Atmospheric pollution monitoring method, device, equipment, storage medium and product

Abstract

The application provides an atmospheric pollution monitoring method, device, equipment, storage medium and product. The method comprises: collecting atmospheric data based on a working parameter by a collection device to obtain atmospheric pollutant concentration data in a preset time range; adaptively adjusting the working parameter according to an adjustment coefficient based on statistical characteristics of the atmospheric pollutant concentration data in the preset time range, wherein the adaptively adjusted working parameter corresponds to the statistical characteristics; calculating a first performance index value corresponding to an atmospheric data collection process under the condition that the working parameter of the collection device is adjusted; and optimizing the adjustment coefficient by a gradient descent method according to the first performance index value, so that a second performance index value of a corresponding data collection process under the condition that the working parameter is adjusted based on the optimized adjustment coefficient is greater than the first performance index value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of environmental monitoring, and in particular to an atmospheric pollution monitoring method, device, equipment, storage medium and product. BACKGROUND

[0002] With the rapid development of industries and transportation, especially the large use of coal and oil, atmospheric pollution has become a major concern. Therefore, atmospheric pollution monitoring technology has also begun to develop gradually. Atmospheric pollution monitoring refers to the process of determining the types and concentrations of pollutants in the atmosphere, observing their spatial and temporal distribution and variation. The purpose of atmospheric pollution monitoring is to identify atmospheric pollutants, understand their distribution and diffusion rules, and monitor the emission and control of atmospheric pollution sources.

[0003] In the prior art, atmospheric pollution monitoring mainly relies on fixed site monitoring equipment, but atmospheric pollutants may contain many different components and vary with time and environmental conditions. The traditional monitoring method may not effectively capture these complex data characteristics when using fixed monitoring equipment and monitoring according to preset working parameters. The real-time performance of atmospheric pollution monitoring is not high, and the atmospheric pollution situation cannot be quickly determined. SUMMARY

[0004] The embodiments of the present application provide an atmospheric pollution monitoring method, device, equipment, storage medium and product, which can quickly respond to environmental changes and determine the atmospheric pollution situation.

[0005] In a first aspect, the present application provides an atmospheric pollution monitoring method, comprising:

[0006] acquiring atmospheric data based on working parameters by a collection device to obtain atmospheric pollutant concentration data in a preset time range;

[0007] According to the statistical characteristics of the atmospheric pollutant concentration data in the preset time range, the working parameters are adaptively adjusted according to an adjustment coefficient, and the adaptively adjusted working parameters correspond to the statistical characteristics;

[0008] In the case of adjusting the working parameters of the collection device, a first performance index value corresponding to the atmospheric data acquisition process is calculated;

[0009] According to the first performance index value, the adjustment coefficient is optimized by gradient descent method, so that in the case of adjusting the working parameters based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value.

[0010] In some possible implementation manners, the collecting, by the collecting device, of the atmospheric data based on the working parameters to obtain the atmospheric pollutant concentration data in the preset time range comprises:

[0011] The collecting device collects pollutant concentration values corresponding to each time point in the preset time range according to the working parameters;

[0012] The average value of the pollutant concentration values corresponding to each time point is calculated to obtain a pollutant concentration mean value;

[0013] The square value of the difference between the pollutant concentration value corresponding to each time point and the pollutant concentration mean value is calculated;

[0014] The average value of each square is calculated, and the square root of the average value of the square is taken to obtain a pollutant concentration standard deviation;

[0015] According to the rate of change of the pollutant concentration values over time, a pollutant concentration change trend is determined;

[0016] The pollutant concentration values corresponding to each time point, the pollutant concentration mean value, the pollutant concentration standard deviation, and the pollutant concentration change trend are taken as the atmospheric pollutant concentration data in the preset time range.

[0017] In some possible implementation manners, the working parameters comprise a sampling frequency, a quantization precision, and a data compression rate, and the adaptive adjustment of the working parameters according to the statistical characteristics of the atmospheric pollutant concentration data in the preset time range comprises:

[0018] A sampling frequency improvement value is determined according to the adjustment coefficient and the pollutant concentration standard deviation and the pollutant concentration change trend;

[0019] A target sampling frequency is determined according to a preset basic sampling frequency and the sampling frequency improvement value, and the sampling frequency is adjusted to the target sampling frequency;

[0020] The quantization precision is adjusted according to a current pollutant concentration value and a pollutant concentration extreme value in the preset time range;

[0021] The data compression rate is adjusted according to the ratio of the pollutant concentration to a preset standard deviation.

[0022] In some possible implementation manners, after the collecting, by the collecting device, of the atmospheric data based on the working parameters to obtain the atmospheric pollutant concentration data in the preset time range, the method further comprises:

[0023] The atmospheric pollutant concentration data is decomposed by a preset wavelet basis function to obtain wavelet coefficients, the wavelet coefficients representing local characteristics of the atmospheric pollutant concentration data at different scales;

[0024] The wavelet coefficients are extracted to obtain pollutant concentration features at different scales;

[0025] The pollutant concentration features at different scales are combined to obtain multi-scale fusion features;

[0026] The multi-scale fusion features are input into a preset prediction model to obtain atmospheric pollutant concentration prediction data corresponding to a preset time, the preset time being later than the preset time range.

[0027] In some possible implementations, the atmospheric pollutant concentration data is decomposed by a preset wavelet basis function to obtain wavelet coefficients, including:

[0028] A decomposition scale corresponding to the preset wavelet basis function is obtained;

[0029] The atmospheric pollutant concentration data is decomposed by the preset wavelet basis function according to the decomposition scale to obtain initial wavelet coefficients;

[0030] The initial wavelet coefficients are statistically analyzed to determine statistical features of the initial wavelet coefficients;

[0031] The initial wavelet coefficients are denoised based on a threshold corresponding to the statistical features to obtain wavelet coefficients.

[0032] In some possible implementations, after the adjustment coefficient is optimized by the gradient descent method according to the first performance index value, the method further includes:

[0033] The working parameter is adjusted based on the optimized adjustment coefficient;

[0034] The pollutant concentration data is collected according to the working parameter;

[0035] The pollutant concentration data, geographical data and meteorological data are combined to obtain environmental fusion data;

[0036] The environmental fusion data is processed by a preset neural network model to obtain an atmospheric pollution analysis result.

[0037] In some possible implementations, the preset neural network model includes a convolutional neural network, a graph neural network, a long short-term memory network and a fully connected network, and the environmental fusion data is processed by the preset neural network model to obtain an atmospheric pollution analysis result, including:

[0038] The spatiotemporal features of the environmental fusion data are obtained by extracting features from the environmental fusion data using the convolutional neural network.

[0039] The pollution diffusion characteristics of the environmental fusion data are obtained by extracting features from the environmental fusion data using the graph neural network.

[0040] The spatiotemporal features and the pollution diffusion features are input into the long short-term memory network to obtain the time-dependent features of air pollution.

[0041] The time-dependent features are input into the fully connected network to obtain the air pollution analysis results.

[0042] In some possible implementations, calculating the first performance index value corresponding to the atmospheric data acquisition process after adjusting the operating parameters of the acquisition device includes:

[0043] With the operating parameters adjusted, the data quality index values, energy consumption, and data transmission volume during the data acquisition process are obtained.

[0044] The data quality index value, the energy consumption, and the data transmission volume are weighted and summed to obtain the first performance index value corresponding to the data acquisition process.

[0045] Secondly, this application provides an air pollution monitoring device, the device comprising:

[0046] The data acquisition module is used to collect atmospheric data based on operating parameters through the acquisition device, and obtain atmospheric pollutant concentration data within a preset time range;

[0047] The adjustment module is used to adaptively adjust the working parameters according to the statistical characteristics of the atmospheric pollutant concentration data within the set time range, and the adaptively adjusted working parameters correspond to the statistical characteristics.

