label and classify high-dimensional data, and standardize high-dimensional data;

[0014] For each category c in the high-dimensional data, calculate its mean vector μ c , calculate the intra-class scatter matrix S based on the sum of all class scatter matrices W , the calculation expression is Where C is the total number of categories in the high-dimensional data, x i is the i-th sample vector belonging to category c, with dimension n, μ c is the mean vector dimension of category c, with dimension n, and T is the transposition operator;

[0015] Calculate the mean vector μ of all high-dimensional data. The calculation expression is Where N is the total number of data samples, x j is the jth sample vector, and the inter-class scatter matrix S is calculated based on the degree of scatter between the means of different categories. B , the calculation expression is Where N c is the number of samples in category c, T is the transposition operator;

[0016] The optimal projection direction is obtained by maximizing the ratio of the between-class scatter matrix to the within-class scatter matrix. The calculation method is: Among them, ω is the projection vector, which is obtained by solving the generalized eigenvalue. The solution method of the generalized eigenvalue is: Where λ is the generalized eigenvalue;

[0017] Project high-dimensional data into low-dimensional space. The projection method is In the formula, x i For high-dimensional data, y i is the low-dimensional data after projection.

[0018] Preferably, the method for standardizing the data is In the formula, x0 is the data before standardization, x1 is the data after standardization, Me is the mean of the data, and Sd is the standard deviation of the data.

[0019] Preferably, the method for comparing and analyzing the air quality data outside and inside the vehicle is: In the formula, E is the filtration efficiency index, C out is the air quality data outside the car, C in For the air quality data in the car, preset the filtering critical threshold E t , when the calculated filtration efficiency index E is less than or equal to the filtration critical threshold E t When the calculated filter efficiency index E is greater than the critical filter threshold E, the air conditioning filter is marked as failed and a prompt is given to replace the air conditioning filter. t , it indicates that the air conditioning filter is effective and that the air conditioning filter is working properly.

[0020] Preferably, the method for analyzing the change in the effectiveness of the air conditioning filter according to the air volume data is:

[0021] Pressure sensors are installed at the air inlet and outlet of the air conditioner to monitor the pressure difference at both ends of the air conditioner filter in real time as an indicator of pressure loss. The calculation expression of the pressure loss value is ΔP = P in -P out , where ΔP is the pressure loss value, P in is the inlet pressure of the air conditioning filter, P out is the outlet pressure of the air conditioning filter.

[0022] Preferably, the method for verifying the timeliness of data is:

[0023] Obtain the duration of the transmission signal received by the sensor at different times within the time T, and calibrate the duration of the transmission signal received by the sensor as S e , e represents the serial number of the duration of the transmission signal received by the sensor, and e={12,3…E}, E is the total number of serial numbers of the duration of the transmission signal received by the sensor, and E is a positive integer;

[0024] Calculate the duration S of the transmission signal received by the sensor at different times within the time T e The standard deviation of the signal continuity coefficient S cn , the calculation formula is:

[0025]

[0026] in, The duration S of the transmission signal received by the sensor at different times within the time T e The average value of

[0027] The transmission signals received by the sensor at different times within the time T whose duration is less than the preset minimum threshold value Et of the effective transmission signal are re-marked and marked as S a , a represents the number of the transmission signal received by the sensor whose duration is less than the preset minimum threshold value Et of the effective transmission signal, and a={12,3…A}, A is a positive integer;

[0028] Calculate the interruption transmission coefficient I n , the calculation formula is

[0029] When the calculated I n Less than or equal to the preset interruption threshold I t , then the sensor signal transmission is valid, otherwise the sensor signal transmission is invalid.

[0030] Preferably, the collected data is normalized, and the air volume, pressure loss and sensor time-efficiency data in a continuous time period are divided into sliding windows, and the data in each window is used as an input sample;

[0031] The dynamic characteristics of air volume are captured through the rate of change of air volume and wind speed. The dynamic characteristics of the change of air conditioning filter blockage pressure are captured through the rate of change of pressure loss. The dynamic characteristics of the impact of sensor aging or failure on data quality are captured through sensor time data to construct time series data.

[0032] The time series data is processed by sliding window and input into the neural network model. The time series data is processed by 1D convolution layer, the lag degree of air volume and pressure loss in the time series is processed by long short-term memory network layer, and the predicted life value is output.

