CN116881671A - Atmospheric pollution tracing method and system based on neural network - Google Patents
Atmospheric pollution tracing method and system based on neural network Download PDFInfo
- Publication number
- CN116881671A CN116881671A CN202311126298.6A CN202311126298A CN116881671A CN 116881671 A CN116881671 A CN 116881671A CN 202311126298 A CN202311126298 A CN 202311126298A CN 116881671 A CN116881671 A CN 116881671A
- Authority
- CN
- China
- Prior art keywords
- data
- pollution
- monitoring data
- sub
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 113
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 35
- 238000012544 monitoring process Methods 0.000 claims abstract description 106
- 239000003344 environmental pollutant Substances 0.000 claims abstract description 63
- 231100000719 pollutant Toxicity 0.000 claims abstract description 63
- 238000004458 analytical method Methods 0.000 claims abstract description 33
- 238000004088 simulation Methods 0.000 claims abstract description 22
- 238000012937 correction Methods 0.000 claims abstract description 18
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000003915 air pollution Methods 0.000 claims abstract description 6
- 238000003062 neural network model Methods 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims abstract description 5
- 238000005065 mining Methods 0.000 claims abstract description 4
- 230000008859 change Effects 0.000 claims description 28
- 230000006870 function Effects 0.000 claims description 26
- 230000008569 process Effects 0.000 claims description 23
- 238000009826 distribution Methods 0.000 claims description 18
- 230000002159 abnormal effect Effects 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 12
- 241001123248 Arma Species 0.000 claims description 10
- 238000007476 Maximum Likelihood Methods 0.000 claims description 10
- 239000000654 additive Substances 0.000 claims description 10
- 230000000996 additive effect Effects 0.000 claims description 10
- 230000007613 environmental effect Effects 0.000 claims description 9
- 230000033001 locomotion Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000005311 autocorrelation function Methods 0.000 claims description 6
- 238000010206 sensitivity analysis Methods 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 3
- 238000001744 unit root test Methods 0.000 claims description 3
- 238000003648 Ljung–Box test Methods 0.000 claims description 2
- 238000012300 Sequence Analysis Methods 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 description 15
- 239000011159 matrix material Substances 0.000 description 10
- 239000000356 contaminant Substances 0.000 description 7
- 239000013598 vector Substances 0.000 description 7
- 238000009792 diffusion process Methods 0.000 description 6
- 238000000540 analysis of variance Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000013210 evaluation model Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 101100365516 Mus musculus Psat1 gene Proteins 0.000 description 2
- CBENFWSGALASAD-UHFFFAOYSA-N Ozone Chemical compound [O-][O+]=O CBENFWSGALASAD-UHFFFAOYSA-N 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000011109 contamination Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000012731 temporal analysis Methods 0.000 description 2
- 238000000700 time series analysis Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000001373 regressive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004062 sedimentation Methods 0.000 description 1
- 238000013112 stability test Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/0004—Gaseous mixtures, e.g. polluted air
- G01N33/0009—General constructional details of gas analysers, e.g. portable test equipment
- G01N33/0027—General constructional details of gas analysers, e.g. portable test equipment concerning the detector
- G01N33/0036—General constructional details of gas analysers, e.g. portable test equipment concerning the detector specially adapted to detect a particular component
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Combustion & Propulsion (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses an atmosphere pollution tracing method and system based on a neural network, belonging to the technical field of air pollution simulation tracing, wherein the method comprises the following steps: reading and preprocessing mobile monitoring data of a mobile detection station to obtain standardized mobile monitoring data; mining a time-space evolution trend of the pollutants based on the mobile monitoring data to obtain a trend analysis result; reading multi-source pollutant data, constructing a pollution source correction module, taking the multi-source pollutant as input to obtain a predicted result of the pollutant, comparing the predicted result with a simulation result, and correcting the simulation result; and constructing a graph neural network model, and coding and training based on the corrected simulation result to obtain traceability data. The application has higher efficiency of tracing the atmospheric pollution and greatly improves the accuracy.
Description
Technical Field
The application relates to an atmosphere pollution tracing technology, in particular to an atmosphere pollution tracing method and system based on a neural network.
Background
With the development of industrialization and city, the problem of air pollution is increasingly serious, and great harm is brought to human health and ecological environment. In order to effectively control and remediate atmospheric pollution, accurate identification and localization of the pollution sources is required. However, due to the complex processes of diffusion, conversion, sedimentation and the like of atmospheric pollutants, and the characteristics of diversity, dispersibility, space-time variability and the like of pollution sources, the conventional atmospheric pollution source tracing method based on a fixed monitoring station is difficult to meet actual demands.
In order to solve the problem, an atmospheric pollution source tracing method based on a mobile monitoring station appears in recent years. The mobile monitoring station refers to equipment capable of performing air quality monitoring at different places, such as unmanned aerial vehicles, unmanned boats, automobiles and the like. Compared with a fixed monitoring station, the mobile monitoring station has the following advantages: (1) The method can cover a wider space range and improve the space resolution; (2) The monitoring route and time can be flexibly adjusted according to actual conditions, and the time sequence resolution is improved; (3) More dimensionality data such as position, speed, direction and the like can be acquired, and the data information quantity is improved. However, the atmospheric pollution source tracing method based on the mobile monitoring station also faces some challenges, such as: (1) how to deal with data noise and deviation generated by the mobile monitoring station under the influence of various factors in the movement process; (2) how to effectively control and fuse the quality of the data from different sources, different frequencies and different accuracies; (3) how to utilize mobile monitoring data to mine the time-space distribution trend of pollutants, and reversely deduce the position and the intensity of a pollution source based on the trend, and the accuracy and the high efficiency of the analysis method.