[0048] The calculation module is used to calculate the first performance index value corresponding to the atmospheric data acquisition process after adjusting the operating parameters of the acquisition device.

[0049] An optimization module is used to optimize the adjustment coefficient according to the first performance index value using the gradient descent method, so that when the working parameters are adjusted based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value.

[0050] Thirdly, this application provides an air pollution monitoring device, the device comprising: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the air pollution monitoring method described above.

[0051] Fourthly, this application provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the air pollution monitoring method described above.

[0052] Fifthly, this application provides a computer program product in which the instructions, when executed by a processor of an electronic device, cause the electronic device to perform the air pollution monitoring method described above.

[0053] The air pollution monitoring method, apparatus, equipment, storage medium, and product provided in this application continuously adjust the operating parameters of the acquisition equipment according to real-time air pollutant concentration data and adjustment coefficients, enabling more rational allocation of system resources and improving acquisition efficiency during the acquisition process. Furthermore, a corresponding first performance index value is calculated, and the adjustment coefficient is optimized according to the first performance index value, continuously optimizing the adjustment coefficient through iterative processes. Since the operating parameters are adjusted according to actual atmospheric conditions to adapt to the current situation, the adjusted acquisition operating parameters allow for more efficient acquisition of changes in air pollutant concentrations based on the actual environment, enabling rapid response to environmental changes and quick assessment of air pollution conditions. Attached Figure Description

[0054] This application can be better understood from the following description of specific embodiments in conjunction with the accompanying drawings, wherein:

[0055] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings, wherein the same or similar reference numerals denote the same or similar features.

[0056] Figure 1 This is a flowchart of an air pollution monitoring method provided in one embodiment of this application;

[0057] Figure 2 This is a flowchart of an environmental data extraction process provided in one embodiment of this application;

[0058] Figure 3 This is a flowchart of a multi-source data fusion and mining process provided in one embodiment of this application;

[0059] Figure 4This is a schematic diagram of the structure of an air pollution monitoring device provided in one embodiment of this application;

[0060] Figure 5 This is a schematic diagram of the hardware structure of the air pollution monitoring equipment provided in the embodiments of this application. Detailed Implementation

[0061] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0062] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0063] Integrated Sensing and Communication (ISAC) is a technology that shares the same frequency band and hardware for sensing and communication. It can fully share the time, space and other multi-dimensional resources of wireless communication and radar sensing, and realize the coexistence and mutual assistance of the two technologies.

[0064] In this embodiment, environmental monitoring is performed by an ISAC device. Specifically, the ISAC device uses high-sensitivity, low-power, and miniaturized sensors such as lidar and electrochemical sensors to monitor the concentrations of air pollutants such as PM2.5, CO, SO2, and NO2 in real time.

[0065] Design a joint processing circuit for radio frequency (RF) signals and environmental sensing signals to efficiently separate and extract the two types of signals. The RF communication signal is processed by filtering, amplification, and down-conversion before being sent to the baseband processing unit for communication demodulation. The environmental sensing signal is processed by a dedicated signal conditioning circuit for filtering, amplification, and AD conversion before being sent to the environmental signal processing unit for feature extraction and concentration calculation.

[0066] Based on communication traffic volume and environmental monitoring needs, the allocation ratio of radio frequency resources and computing resources between communication and sensing tasks is dynamically configured. During peak communication periods, priority is given to ensuring communication quality; during off-peak periods, more resources are used for the acquisition and processing of environmental monitoring data to improve the temporal resolution of monitoring.

[0067] To address the problems of the prior art, embodiments of this application provide an air pollution monitoring method, apparatus, device, storage medium, and product. The air pollution monitoring method provided in this application embodiment is described first, and this method can be executed by the aforementioned air pollution monitoring apparatus.

[0068] Figure 1 A schematic flowchart of an air pollution monitoring method according to an embodiment of this application is shown. Figure 1 As shown, the method includes steps S101 to S105.

[0069] Step S101: Collect atmospheric data based on the working parameters using the data acquisition device to obtain atmospheric pollutant concentration data within a preset time range.

[0070] In practice, the data acquisition device is pre-set with operating parameters, which may include sampling frequency, sensor sensitivity, etc. Data is collected according to a preset time range. The acquisition device continuously measures and records the concentration of air pollutants to obtain air pollutant concentration data within the preset time range.

[0071] Step S102: Based on the statistical characteristics of the above-mentioned air pollutant concentration data within the above-mentioned time range, the above-mentioned working parameters are adaptively adjusted according to the adjustment coefficient, and the above-mentioned working parameters after adaptive adjustment correspond to the above-mentioned statistical characteristics.

[0072] In the specific implementation, statistical analysis is performed on the data collected in S101 to calculate the statistical characteristics of the data, such as mean and variance. Adjustment coefficients are then set based on these statistical characteristics to adjust the operating parameters. The operating parameters are then adaptively adjusted according to the calculated adjustment coefficients.

[0073] As another example, if the data variance is large, the sensor sensitivity or sampling frequency can be adjusted to improve data quality.

[0074] Step S103: After adjusting the above-mentioned operating parameters of the acquisition equipment, calculate the first performance index value corresponding to the atmospheric data acquisition process.

[0075] In practice, in order to evaluate the adjusted acquisition effect and optimize the adjustment coefficient, data is acquired again under the adjusted working parameters, and the first performance index value is calculated based on the newly acquired data.

[0076] As another example, when calculating the first performance metric value, metrics such as the accuracy and stability of the data can be calculated.

[0077] Step S104: Based on the first performance index value, optimize the adjustment coefficient using the gradient descent method so that, when the working parameters are adjusted based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value.

[0078] In the specific implementation, an objective function is set, with the first performance index value as part of the objective function, and an optimization objective is set, such as the second performance index value should be greater than the first performance index value. Based on the objective function, the gradient of the adjustment coefficient is calculated, and the coefficient is adjusted according to the gradient. The gradient descent step is repeated until the objective function value meets the optimization requirements, thus completing the optimization of the adjustment coefficient.

[0079] The air pollution monitoring method provided in this application continuously adjusts the operating parameters of the acquisition equipment according to real-time air pollutant concentration data and adjustment coefficients, enabling more rational allocation of system resources and improving acquisition efficiency during the acquisition process. Furthermore, a corresponding first performance index value is calculated, and the adjustment coefficient is optimized according to this first performance index value, continuously improving the adjustment coefficient through iterative processes. Since the operating parameters are adjusted according to actual atmospheric conditions to adapt to the current situation, the adjusted acquisition operating parameters allow for more efficient acquisition of changes in air pollutant concentrations based on the actual environment, enabling rapid response to environmental changes and quick assessment of air pollution conditions.

[0080] In order to ensure that the atmospheric pollutant concentration data can comprehensively reflect the actual situation of atmospheric pollution, in some embodiments, the above-mentioned S101 includes steps A1 to F1:

[0081] Step A1: Collect pollutant concentration values ​​at each time point within a preset time range using the data acquisition device according to the operating parameters.

[0082] In practice, the data acquisition device is first configured, and its operating parameters are set, such as sampling frequency, sampling time interval, and sensor sensitivity. The device is then started to collect data, ensuring continuous sampling at the set frequency within the preset time range, and recording the pollutant concentration value at each time point.

[0083] Step B1: Calculate the average value of pollutant concentration at each of the above time points to obtain the average pollutant concentration.

[0084] In practice, the concentration values ​​of pollutants at each of the above time points are added together to obtain the total concentration. The total concentration is then divided by the number of time points to obtain the average pollutant concentration.

[0085] Step C1: Calculate the square of the difference between the above pollutant concentration values ​​and the average pollutant concentration values ​​at each time point.

[0086] In practice, for each pollutant concentration value at each time point, the difference between it and the mean is calculated, and each deviation value is squared to obtain the square of the difference between the pollutant concentration value at each time point and the mean pollutant concentration.