[0033] Preferably, a supervision strategy is formulated according to the actual efficacy of the air conditioning filter and the actual working condition of the vehicle. When the predicted life value output by the neural network model is lower than the preset life critical value, the efficacy loss of the air conditioning filter is marked as irreversible, and a prompt is given to replace the air conditioning filter;

[0034] When the predicted life value output by the neural network model is greater than or equal to a preset life critical value, it is marked that the air conditioning filter has limited loss of effectiveness and a prompt is given that the air conditioning filter is available.

[0035] An air conditioning filter supervision system includes a quality data acquisition module, an efficacy analysis module, a model compound module, and a strategy formulation module;

[0036] The quality data acquisition module is used to detect the air quality data outside and inside the vehicle through sensors and collect and pre-process the air quality data;

[0037] The efficacy analysis module is used to compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering efficacy of the air conditioning filter. It analyzes the efficacy changes of the air conditioning filter based on the air volume data and conducts a timeliness evaluation on the collected data to verify the validity of the data.

[0038] The model composite module is used to model the air conditioning filter by combining convolutional neural network and long short-term memory network, and analyze the actual efficacy of the air conditioning filter in combination with data validity;

[0039] The strategy formulation module is used to formulate a supervision strategy based on the actual efficacy of the air conditioning filter and the actual operating conditions of the vehicle.

[0040] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0041] The present application predicts the useful life of the air conditioning filter by combining the advantages of convolutional neural networks and long short-term memory networks, verifies the validity of real-time data by analyzing the timeliness of sensor data, extracts advantageous local features through a 1D convolution layer, and then uses a long short-term memory network layer to amplify and capture the local features. This can effectively improve the accuracy of the predicted life of the air conditioning filter, avoid unnecessary replacement or maintenance during the life of the air conditioning filter, and improve the actual use efficiency of the air conditioning filter. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application 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 recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0043] Figure 1 The figure is a flow chart of the method of the present invention.

[0044] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Example 1: Please refer to Figure 1As shown, the present invention is an air conditioning filter supervision method, and the specific steps include detecting the air quality data outside and inside the vehicle through a sensor and collecting and preprocessing the air quality data;

[0047] Compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering effect of the air conditioning filter. Analyze the effect change of the air conditioning filter based on the air volume data, and evaluate the timeliness of the collected data to verify the validity of the data.

[0048] The air conditioning filter is modeled by combining convolutional neural networks and long short-term memory networks, and the actual effectiveness of the air conditioning filter is analyzed based on data validity;

[0049] Develop a regulatory strategy based on the actual performance of the cabin air filter and the actual operating conditions of the vehicle.

[0050] The air quality data outside and inside the vehicle are collected through sensors, including the particle concentration data outside the vehicle, the particle concentration and temperature and humidity data inside the vehicle, and the air volume data of the air conditioner including the supply air volume, return air volume, wind speed and wind pressure. The sensors used to collect the above data include particle concentration sensors, temperature and humidity sensors, air volume sensors, wind speed sensors and wind pressure sensors;

[0051] In a specific embodiment, the timeliness of the collected data is evaluated by preprocessing the collected data. The method of preprocessing the data is:

[0052] Remove invalid data, delete or fill in data rows with missing values. Methods for filling missing values ​​include using the mean, median, mode, and using cluster analysis methods to identify and process data points that deviate from the range;

[0053] Integrate data into a unified data warehouse or data set, deduplicate and match data, and normalize data dimensions.

[0054] The dimensions of collected data are not uniform. For example, the collected air quality data include not only pollutant concentrations, but also volatile organic compounds, gas composition, and particulate matter composition analysis information. The air quality data constitutes high-dimensional data, covering multiple different pollutants and chemical components, and each component is an independent dimension. When analyzing and processing high-dimensional data, more variables and dimensions need to be processed, resulting in a significant increase in the amount of calculation. In high-dimensional data, there is both redundant information and noise, which are harmful to the accuracy and effectiveness of data analysis. In data mining, high-dimensional data often leads to the dimensionality disaster, that is, as the dimension increases, the sparsity of the data increases, making the model training difficult and prone to overfitting.