[0087] Step D1: Calculate the average of each of the above squares, and take the square root of the average of the above squares to obtain the standard deviation of the pollutant concentration.

[0088] In practice, the squared deviations at all time points are summed and their average value is calculated. The square root of the average value is then taken to obtain the standard deviation of the pollutant concentration.

[0089] Step E1: Determine the trend of pollutant concentration change based on the rate of change of each pollutant concentration over time.

[0090] In practice, linear regression or other trend analysis methods are used to fit the time series data. The slope of the regression line is calculated, and the trend of pollutant concentration changes is determined based on the magnitude of the slope.

[0091] Step F1: Use the pollutant concentration values ​​at each of the above time points, the average pollutant concentration, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change as the atmospheric pollutant concentration data within the preset time range.

[0092] The embodiments described above in this application collect pollutant concentration values ​​at various time points within a preset time range according to the working parameters of the data acquisition device. Then, the mean concentration and standard deviation of each pollutant concentration value are calculated. The pollutant concentration values ​​at each time point, the mean concentration, the standard deviation, and the trend of pollutant concentration change are then used as atmospheric pollutant concentration data within the preset time range. By comprehensively utilizing multiple data sources as atmospheric pollutant concentration data within the preset time range, the atmospheric pollutant concentration data can fully reflect the actual situation of air pollution.

[0093] In order to accurately adjust the various operating parameters, in some embodiments, the above-mentioned operating parameters include sampling frequency, quantization accuracy, and data compression ratio. The above-mentioned step S102 includes steps A2 to D2:

[0094] Step A2: Determine the sampling frequency increase value according to the adjustment coefficient, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change.

[0095] In practice, the adjustment coefficient and the standard deviation of pollutant concentration calculated from step D1 are first obtained. These reflect the dispersion of the data and the trend of pollutant concentration changes obtained from step E1. Based on the above information, the sampling frequency increase value is determined. For example, if the pollutant concentration increases significantly, the sampling frequency is appropriately increased; if it decreases or stabilizes, the sampling frequency can be maintained or moderately reduced.

[0096] Step B2: Determine the target sampling frequency based on the preset base sampling frequency and the above-mentioned sampling frequency boost value, and adjust the sampling frequency to the above-mentioned target sampling frequency.

[0097] In the specific implementation, the current base sampling frequency is obtained, and the base sampling frequency is added to the sampling frequency boost value to obtain the target sampling frequency. The calculated target sampling frequency is then applied to adjust the sampling frequency to the target sampling frequency.

[0098] As another example, after calculating the target sampling frequency, you can check whether the new target sampling frequency is within the range allowed by the device. If it is outside the range, take appropriate measures, such as returning to the default maximum or minimum frequency.

[0099] Step C2: Adjust the quantification accuracy based on the current pollutant concentration value and the extreme value of pollutant concentration within the preset time range.

[0100] In practice, the extreme values ​​of pollutant concentrations within a preset time range are calculated, including both maximum and minimum values. A new quantification precision is determined based on the relationship between the current concentration value and the extreme values; for example, the closer the current concentration value is to the extreme value, the higher the quantification precision. The quantification precision is then adjusted based on this new precision.

[0101] Step D2: Adjust the data compression rate based on the ratio of the pollutant concentration to the preset standard deviation.

[0102] In practice, the ratio of the current pollutant concentration value to the standard deviation is calculated. If the ratio is greater than a preset threshold, the compression ratio can be increased to reduce storage requirements. If the ratio is close to or lower than 1, the compression ratio is reduced to retain more data details, thereby adjusting the data compression ratio.

[0103] The above-described implementation method of this application adjusts the sampling frequency according to the adjustment coefficient, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change. It adjusts the quantization accuracy based on the current pollutant concentration value and the extreme value of pollutant concentration within a preset time range. It adjusts the data compression rate according to the ratio of the pollutant concentration to the preset standard deviation, thereby accurately adjusting each working parameter.

[0104] In order to obtain accurate predicted data on atmospheric pollutant concentrations, in some embodiments, the above-mentioned S101 includes steps A3 to D3:

[0105] Step A3: Perform wavelet decomposition on the above air pollutant concentration data using a preset wavelet basis function to obtain wavelet coefficients. The wavelet coefficients represent the local characteristics of the above air pollutant concentration data at multiple different scales.

[0106] In practical implementation, time-series data of air pollutant concentrations are acquired. This data typically exists in time-series format, such as hourly or daily pollutant concentration values. Based on the application scenario and data characteristics, a preset wavelet basis function is determined. The selected wavelet basis function is then used to perform a wavelet transform on the pollutant concentration data to obtain wavelet coefficients.

[0107] As another example, the wavelet transform process involves decomposing the original time series data into coefficients of different frequency components. Specifically, the wavelet transform decomposes the data into approximate coefficients (low frequencies) and detail coefficients (high frequencies), layer by layer, to obtain information at different scales. Each decomposition layer generates a set of wavelet coefficients that represent the data characteristics at different scales.

[0108] Step B3: Extract features from the above wavelet coefficients to obtain pollutant concentration features at multiple different scales.

[0109] In practice, the appropriate feature type is selected according to the analysis purpose, and the selected feature is calculated for the wavelet coefficients at each scale to obtain pollutant concentration features at multiple different scales.

[0110] As another example, features that might be extracted include: statistical features such as mean, variance, kurtosis, and skewness; spectral characteristics of wavelet coefficients such as frequency domain amplitude; and local features such as local extrema and points of change.

[0111] Step C3: Combine the above pollutant concentration characteristics at multiple different scales to obtain multi-scale fusion characteristics.

[0112] In the specific implementation, features at each scale are combined into a comprehensive feature set according to a preset method to obtain multi-scale fused features.

[0113] Step D3: Input the above multi-scale fusion features into the preset prediction model to obtain the predicted atmospheric pollutant concentration data corresponding to the preset time. The preset time is later than the preset time range.

[0114] In practice, multi-scale fused feature data is input into the prediction model, and a prediction operation is performed. The model uses the given feature data to infer the concentration of air pollutants at future times, thus obtaining the predicted air pollutant concentration data corresponding to the preset time. For example, if the model is a regression model, it generates a continuous concentration prediction value; if it is a classification model, it may output a prediction of the pollution level.

[0115] The above-described implementation method of this application performs wavelet decomposition on the atmospheric pollutant concentration data using a preset wavelet basis function to obtain wavelet coefficients. Then, feature extraction is performed on the wavelet coefficients to obtain multi-scale fusion features. A preset prediction model is used to process these multi-scale fusion features to obtain atmospheric pollutant concentration prediction data corresponding to a preset time. By using wavelet decomposition and multi-scale feature fusion as concentration prediction data, accurate atmospheric pollutant concentration prediction data can be obtained.

[0116] In order to obtain more accurate wavelet coefficients, in some implementations, A3 above includes steps A31 to A32:

[0117] Step A31: Obtain the decomposition scale corresponding to the preset wavelet basis function.

[0118] In practice, the decomposition scale is set according to the properties of the preset wavelet basis functions. For example, for the Daubechies wavelet, an 8-level decomposition might be chosen. Each scale represents a different frequency component of the data.

[0119] Step A32 involves performing wavelet decomposition on the atmospheric pollutant concentration data according to the decomposition scale corresponding to the preset wavelet basis function to obtain the initial wavelet coefficients.

[0120] In the specific implementation, the data is decomposed according to a preset decomposition scale. Wavelet coefficients with different frequency ranges are generated at each level. Wavelet basis functions are applied to the data, and wavelet coefficients at each level are calculated through convolution operations to obtain approximate coefficients and detail coefficients, thus obtaining the initial wavelet coefficients.

[0121] Step A33 involves statistical analysis of the initial wavelet coefficients to determine their statistical characteristics.