[0055] In order to alleviate the complexity of high-dimensional data processing, high-dimensional data must be processed by dimensionality reduction. Dimensionality reduction maps data from high-dimensional space to low-dimensional space, thereby reducing the complexity and time cost of calculation. Dimensionality reduction can also remove redundancy and noise, retain the main features and useful information of the data, and reduce the dimension of the data. Dimensionality reduction can improve the training efficiency and generalization ability of the model, and extract the key features in the data through dimensionality reduction, thereby revealing the association and pattern between the data. Therefore, the data dimension must be normalized;

[0056] The method to normalize the data dimensions is:

[0057] Label and classify high-dimensional data, and standardize high-dimensional data;

[0058] For each category c in the high-dimensional data, calculate its mean vector μ c , calculate the intra-class scatter matrix S based on the sum of all class scatter matrices W , the calculation expression is Where C is the total number of categories in the high-dimensional data, x i is the i-th sample vector belonging to category c, with dimension n, μ c is the mean vector dimension of category c, with dimension n, and T is the transposition operator;

[0059] Calculate the mean vector μ of all high-dimensional data. The calculation expression is Where N is the total number of data samples, x j is the jth sample vector, and the inter-class scatter matrix S is calculated based on the degree of scatter between the means of different categories. B , the calculation expression is Where N c is the number of samples in category c, T is the transposition operator;

[0060] The optimal projection direction is obtained by maximizing the ratio of the between-class scatter matrix to the within-class scatter matrix. The calculation method is: Among them, ω is the projection vector, which is obtained by solving the generalized eigenvalue. The solution method of the generalized eigenvalue is: Where λ is the generalized eigenvalue;

[0061] Project high-dimensional data into low-dimensional space. The projection method is In the formula, x i For high-dimensional data, y i is the low-dimensional data after projection.

[0062] Different data features have different dimensions and dimensional units. When data of different dimensions are directly involved in calculations, some features may be overemphasized or ignored due to the difference in dimensions, thus affecting the results of data analysis. Distance-based algorithms or gradient-based optimization algorithms are sensitive to the scale of data. If the data is not standardized, the performance of the algorithm may be seriously affected because the large values ​​of some features may dominate the entire calculation process, resulting in the algorithm being unable to correctly capture the actual patterns and relationships in the data.

[0063] The method to standardize the data is In the formula, x0 is the data before standardization, x1 is the data after standardization, Me is the mean of the data, and Sd is the standard deviation of the data.

[0064] It should be noted that, in a specific embodiment, high-dimensional data are marked and classified, and a group of air quality data is classified according to the calculation method of the air quality index, and is divided into excellent, good, lightly polluted, moderately polluted, etc. according to the value of the air quality index. The logic of marking and classification is set by professional and technical personnel in this field according to the data type.

[0065] The air quality data outside and inside the vehicle are compared and analyzed to evaluate the filtering effect of the air conditioning filter. The method for comparing and analyzing the air quality data outside and inside the vehicle is as follows: In the formula, E is the filtration efficiency index, C out is the air quality data outside the car, C in For the air quality data in the car, preset the filtering critical threshold E t , when the calculated filtration efficiency index E is less than or equal to the filtration critical threshold E t When the calculated filter efficiency index E is greater than the critical filter threshold E, the air conditioning filter is marked as failed and a prompt is given to replace the air conditioning filter. t When the air conditioning filter is marked as valid, it indicates that the air conditioning filter is working properly;

[0066] The efficiency change of the air conditioning filter is analyzed according to the air volume data. Pressure sensors are installed at the air inlet and outlet of the air conditioner to monitor the pressure difference at both ends of the air conditioning filter in real time. The pressure loss value is calculated as ΔP = P in -P out , where ΔP is the pressure loss value, P in is the inlet pressure of the air conditioning filter, P out is the outlet pressure of the air conditioning filter;

[0067] As the air volume increases, the pressure loss usually increases, resulting in a decrease in filter efficiency. Therefore, by plotting the relationship between pressure loss and air volume, the pressure loss of the filter at different air volumes is observed to evaluate its effectiveness.

[0068] The collected data is evaluated for timeliness to verify the validity of the data. The method for verifying the timeliness of the data is:

[0069] Obtain the duration of the transmission signal received by the sensor at different times within the time T, and calibrate the duration of the transmission signal received by the sensor as S e , e represents the serial number of the duration of the transmission signal received by the sensor, and e={12,3…E}, E is the total number of serial numbers of the duration of the transmission signal received by the sensor, and E is a positive integer;

[0070] Calculate the duration S of the transmission signal received by the sensor at different times within the time T e The standard deviation of the signal continuity coefficient S cn , the calculation formula is:

[0071]

[0072] in, The duration S of the transmission signal received by the sensor at different times within the time T e The average value of

[0073] The transmission signals received by the sensor at different times within the time T whose duration is less than the preset minimum threshold value Et of the effective transmission signal are re-marked and marked as S a , a represents the number of the transmission signal received by the sensor whose duration is less than the preset minimum threshold value Et of the effective transmission signal, and a={12,3…A}, A is a positive integer;

[0074] Calculate the interruption transmission coefficient I n , the calculation formula is

[0075] When the calculated I n Less than or equal to the preset interruption threshold I t , then the sensor signal transmission is valid, otherwise the sensor signal transmission is invalid.

[0076] The air conditioning filter is modeled by combining convolutional neural networks and long short-term memory networks, and the actual effectiveness of the air conditioning filter is analyzed based on data validity;

[0077] The collected data is normalized, and the air volume, pressure loss and sensor time data in a continuous time period are divided into sliding windows, and the data in each window is used as an input sample;

[0078] The dynamic characteristics of air volume are captured through the rate of change of air volume and wind speed. The dynamic characteristics of the change of air conditioning filter blockage pressure are captured through the rate of change of pressure loss. The dynamic characteristics of the impact of sensor aging or failure on data quality are captured through sensor time data to construct time series data.

[0079] The time series data is processed by sliding window and input into the neural network model. The time series data is processed by 1D convolution layer, the lag degree of air volume and pressure loss in the time series is processed by long short-term memory network layer, and the predicted life value is output.

[0080] It should be noted that there are lag terms in real-time data. The lag values ​​of air volume and pressure loss are added to capture time dependence for constructing time series features. The mean square error is used as the loss function to train the model. Adaptive optimizers such as Adam or RMSprop are used to quickly converge the model. The generalization ability of the model is improved through cross-validation methods. The importance of the model to the input features is analyzed to analyze the impact of air volume, pressure loss and sensor timeliness on the working state of the air conditioning filter.

[0081] The model is deployed in the actual system, and the current air volume, pressure loss and sensor data are input in real time to predict the filter efficiency.

[0082] Convolutional neural networks are suitable for processing sequence data with local correlation, such as spatial correlation in images or temporal correlation in time series. For sensor data, 1D convolutional neural networks can provide modeling advantages in the following aspects:

[0083] 1D convolution can capture short-term local patterns in sensor data through sliding filters. For example, the changes in wind volume or pressure recorded by the sensor over certain time periods can be used to identify trends or anomalies through convolution operations;

[0084] 1D convolutional layers can reduce data noise through filtering operations and extract more representative features, thereby improving the robustness of the model. For example, convolutional layers can detect periodic fluctuations or sudden changes in data;

[0085] The convolution operation can reduce the dimension of the data while retaining important features, reducing model complexity and improving computational efficiency;

[0086] Through multiple layers of convolution, convolutional neural networks can gradually abstract and integrate information in the time dimension, and ultimately extract high-level temporal features to help better predict and understand the dynamic changes of sensor data.

[0087] The LSTM network is good at processing and modeling time series data. Unlike ordinary recurrent neural networks, the LSTM network can effectively remember dependencies over a longer time range by introducing memory units and gating mechanisms. In the state prediction of the air conditioning filter, the historical changes in air volume and pressure loss may have a long-term dependency on the current state, such as the cumulative effect of filter blockage. The LSTM network can capture and utilize this long-term dependency to improve the accuracy of the prediction;

[0088] In filter effectiveness modeling, historical data with a long time span must be input, which may span days, weeks or even months. Long short-term memory networks can effectively memorize and process such long-term data, helping the model to consider the impact of historical states when making decisions;

[0089] The time-varying features of sensor data, air volume, and pressure loss usually present complex nonlinear patterns, such as sudden changes, gradual changes, etc. The LSTM network can capture and learn these complex nonlinear dynamic patterns through its internal structure, thereby more accurately modeling the state changes of the filter;

[0090] LSTM can detect abnormal patterns in data, such as sudden changes in air volume or surges in pressure loss, by analyzing historical data, and predict future conditions. For example, LSTM can identify mutations or drifts in sensor signals, thereby issuing early warnings that filter or sensor maintenance is needed;

[0091] Convolutional neural network layers are often used to extract local temporal features, such as wind volume changes or pressure loss patterns in the short term. The LSTM layer can receive the local features extracted by the convolutional neural network layer and further process and understand these features in a longer time context. This combination allows the model to capture both short-term and long-term temporal dependencies;

[0092] Stacking multiple LSTM layers allows the model to capture more complex temporal dependencies. Multi-layer LSTM networks help extract features from different levels of time scales and improve the model's expressiveness.