[0122] In practical implementation, statistical characteristics such as mean, variance, kurtosis, and skewness are calculated for the wavelet coefficients at each scale to determine the statistical characteristics of the initial wavelet coefficients. These characteristics describe the distribution and variation properties of the data.

[0123] Step A34 denoises the initial wavelet coefficients based on the thresholds corresponding to the above statistical features to obtain the wavelet coefficients.

[0124] In the specific implementation, based on the statistical characteristics in step A33, an appropriate threshold is selected, including soft thresholding and hard thresholding. In soft thresholding denoising, wavelet coefficients with absolute values ​​below the threshold are set to zero, and the threshold is subtracted from the remaining coefficients. In hard thresholding denoising, wavelet coefficients with absolute values ​​below the threshold are set to zero, and the remaining coefficients remain unchanged. After denoising, wavelet coefficients are obtained.

[0125] The above-described implementation method of this application obtains the decomposition scale corresponding to the preset wavelet basis function, performs wavelet decomposition on the atmospheric pollutant concentration data according to this decomposition scale, and then performs denoising processing on the initial wavelet coefficients to remove useless information, thereby obtaining more accurate wavelet coefficients.

[0126] To obtain accurate air pollution analysis results, in some embodiments, after S104 above, the method further includes steps A4 to D4:

[0127] Step A4: Adjust the working parameters based on the optimized adjustment coefficients.

[0128] In practice, the relevant working parameters are updated based on the optimized adjustment coefficients, thereby adjusting the working parameters.

[0129] As another example, the sensitivity settings of the sensor can be adjusted to improve the accuracy of pollutant concentration measurements. Alternatively, the frequency of data acquisition can be adjusted based on optimization coefficients to ensure the timeliness and accuracy of the data.

[0130] Step B4: Collect pollutant concentration data according to the above working parameters.

[0131] In practice, the system is equipped with parameters that are adjusted to collect atmospheric pollutant concentration data.

[0132] Step C4: Combine the above pollutant concentration data, geographical data, and meteorological data to obtain environmental fusion data.

[0133] Geographic data can include information such as geographic location and topography, which can be obtained through map services or geographic information systems (GIS).

[0134] Meteorological data can include data on meteorological conditions such as temperature, humidity, and wind speed.

[0135] In practice, these data are matched in time and space, and the matched data are merged to generate environmental fusion data.

[0136] Step D4: Process the above-mentioned environmental fusion data using a preset neural network model to obtain the air pollution analysis results.

[0137] In practice, the environmental fusion data is transformed into a format suitable for input to the neural network model. A pre-defined neural network model is then loaded, and the prepared environmental fusion data is input into the model. The neural network model processes the input data to calculate predicted values ​​or analytical results of air pollution. Finally, analytical results, such as pollutant concentration predictions and pollution trends, are extracted from the model output.

[0138] The above-described implementation method of this application adjusts the working parameters by optimizing the adjustment coefficient, using the working parameters to collect pollutant concentration data, and combining the collected pollutant concentration data with geographical and meteorological data to obtain environmental fusion data that comprehensively reflects the environmental conditions. Then, the environmental fusion data is processed by a preset neural network model to obtain accurate air pollution analysis results.

[0139] To obtain accurate air pollution analysis results, in some implementations, the aforementioned preset neural network model includes convolutional neural networks, graph neural networks, long short-term memory networks, and fully connected networks. D4 includes steps D41 to D44:

[0140] Step D41: Extract features from the above-mentioned environmental fusion data using the above-mentioned convolutional neural network to obtain the spatiotemporal features of the above-mentioned environmental fusion data.

[0141] In its implementation, environmental fusion data combines pollutant concentrations, geographic information, and meteorological data. The data is typically represented in matrix form, containing time series data, spatial location information, and corresponding environmental variables. Convolutional operations are applied to extract spatial features from the data; the convolution kernel scans the spatial dimensions of the environmental fusion data to extract local features. The convolutional operations extract the spatial features of the data and combine them with time series information to generate spatiotemporal features. These features can represent pollutant concentration patterns and trends at different time points and spatial locations.

[0142] As another example, the input data can be standardized or normalized so that convolutional neural networks can process it better.

[0143] Step D42: Extract features from the above-mentioned environmental fusion data using the graph neural network described above to obtain the pollution diffusion characteristics of the above-mentioned environmental fusion data.

[0144] In the specific implementation, the environmental fusion data is transformed into a graph structure, and a feature vector is assigned to each node in the graph. These vectors typically include pollutant concentrations, meteorological data, etc. Graph convolution operations are applied to process the graph data, and the feature representations of the nodes are updated through graph convolutional layers, so that the features of each node include information from its neighboring nodes, thereby better capturing the spatial characteristics of pollution diffusion and obtaining the pollution diffusion characteristics of the aforementioned environmental fusion data.

[0145] Step D43: Input the above spatiotemporal features and pollution diffusion features into the above long short-term memory network to obtain the temporal dependence features of air pollution.

[0146] In the specific implementation, the spatiotemporal features extracted by the convolutional neural network are fused with the pollution diffusion features extracted by the graph neural network. For example, the two feature sets are merged into a larger feature vector or matrix. The fused feature vector is then input into a Long Short-Term Memory (LSTM) layer. The LSTM network processes the fused features and extracts temporal dependent features to obtain the temporal dependent features of air pollution. These features represent the changing patterns of air pollution over time and its long-term trends.

[0147] Step D44: Input the above time-series dependent features into the above fully connected network to obtain the air pollution analysis results.

[0148] In the specific implementation, the temporal dependency features obtained from the LSTM network are used as the input to the fully connected network. The fully connected network includes one or more fully connected layers, and each neuron in each layer is connected to all neurons in the previous layer. Based on the output of the fully connected network, interpretable analysis results are generated, such as predicted values, trend charts, or classification results of air pollution, thus obtaining the air pollution analysis results.

[0149] The above-described implementation method of this application extracts features from the environmental fusion data using a convolutional neural network to obtain the spatiotemporal features of the environmental fusion data. Then, it extracts features from the environmental fusion data using a graph neural network to obtain the pollution diffusion features of the environmental fusion data. Finally, it uses a long short-term memory network to integrate the pollution diffusion features and spatiotemporal features to obtain the time-dependent features. Finally, it obtains the air pollution analysis results based on a fully connected network. By performing feature extraction from different perspectives and in multiple aspects, accurate air pollution analysis results can be obtained.

[0150] In order to accurately obtain the first performance index value corresponding to the data acquisition process, in some embodiments, the above-mentioned S103 includes steps A5 to B5:

[0151] Step A5: After adjusting the above operating parameters, obtain the data quality index values, energy consumption, and data transmission volume during the data acquisition process.

[0152] In practice, after adjusting the aforementioned operating parameters, the data acquisition process begins, recording relevant quality indicator values ​​and data transmission volume. Energy consumption data is recorded by monitoring energy usage.

[0153] Step B5: Perform a weighted summation of the above data quality index values, the above energy consumption, and the above data transmission volume to obtain the first performance index value corresponding to the data acquisition process.

[0154] In the implementation, to comprehensively consider the impact of data quality, energy consumption, and data transmission volume, appropriate weights are pre-assigned to each indicator. These weights represent the importance of each indicator in the overall performance evaluation. The different indicators are then standardized, converting them into a unified range. Finally, the weighted values ​​of all indicators are summed to obtain the first performance indicator value corresponding to the data acquisition process.

[0155] The above-described embodiments of this application obtain the data quality index value, energy consumption, and data transmission volume during the data acquisition process by adjusting the working parameters, and then perform a weighted summation to accurately obtain the first performance index value corresponding to the data acquisition process.

[0156] In some implementations, a fixed-size sliding window W is defined during the acquisition of pollutant concentrations to assess the environmental state in real time. For the pollutant concentration c(t) at time t, we calculate the following indicators:

[0157] Mean μ(t):μ(t)=(1 / W)*Σ(c(i)), i from t-W+1to t;

[0158] Standard deviation σ(t):σ(t)=sqrt((1 / W)*Σ((c(i)-μ(t))^2)), i from t-W+1to t;

[0159] Trend of change (linear regression slope): k(t): k(t)=(W*Σ(i*c(i))-Σ(i)*Σ(c(i))) / (W*Σ(i^2)-(Σ(i))^2), i from t-W+1to t.

[0160] Based on the environmental condition assessment results, the operating parameters of the ISAC equipment are dynamically adjusted, for example:

[0161] The sampling frequency is adjusted to: f(t): f(t)=f_base+α*|σ(t)|+β*|k(t)|, where f_base is the base sampling frequency, and α and β are adjustment coefficients.

[0162] The quantization precision is adjusted to: b(t): b(t)=b_min+floor((c(t)-c_min) / (c_max-c_min)*(b_max-b_min)), where b_min and b_max are the minimum and maximum quantization bits, respectively, and c_min and c_max are the historical minimum and maximum values ​​of pollutant concentration.

[0163] The data compression ratio is adjusted to: r(t): r(t)=r_max-γ*(|σ(t)| / σ_max), where r_max is the maximum compression ratio, γ is the adjustment coefficient, and σ_max is the preset maximum standard deviation.

[0164] Then we can define a comprehensive performance index J(t):

[0165] J(t)=w1*Q(t)+w2*(1 / E(t))+w3*(1 / T(t)).

[0166] Where Q(t) is the data quality indicator, which can be represented by the signal-to-noise ratio (SNR); E(t) is the energy consumption; T(t) is the data transmission volume; and w1, w2, and w3 are weighting coefficients.

[0167] To maximize J(t), the gradient descent method can be used to continuously optimize the parameter adjustment strategy. The specific adjustment formula is as follows:

[0168]

[0169] Where θ represents the parameters to be optimized (such as α, β, γ), and η is the learning rate.

[0170] Anomaly detection is performed using the 3-sigma rule. If |c(t)-μ(t)|>3σ(t), an anomaly handling mechanism is triggered, temporarily increasing the sampling frequency and quantization precision to their maximum values, while reducing the data compression rate to ensure data integrity. Furthermore, adjacent ISAC devices can exchange data and perform collaborative optimization. Specifically, a collaborative metric is defined as: C(t) = ρ*J_local(t) + (1-ρ)*avg(J_neighbor(t)), and this metric is optimized, where ρ is the trade-off coefficient between local and neighbor performance.

[0171] In addition, the process of environmental signal feature extraction can refer to Figure 2 ,like Figure 2 As shown, the specific algorithm flow includes the following steps:

[0172] Step S201: Perform preprocessing such as normalization and detrending on the acquired raw environmental signals to eliminate the effects of sensor bias and long-term drift.

[0173] Step S202: Based on the statistical characteristics of the signal, adaptively select the optimal wavelet basis function.

[0174] Step S203: Perform multi-scale time-frequency decomposition on the signal based on the optimal wavelet basis function and decomposition scale. The wavelet coefficients reflect the energy distribution of the signal at different scales.

[0175] Specifically, let the environmental signal be f(t) and the wavelet basis function be ψ(t), then the wavelet decomposition formula is:

[0176]

[0177] Where a is the scale parameter, b is the translation parameter, and * denotes complex conjugation. W f (a,b) are wavelet coefficients, which reflect the projection coefficients of the signal at scale a and time b.

[0178] Step S204: Determine the denoising threshold, which includes at least one of a soft threshold or a hard threshold.

[0179] Step S205: Denoise the wavelet decomposition results according to the denoising threshold. At each scale, the threshold is adaptively set according to the statistical distribution of the wavelet coefficients. Soft or hard thresholding is applied to the wavelet coefficients to filter out high-frequency noise while retaining useful signal components that reflect changes in pollutant concentration.

[0180] Specifically, let the wavelet coefficients at the j-th scale be W. j The adaptive threshold is λ j The soft threshold denoising formula is:

[0181]

[0182] in, These are the denoised wavelet coefficients. Threshold λ j The settings can be adaptively configured based on the statistical characteristics of the wavelet coefficients, such as: Where, σ j Let N be the standard deviation of the wavelet coefficients at the j-th scale. j This represents the number of coefficients at this scale.

[0183] Step S206: Extract features from the denoised wavelet decomposition results. Calculate the energy distribution at different scales in the denoised wavelet coefficients and extract key frequency band energy features that reflect changes in pollutant concentration.

[0184] Specifically, let the wavelet coefficients after denoising at the j-th scale be... The energy characteristics at this scale are:

[0185]

[0186] Step S207: Using methods such as weighted fusion or nonlinear mapping, the energy features extracted at different scales are combined to form a multi-scale fused feature vector. This vector achieves data compression while preserving key signal information.

[0187] Specifically, let the extracted M scale energy features be E1, E2, ..., E M Then the multi-scale fusion feature is: F=[w1E1,w2E2,…,w M E M ] T Among them, w1, w2, etc. are the weight coefficients for each scale, which can be adaptively adjusted according to environmental conditions.

[0188] Step S208: Based on changes in environmental conditions, dynamically adjust the weighting coefficients of features at each scale, giving higher weights to scales with stronger ability to distinguish pollutant concentrations, thereby improving the adaptability and robustness of the features.

[0189] Step S209: Feature selection is performed based on the fused features and weights.

[0190] Step S210: Train the inversion model to obtain a concentration inversion model that can make accurate predictions.

[0191] Step S211: Process the selected features based on the inversion model to obtain pollutant concentration estimates. Input the extracted multi-scale fusion features into the pre-trained concentration inversion model to achieve real-time estimation of pollutant concentration. The inversion model can be a machine learning-based regression model, such as support vector regression or random forest.

[0192] Specifically, let the multi-scale fusion feature be F, and the concentration inversion model be g(·), then the estimated pollutant concentration is... The inversion model g(·) can be trained using machine learning methods such as support vector regression.

[0193] Further reference Figure 3 To address the characteristics of heterogeneous data, a multi-view autoencoder network can be designed to map data from different sources and modalities to a shared latent space, uncovering the intrinsic connections between them. Based on this, a graph convolutional neural network is used to model the complex relationships between "site-pollutant-influencing factors" during the pollution process. A causal inference mechanism is also introduced to explore the causes of pollution and clarify the causal chains of various factors. Through the fusion of multi-dimensional information, the spatiotemporal distribution patterns of pollution are comprehensively depicted, specifically including the following steps.

[0194] Step S301: Input multi-source heterogeneous data, including environmental monitoring data, meteorological data, geographic information data, emission source data, and satellite remote sensing data.

[0195] Step S302: Clean and standardize heterogeneous data through data preprocessing.

[0196] Specifically, step S302 includes operations such as step S3021 missing value imputation, step S3022 outlier detection, and step S3023 data standardization.

[0197] Step S303: Perform data association on the preprocessed data. During the data association process, the association and mapping of data from different sources and in different formats are achieved through techniques such as spatiotemporal alignment in step S3031, feature matching in step S3032, and semantic mapping in step S3033.

[0198] Specifically, in the S3031 spatiotemporal alignment process, firstly, spatial interpolation algorithms such as Kriging are used to unify monitoring data with different spatial resolutions onto a regular grid; then, Kalman filtering data assimilation technology is applied to align data with different time granularities to a unified time reference. Through this spatiotemporal alignment, we have achieved a consistent representation of multi-source data such as pollutant concentrations, meteorological parameters, and emission source inventories in the spatiotemporal dimensions.