[0093] The gating mechanism includes input gate, forget gate, and output gate.

[0094] Formulate a supervision strategy based on the actual performance of the air conditioning filter and the actual working conditions of the vehicle. When the predicted life value output by the neural network model is lower than the preset life critical value, mark the air conditioning filter's performance loss as irreversible and prompt to replace the air conditioning filter;

[0095] When the predicted life value output by the neural network model is greater than or equal to a preset life critical value, it is marked that the air conditioning filter has limited loss of effectiveness and a prompt is given that the air conditioning filter is available.

[0096] Example 2: Please refer to Figure 2 As shown, an air conditioning filter supervision system includes a quality data acquisition module, an efficacy analysis module, a model compound module, and a strategy formulation module;

[0097] The quality data acquisition module is used to detect the air quality data outside and inside the vehicle through sensors and collect and pre-process the air quality data;

[0098] The efficacy analysis module is used to compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering efficacy of the air conditioning filter. It analyzes the efficacy changes of the air conditioning filter based on the air volume data and conducts a timeliness evaluation on the collected data to verify the validity of the data.

[0099] The model composite module is used to model the air conditioning filter by combining convolutional neural network and long short-term memory network, and analyze the actual efficacy of the air conditioning filter in combination with data validity;

[0100] The strategy formulation module is used to formulate a supervision strategy based on the actual efficacy of the air conditioning filter and the actual operating conditions of the vehicle.

[0101] The present application predicts the useful life of the air conditioning filter by combining the advantages of convolutional neural networks and long short-term memory networks, verifies the validity of real-time data by analyzing the timeliness of sensor data, extracts advantageous local features through a 1D convolution layer, and then uses a long short-term memory network layer to amplify and capture the local features. This can effectively improve the accuracy of the predicted life of the air conditioning filter, avoid unnecessary replacement or maintenance during the life of the air conditioning filter, and improve the actual use efficiency of the air conditioning filter.

[0102] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0103] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0104] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the method described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

Claims

1. A method for monitoring an air conditioning filter, characterized in that: The specific steps include detecting the air quality data outside and inside the vehicle through sensors and collecting and preprocessing the air quality data; Compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering effect of the air conditioning filter. Analyze the effect change of the air conditioning filter based on the air volume data, and evaluate the timeliness of the collected data to verify the validity of the data. The air conditioning filter is modeled by combining convolutional neural networks and long short-term memory networks, and the actual effectiveness of the air conditioning filter is analyzed based on data validity; Develop a regulatory strategy based on the actual performance of the cabin air filter and the actual operating conditions of the vehicle; The method for verifying the timeliness of data is: Obtain the duration of the transmission signal received by the sensor at different times within the time T, and calibrate the duration of the transmission signal received by the sensor as S e , e represents the serial number of the duration of the transmission signal received by the sensor, and e={12,3…E}, E is the total number of serial numbers of the duration of the transmission signal received by the sensor, and E is a positive integer; Calculate the duration S of the transmission signal received by the sensor at different times within the time T e The standard deviation of the signal continuity coefficient S cn , the calculation formula is: in, The duration S of the transmission signal received by the sensor at different times within the time T e The average value of The transmission signals received by the sensor at different times within the time T whose duration is less than the preset minimum threshold value Et of the effective transmission signal are re-marked and marked as S a , a represents the number of the transmission signal received by the sensor whose duration is less than the preset minimum threshold value Et of the effective transmission signal, and a={12,3…A}, A is a positive integer; Calculate the interruption transmission coefficient I n , the calculation formula is When the calculated I n Less than or equal to the preset interruption threshold I t When , the signal transmission of the marking sensor is valid, otherwise the signal transmission of the marking sensor is invalid; The collected data is normalized, and the air volume, pressure loss and sensor time data in a continuous time period are divided into sliding windows, and the data in each window is used as an input sample; The dynamic characteristics of air volume are captured through the rate of change of air volume and wind speed. The dynamic characteristics of the change of air conditioning filter blockage pressure are captured through the rate of change of pressure loss. The dynamic characteristics of the impact of sensor aging or failure on data quality are captured through sensor time data to construct time series data. The time series data is processed by sliding window and input into the neural network model. The time series data is processed by 1D convolution layer, the lag degree of air volume and pressure loss in the time series is processed by long short-term memory network layer, and the predicted life value is output.