[0199] In the S3032 feature matching process, considering the differences in physical quantities and units of measurement among monitoring data from different sources, firstly, physical mechanism models of atmospheric pollution processes, such as convection-diffusion equations, are used to establish conversion relationships between different physical quantities. Then, statistical learning methods such as canonical correlation analysis are employed to identify homogeneous features from different data sources from a data-driven perspective. Based on this, a unified feature representation space for multi-source monitoring data is constructed, achieving feature normalization and standardization under different units of measurement such as mass concentration, volume concentration, and emission flux.

[0200] In the S3033 semantic mapping process, to address the differences in concepts and semantics across environmental monitoring, meteorology, and emission inventories, a general ontology covering all areas of air pollution monitoring was first constructed, defining core concepts and relationships such as pollutants, meteorological parameters, and emission source types. Then, ontology matching algorithms, such as semantic similarity calculation, were used to map ontologs from different fields onto this general ontology, achieving a unified semantic representation. Simultaneously, knowledge graph technology was utilized to link concepts, entities, and relationships within the ontology with multi-source monitoring data, forming a semantically rich knowledge base for air pollution monitoring.

[0201] Step S304: Perform data fusion, using algorithms such as multi-view representation learning, graph neural networks, and causal inference to mine the inherent connections and patterns in the data from multiple perspectives and multiple relationships.

[0202] Step S305: Perform environmental process modeling, input the fused data into the environmental process modeling, and establish a machine learning model that can reflect the spatiotemporal evolution of pollution.

[0203] Specifically, based on data-driven approaches, a machine learning model of the pollution process is constructed by integrating classic atmospheric diffusion mechanism models. Employing a spatiotemporal series prediction paradigm, an end-to-end prediction model is established, taking pollution monitoring data, meteorological conditions, and emission source information as inputs, and outputting the spatiotemporal distribution of pollution concentrations. Attention mechanisms and physical constraints are introduced to enhance the model's interpretability and generalization ability. The model can be used for tasks such as pollution source tracing and pollution concentration forecasting.

[0204] For example, after inputting the multi-source air pollution monitoring data after correlation and fusion, an end-to-end spatiotemporal sequence prediction modeling paradigm is adopted. A deep neural network model that integrates spatiotemporal feature extraction, long short-term memory, and attention mechanism is designed. Then, convolutional neural networks are used to extract local spatiotemporal features, and graph neural networks are used to extract inter-regional propagation correlation features.

[0205] Specifically:

[0206] Convolutional Neural Network Feature Extraction: Let the input spatiotemporal contamination data be... Where T is the time step, H and W are the spatial dimensions, and C is the type of pollutant, the formula for calculating the convolutional layer is: Where * represents the convolution operation, W (l) and b (l) These are the weights and biases of the l-th convolutional layer, respectively, X (l-1) This is the output of the (l-1)th layer.

[0207] Graph Neural Network Feature Extraction: Let the graph formed by pollution monitoring stations be G = (V, E), where V is the set of nodes and E is the set of edges. The formula for calculating the graph convolutional layer is:

[0208]

[0209] in, To add self-connected adjacency matrices, for The degree matrix, H (l) Let W be the feature matrix of the nodes in the l-th layer. (l) Let σ be the weight matrix of the l-th layer, and σ be the activation function.

[0210] The extracted spatiotemporal features are then input into a Long Short-Term Memory (LSTM) network to characterize the temporal dependence and trend of the pollution process.

[0211] An attention mechanism can be introduced on top of LSTM to focus on the most important historical moments and spatial regions for prediction. The learned spatiotemporal features are mapped to predicted concentration distribution values ​​for future moments through fully connected network layers. A supervised learning paradigm incorporating physical constraints is used to train the model, introducing physical constraint regularization terms such as the mass conservation equation and convection-diffusion equation into the loss function. Data augmentation and adversarial learning techniques are employed to enhance the model's generalization and robustness.

[0212] Step S306: Based on the environmental process model, further conduct pollution source analysis and pollution concentration forecasting, and output intelligent analysis results of air pollution.

[0213] In the embodiments described above, the correlation and fusion analysis of multi-source data expands the depth and breadth of applications for environmental monitoring data. Traditional environmental monitoring data analysis is mostly limited to a single data source, making it difficult to comprehensively depict complex pollution processes. The multi-source data correlation and fusion method proposed in this application can integrate environmental monitoring data with meteorological, geographical, and emission data in multiple dimensions, constructing a high-dimensional feature space that comprehensively reflects the pollution process. This provides a data foundation for a deeper understanding of pollution formation mechanisms and for conducting multi-scenario analysis applications. The modeling method that integrates physical mechanisms and machine learning achieves accurate characterization and prediction of pollution processes. Existing pollution process models are mostly based on simplified physical mechanisms or statistical relationships, making it difficult to accurately reflect the complexity of the real world. The fusion modeling method proposed in this application, by combining classical physical models with deep learning models, ensures the accuracy of physical mechanisms while fully utilizing data-driven machine learning algorithms to characterize the complex nonlinear relationships of pollution processes. This achieves accurate, efficient, and universal modeling of pollution processes, providing strong support for intelligent analysis and decision-making applications.

[0214] Based on the air pollution monitoring method provided in the above embodiments, this application also provides specific implementation methods for air pollution monitoring devices. Please refer to the following embodiments.

[0215] First see Figure 4 The air pollution monitoring device 400 provided in this application embodiment includes the following modules:

[0216] The acquisition module 401 is used to acquire atmospheric data based on the working parameters of the acquisition device to obtain atmospheric pollutant concentration data within a preset time range.

[0217] The adjustment module 402 is used to adaptively adjust the working parameters according to the statistical characteristics of the atmospheric pollutant concentration data within the set time range and the adjustment coefficient. The adaptively adjusted working parameters correspond to the statistical characteristics.

[0218] The calculation module 403 is used to calculate the first performance index value corresponding to the atmospheric data acquisition process after adjusting the above-mentioned operating parameters of the above-mentioned acquisition equipment.

[0219] The optimization module 404 is used to optimize the adjustment coefficient according to the first performance index value by using the gradient descent method, so that when the working parameters are adjusted based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value.

[0220] The air pollution monitoring device provided in this application continuously adjusts the operating parameters of the acquisition equipment according to real-time air pollutant concentration data and adjustment coefficients, enabling more rational allocation of system resources and improving acquisition efficiency during the acquisition process. Furthermore, a corresponding first performance index value is calculated, and the adjustment coefficient is optimized according to this first performance index value, continuously improving the adjustment coefficient through iterative processes. Since the operating parameters are adjusted according to actual atmospheric conditions to adapt to the current situation, the adjusted acquisition parameters allow for more efficient acquisition of changes in air pollutant concentrations based on the actual environment, enabling rapid response to environmental changes and quick assessment of air pollution conditions.

[0221] As one implementation of this application, the acquisition module 401 may include:

[0222] The acquisition unit is used to acquire pollutant concentration values ​​at various time points within a preset time range according to the working parameters of the acquisition device.

[0223] The calculation unit is used to calculate the average value of pollutant concentration at each of the above time points to obtain the average pollutant concentration.

[0224] The calculation unit is also used to calculate the square of the difference between the above-mentioned pollutant concentration value and the above-mentioned average pollutant concentration at each time point.

[0225] The calculation unit is also used to calculate the average of each of the above squares, and take the square root of the average of the above squares to obtain the standard deviation of the pollutant concentration.

[0226] The determination unit is used to determine the trend of pollutant concentration change based on the rate of change of each pollutant concentration value over time.

[0227] The determining unit is also used to take the pollutant concentration values ​​corresponding to each of the above time points, the average pollutant concentration, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change as atmospheric pollutant concentration data within a preset time range.

[0228] The embodiments described above in this application collect pollutant concentration values ​​at various time points within a preset time range according to the working parameters of the data acquisition device. Then, the mean concentration and standard deviation of each pollutant concentration value are calculated. The pollutant concentration values ​​at each time point, the mean concentration, the standard deviation, and the trend of pollutant concentration change are then used as atmospheric pollutant concentration data within the preset time range. By comprehensively utilizing multiple data sources as atmospheric pollutant concentration data within the preset time range, the atmospheric pollutant concentration data can fully reflect the actual situation of air pollution.

[0229] As one implementation of this application, adjustment module 402 may include:

[0230] The determination unit is used to determine the sampling frequency increase value according to the adjustment coefficient, the standard deviation of the pollutant concentration and the trend of the pollutant concentration change.

[0231] The adjustment unit is used to determine the target sampling frequency based on the preset basic sampling frequency and the above-mentioned sampling frequency boost value, and adjust the sampling frequency to the above-mentioned target sampling frequency.

[0232] The adjustment unit is also used to adjust the quantification accuracy based on the current pollutant concentration value and the extreme value of pollutant concentration within a preset time range.

[0233] The adjustment unit is also used to adjust the data compression rate based on the ratio of the pollutant concentration to the preset standard deviation.

[0234] The above-described implementation method of this application adjusts the sampling frequency according to the adjustment coefficient, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change. It adjusts the quantization accuracy based on the current pollutant concentration value and the extreme value of pollutant concentration within a preset time range. It adjusts the data compression rate according to the ratio of the pollutant concentration to the preset standard deviation, thereby accurately adjusting each working parameter.

[0235] As one implementation of this application, the air pollution monitoring device 400 may further include:

[0236] The decomposition module is used to perform wavelet decomposition on the above-mentioned atmospheric pollutant concentration data through a preset wavelet basis function to obtain wavelet coefficients, which represent the local characteristics of the above-mentioned atmospheric pollutant concentration data at multiple different scales.

[0237] The extraction module is used to extract features from the wavelet coefficients mentioned above to obtain pollutant concentration features at multiple different scales.

[0238] The combination module is used to combine the above-mentioned pollutant concentration features at multiple different scales to obtain multi-scale fused features.

[0239] The determination module is used to input the above-mentioned multi-scale fusion features into the preset prediction model to obtain the predicted atmospheric pollutant concentration data corresponding to the preset time, wherein the preset time is later than the preset time range.

[0240] The above-described implementation method of this application performs wavelet decomposition on the atmospheric pollutant concentration data using a preset wavelet basis function to obtain wavelet coefficients. Then, feature extraction is performed on the wavelet coefficients to obtain multi-scale fusion features. A preset prediction model is used to process these multi-scale fusion features to obtain atmospheric pollutant concentration prediction data corresponding to a preset time. By using wavelet decomposition and multi-scale feature fusion as concentration prediction data, accurate atmospheric pollutant concentration prediction data can be obtained.

[0241] As one implementation of this application, the decomposition module may include:

[0242] The acquisition unit is used to acquire the decomposition scale corresponding to the preset wavelet basis function.

[0243] The decomposition unit is used to perform wavelet decomposition on the above-mentioned atmospheric pollutant concentration data according to the decomposition scale corresponding to the preset wavelet basis function, so as to obtain the initial wavelet coefficients.

[0244] The analysis unit is used to perform statistical analysis on the initial wavelet coefficients and determine their statistical characteristics.

[0245] The denoising unit is used to denoise the initial wavelet coefficients based on the threshold corresponding to the above statistical features to obtain the wavelet coefficients.

[0246] The above-described implementation method of this application obtains the decomposition scale corresponding to the preset wavelet basis function, performs wavelet decomposition on the atmospheric pollutant concentration data according to this decomposition scale, and then performs denoising processing on the initial wavelet coefficients to remove useless information, thereby obtaining more accurate wavelet coefficients.

[0247] As one implementation of this application, the air pollution monitoring device 400 may further include:

[0248] The adjustment module is used to adjust the working parameters based on the optimized adjustment coefficients.

[0249] The data acquisition module is used to collect pollutant concentration data according to the above operating parameters.

[0250] The combination module is used to combine the above-mentioned pollutant concentration data, geographical data, and meteorological data to obtain environmental fusion data.

[0251] The processing module is used to process the above-mentioned environmental fusion data through a preset neural network model to obtain air pollution analysis results.

[0252] The above-described implementation method of this application adjusts the working parameters by optimizing the adjustment coefficient, using the working parameters to collect pollutant concentration data, and combining the collected pollutant concentration data with geographical and meteorological data to obtain environmental fusion data that comprehensively reflects the environmental conditions. Then, the environmental fusion data is processed by a preset neural network model to obtain accurate air pollution analysis results.

[0253] As one implementation of this application, the processing module may include:

[0254] The extraction unit is used to extract features from the environmental fusion data through the convolutional neural network to obtain the spatiotemporal features of the environmental fusion data.

[0255] The extraction unit is also used to extract features from the environmental fusion data using the graph neural network described above, so as to obtain the pollution diffusion features of the environmental fusion data.

[0256] The determination unit is used to input the above-mentioned spatiotemporal features and pollution diffusion features into the above-mentioned long short-term memory network to obtain the temporal dependence features of air pollution.

[0257] The determination unit is also used to input the aforementioned time-dependent features into the aforementioned fully connected network to obtain the air pollution analysis results.

[0258] The above-described implementation method of this application extracts features from the environmental fusion data using a convolutional neural network to obtain the spatiotemporal features of the environmental fusion data. Then, it extracts features from the environmental fusion data using a graph neural network to obtain the pollution diffusion features of the environmental fusion data. Finally, it uses a long short-term memory network to integrate the pollution diffusion features and spatiotemporal features to obtain the time-dependent features. Finally, it obtains the air pollution analysis results based on a fully connected network. By performing feature extraction from different perspectives and in multiple aspects, accurate air pollution analysis results can be obtained.

[0259] As one implementation of this application, the computing module 403 may include:

[0260] The acquisition unit is used to acquire data quality index values, energy consumption, and data transmission volume during the data acquisition process, after adjusting the above-mentioned working parameters.

[0261] The summation unit is used to perform a weighted summation of the above-mentioned data quality index values, the above-mentioned energy consumption, and the above-mentioned data transmission volume to obtain the first performance index value corresponding to the data acquisition process.

[0262] The above-described embodiments of this application obtain the data quality index value, energy consumption, and data transmission volume during the data acquisition process by adjusting the working parameters, and then perform a weighted summation to accurately obtain the first performance index value corresponding to the data acquisition process.

[0263] Each module in the air pollution monitoring device provided in this application embodiment can realize each step in the above-mentioned air pollution monitoring method and achieve the corresponding effect. For the sake of brevity, it will not be described in detail here.

[0264] Figure 5 A schematic diagram of the structure of the air pollution monitoring hardware provided in the embodiments of this application is shown.

[0265] The air pollution monitoring equipment may include a processor 501 and a memory 502 storing computer program instructions.

[0266] Specifically, the processor 501 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0267] Memory 502 may include mass storage for data or instructions. For example, and not limitingly, memory 502 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 502 may include removable or non-removable (or fixed) media. Where appropriate, memory 502 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 502 is non-volatile solid-state memory.

[0268] The memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the air pollution monitoring method according to any embodiment of this disclosure.

[0269] The processor 501 reads and executes computer program instructions stored in the memory 502 to implement any of the air pollution monitoring methods in the above embodiments.

[0270] In one example, the air pollution monitoring device may also include a communication interface 503 and a bus 510. For example, Figure 5 As shown, the processor 501, memory 502, and communication interface 503 are connected through bus 510 and complete communication with each other.

[0271] The communication interface 503 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0272] Bus 510 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 510 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0273] Furthermore, in conjunction with the methods for air pollution monitoring described in the above embodiments, this application embodiment can provide a computer storage medium for implementation. This computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the air pollution monitoring methods described in the above embodiments.