2. The air conditioning filter monitoring method according to claim 1, characterized in that: The method for preprocessing data is: Remove invalid data, delete or fill in data rows with missing values. Methods for filling missing values ​​include using the mean, median, mode, and using cluster analysis methods to identify and process data points that deviate from the range; Integrate data into a unified data warehouse or data set, deduplicate and match data, and normalize data dimensions.

3. The air conditioning filter monitoring method according to claim 2, characterized in that: The method to normalize the data dimensions is: Label and classify high-dimensional data, and standardize high-dimensional data; For each category c in the high-dimensional data, calculate its mean vector μ c , calculate the intra-class scatter matrix S based on the sum of all class scatter matrices W , the calculation expression is Where C is the total number of categories in the high-dimensional data, x i is the i-th sample vector belonging to category c, with dimension n, μ c is the mean vector dimension of category c, with dimension n, and T is the transposition operator; Calculate the mean vector μ of all high-dimensional data. The calculation expression is Where N is the total number of data samples, x j is the jth sample vector, and the inter-class scatter matrix S is calculated based on the degree of scatter between the means of different categories. B , the calculation expression is Where N c is the number of samples in category c, T is the transposition operator; The optimal projection direction is obtained by maximizing the ratio of the between-class scatter matrix to the within-class scatter matrix. The calculation method is: Among them, ω is the projection vector, which is obtained by solving the generalized eigenvalue. The solution method of the generalized eigenvalue is: Where λ is the generalized eigenvalue; Project high-dimensional data to low-dimensional space, the projection method is y i =ω T x i , where x i For high-dimensional data, y i is the low-dimensional data after projection.

4. The air conditioning filter monitoring method according to claim 3, characterized in that: The method to standardize the data is In the formula, x0 is the data before standardization, x1 is the data after standardization, Me is the mean of the data, and Sd is the standard deviation of the data.

5. The air conditioning filter monitoring method according to claim 1, characterized in that: The method for comparing and analyzing the air quality data outside and inside the car is as follows: In the formula, E is the filtration efficiency index, C out is the air quality data outside the car, C in For the air quality data in the car, preset the filtering critical threshold E t , when the calculated filtration efficiency index E is less than or equal to the filtration critical threshold E t When the calculated filter efficiency index E is greater than the critical filter threshold E, the air conditioning filter is marked as failed and a prompt is given to replace the air conditioning filter. t , it indicates that the air conditioning filter is effective and that the air conditioning filter is working properly.

6. The air conditioning filter monitoring method according to claim 1, characterized in that: The method for analyzing the change in the effectiveness of the air conditioning filter based on the air volume data is: Pressure sensors are installed at the air inlet and outlet of the air conditioner to monitor the pressure difference at both ends of the air conditioner filter in real time as an indicator of pressure loss. The calculation expression of the pressure loss value is ΔP = P in -P out , where ΔP is the pressure loss value, P in is the inlet pressure of the air conditioning filter, P out is the outlet pressure of the air conditioning filter.

7. The air conditioning filter monitoring method according to claim 1, characterized in that: Formulate a supervision strategy based on the actual performance of the air conditioning filter and the actual working conditions of the vehicle. When the predicted life value output by the neural network model is lower than the preset life critical value, mark the air conditioning filter's performance loss as irreversible and prompt to replace the air conditioning filter; When the predicted life value output by the neural network model is greater than or equal to a preset life critical value, it is marked that the air conditioning filter has limited loss of effectiveness and a prompt is given that the air conditioning filter is available.

8. An air conditioning filter monitoring system, characterized in that: It includes quality data collection module, efficacy analysis module, model compound module and strategy formulation module; The quality data acquisition module is used to detect the air quality data outside and inside the vehicle through sensors and collect and pre-process the air quality data; The efficacy analysis module is used to compare and analyze the air quality data outside and inside the vehicle and evaluate the filtering efficacy of the air conditioning filter. It analyzes the efficacy changes of the air conditioning filter based on the air volume data and conducts a timeliness evaluation on the collected data to verify the validity of the data. The model composite module is used to model the air conditioning filter by combining convolutional neural network and long short-term memory network, and analyze the actual efficacy of the air conditioning filter in combination with data validity; The strategy formulation module is used to formulate a supervision strategy based on the actual efficacy of the air conditioning filter and the actual operating conditions of the vehicle.

Images