[0274] This application also provides a computer program product, including a computer program that, when executed, implements any of the air pollution monitoring methods described in the above embodiments.

[0275] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0276] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0277] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0278] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0279] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for monitoring air pollution, characterized in that, include: By collecting atmospheric data based on operating parameters using the data acquisition equipment, atmospheric pollutant concentration data within a preset time range can be obtained. The operating parameters include sampling frequency, quantization accuracy, and data compression rate; Based on the statistical characteristics of the air pollutant concentration data within the preset time range, the working parameters are adaptively adjusted according to the adjustment coefficient, and the adaptively adjusted working parameters correspond to the statistical characteristics. With the operating parameters of the acquisition device adjusted, the first performance index value corresponding to the atmospheric data acquisition process is calculated; Based on the first performance index value, the adjustment coefficient is optimized by gradient descent so that when the working parameters are adjusted based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value. The step of acquiring atmospheric data based on operating parameters using a data acquisition device to obtain atmospheric pollutant concentration data within a preset time range includes: The data acquisition device collects pollutant concentration values ​​at various time points within a preset time range according to the operating parameters. Calculate the average value of pollutant concentration at each time point to obtain the average pollutant concentration. Calculate the square of the difference between the pollutant concentration value at each time point and the mean pollutant concentration; Calculate the average of each of the squares, and take the square root of the average of the squares to obtain the standard deviation of the pollutant concentration; The trend of pollutant concentration change is determined based on the rate of change of each pollutant concentration value over time. The pollutant concentration values ​​at each time point, the average pollutant concentration, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change are used as atmospheric pollutant concentration data within a preset time range. The step of adaptively adjusting the working parameters according to an adjustment coefficient based on the statistical characteristics of the air pollutant concentration data within the preset time range includes: The sampling frequency increase value is determined based on the adjustment coefficient, the standard deviation of the pollutant concentration, and the trend of the pollutant concentration change; The target sampling frequency is determined based on the preset base sampling frequency and the sampling frequency boost value, and the sampling frequency is adjusted to the target sampling frequency. The quantification accuracy is adjusted based on the current pollutant concentration value and the extreme value of pollutant concentration within a preset time range; The data compression rate is adjusted based on the ratio of the pollutant concentration to the preset standard deviation; The method further includes, after optimizing the adjustment coefficient using gradient descent according to the first performance index value: Adjust the working parameters based on the optimized adjustment coefficients; Collect pollutant concentration data according to the aforementioned operating parameters; The pollutant concentration data, corresponding geographical data, and meteorological data are combined to obtain environmental fusion data. The environmental fusion data is processed by a preset neural network model to obtain air pollution analysis results; The step of calculating the first performance index value corresponding to the atmospheric data acquisition process after adjusting the operating parameters of the acquisition device includes: With the operating parameters adjusted, the data quality index values, energy consumption, and data transmission volume during the data acquisition process are obtained. The data quality index value, the energy consumption, and the data transmission volume are weighted and summed to obtain the first performance index value corresponding to the data acquisition process.

2. The air pollution monitoring method according to claim 1, characterized in that, After acquiring atmospheric data based on operating parameters using a data acquisition device to obtain atmospheric pollutant concentration data within a preset time range, the method further includes: The atmospheric pollutant concentration data is decomposed by a preset wavelet basis function to obtain wavelet coefficients, which represent the local characteristics of the atmospheric pollutant concentration data at multiple different scales. Feature extraction is performed on the wavelet coefficients to obtain pollutant concentration features at multiple different scales; The pollutant concentration features at multiple different scales are combined to obtain multi-scale fusion features; The multi-scale fusion features are input into a preset prediction model to obtain predicted atmospheric pollutant concentration data corresponding to a preset time, wherein the preset time is later than the preset time range.

3. The air pollution monitoring method according to claim 2, characterized in that, The step of performing wavelet decomposition on the atmospheric pollutant concentration data using a preset wavelet basis function to obtain wavelet coefficients includes: Obtain the decomposition scale corresponding to the preset wavelet basis function; According to the decomposition scale corresponding to the preset wavelet basis function, the atmospheric pollutant concentration data is decomposed by wavelet to obtain the initial wavelet coefficients. Statistical analysis is performed on the initial wavelet coefficients to determine their statistical characteristics; The initial wavelet coefficients are denoised based on the threshold corresponding to the statistical features to obtain the wavelet coefficients.

4. The air pollution monitoring method according to claim 1, characterized in that, The preset neural network model includes convolutional neural networks, graph neural networks, long short-term memory networks, and fully connected networks. The process of processing the environmental fusion data using the preset neural network model to obtain air pollution analysis results includes: The spatiotemporal features of the environmental fusion data are obtained by extracting features from the environmental fusion data using the convolutional neural network. The pollution diffusion characteristics of the environmental fusion data are obtained by extracting features from the environmental fusion data using the graph neural network. The spatiotemporal features and the pollution diffusion features are input into the long short-term memory network to obtain the time-dependent features of air pollution. The time-dependent features are input into the fully connected network to obtain the air pollution analysis results.

5. An air pollution monitoring device, characterized in that, The device includes: The data acquisition module is used to acquire atmospheric data based on operating parameters through the acquisition device to obtain atmospheric pollutant concentration data within a preset time range; the operating parameters include sampling frequency, quantization accuracy, and data compression rate. The adjustment module is used to adaptively adjust the working parameters according to the statistical characteristics of the atmospheric pollutant concentration data within the preset time range, and the adaptively adjusted working parameters correspond to the statistical characteristics. The calculation module is used to calculate the first performance index value corresponding to the atmospheric data acquisition process after adjusting the operating parameters of the acquisition device. The optimization module is used to optimize the adjustment coefficient according to the first performance index value using the gradient descent method, so that when the working parameters are adjusted based on the optimized adjustment coefficient, the second performance index value of the corresponding data acquisition process is greater than the first performance index value. The data acquisition module is further configured to: acquire pollutant concentration values ​​at various time points within a preset time range according to operating parameters using acquisition equipment; calculate the average value of the pollutant concentration values ​​at each time point to obtain the pollutant concentration mean; calculate the square of the difference between the pollutant concentration value at each time point and the pollutant concentration mean; calculate the average of each square, and take the square root of the average of the squares to obtain the pollutant concentration standard deviation; determine the pollutant concentration change trend based on the rate of change of each pollutant concentration value over time; and use the pollutant concentration values ​​at each time point, the pollutant concentration mean, the pollutant concentration standard deviation, and the pollutant concentration change trend as atmospheric pollutant concentration data within the preset time range. The adjustment module is also used to determine the sampling frequency boost value according to the adjustment coefficient, the standard deviation of the pollutant concentration, and the trend of pollutant concentration change; determine the target sampling frequency according to the preset basic sampling frequency and the sampling frequency boost value, and adjust the sampling frequency to the target sampling frequency; adjust the quantization accuracy according to the current pollutant concentration value and the extreme value of pollutant concentration within a preset time range; and adjust the data compression rate according to the ratio of the pollutant concentration to the preset standard deviation. The optimization module is used to adjust the working parameters based on the optimized adjustment coefficients; collect pollutant concentration data according to the working parameters; combine the pollutant concentration data, corresponding geographical data, and meteorological data to obtain environmental fusion data; and process the environmental fusion data through a preset neural network model to obtain air pollution analysis results. The calculation module is used to obtain data quality index values, energy consumption, and data transmission volume during the data acquisition process, after adjusting the operating parameters; and to perform a weighted summation of the data quality index values, energy consumption, and data transmission volume to obtain the first performance index value corresponding to the data acquisition process.

6. An air pollution monitoring device, characterized in that, The device includes: a processor and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the air pollution monitoring method as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the air pollution monitoring method as described in any one of claims 1-4.

8. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the air pollution monitoring method as described in any one of claims 1-4.

Images