Accordingly, there is a need for improvements in the art that overcome the shortcomings of the prior art.
Disclosure of Invention
The application aims to: an air pollution tracing method based on a neural network is provided to solve the problems existing in the prior art.
The technical scheme is as follows: the air pollution tracing method based on the neural network comprises the following steps:
step S1, reading and preprocessing mobile monitoring data of a mobile detection station to obtain standardized mobile monitoring data, wherein the mobile monitoring data at least comprises monitored dye concentration, monitoring time and monitoring position coordinates;
s2, mining a time-space evolution trend of the pollutants based on the mobile monitoring data to obtain a trend analysis result;
s3, reading multi-source pollutant data, constructing a pollution source correction module, taking multi-source pollutants as input, obtaining a predicted result of the pollutants, comparing the predicted result with a simulation result, and correcting the simulation result;
and S4, constructing a graph neural network model, and coding and training based on the corrected simulation result to obtain traceable data, wherein the traceable data comprises pollution source identification, contribution of each pollution source, a pollution propagation path and a sensitivity analysis result.
According to one aspect of the present application, in the step S1, the process of preprocessing the movement monitoring data of the mobile detection station includes:
step S11, detecting and eliminating abnormal values of the mobile monitoring data, and eliminating noise and interference of the data;
step S12, setting an abnormal value judgment standard, and detecting and removing abnormal values in the mobile monitoring data;
step S13, performing spatial interpolation on the mobile monitoring data to fill in the data gaps in the space;
and S14, performing time synchronization on the mobile monitoring data so that the data of different mobile monitoring stations at the same time are corresponding.
According to one aspect of the present application, the step S2 is further:
step S21, dividing mobile monitoring data into a plurality of sub-areas according to the position information of the mobile monitoring stations, wherein each sub-area contains monitoring data of at least one mobile monitoring station in a preset time, and forming a time sequence of the monitoring data;
step S22, carrying out trend analysis on the monitoring data in each subarea based on a time sequence analysis method, and calculating the change rate and variance of the pollutant concentration along with time to be used as pollution trend indexes;
and S23, ranking and screening each sub-region according to the pollution trend index, selecting the sub-region which is ranked at the front and exceeds the threshold value as a candidate pollution source region, estimating the position, the type and the intensity of the pollution source, and outputting pollution source information and a trend analysis result.
According to one aspect of the present application, the step S3 further includes:
step S31, acquiring data of multi-source pollutants, including fixed monitoring station data, remote sensing image data, geographic information data and meteorological information data;
s32, constructing a pollution source correction module, wherein the pollution source correction module comprises a CAMX-OSAT-PSAT module; performing environmental correction on the pollution source correction module based on pre-stored environmental information;
and S33, taking multi-source pollutants as input, obtaining a predicted result of the pollutants, comparing the predicted result with a simulation result, comparing and evaluating the simulation result and the predicted result, calculating the similarity and error between the simulated result and the actual situation, sequencing and screening candidate pollution source areas according to the similarity and error, selecting an optimal pollution source area, and determining the position, the intensity and the type of the pollution source area.
According to one aspect of the present application, the step S4 further includes:
step S41, coding and embedding graph data by utilizing a graph neural network, and learning hidden characteristics of nodes and edges, and relations and similarities among the nodes;
and step S42, decoding and predicting the graph data by utilizing the graph neural network to realize different analysis targets.
According to an aspect of the application, the step S22 is further:
s22a, constructing an autoregressive moving average model, and judging whether the time sequence is stable or not by adopting a unit root test or a stationarity test; judging whether the time sequence is white noise or not by adopting an autocorrelation function or a partial autocorrelation function; if the time sequence does not meet the stationarity or white noise property, adopting a differential method to enable the time sequence to meet the assumption of an ARMA model;
step S22b, selecting an optimal model order by adopting an information criterion method or a residual analysis method, wherein the information criterion method comprises AIC, BIC, HQ; the residual analysis method comprises Ljung-Box detection and normalization detection;
step S22c, estimating parameters of the ARMA model by using a maximum likelihood estimation method or a least square estimation method, predicting future values of the monitoring data by using the estimated parameters and historical data, and calculating a prediction error as a change rate and variance to be used as a pollution trend index.
According to an aspect of the application, the step S23 is further:
step S23a, ranking according to the comprehensive value from high to low, and sequencing pollution trend indexes of each sub-area to obtain a ranking list of the sub-areas;
step S23b, screening pollution trend indexes of each sub-area, and selecting sub-areas with comprehensive values exceeding a threshold value or ranked in the top N bits as candidate pollution source areas according to a preset threshold value or proportion; n is a natural number;
and step S23c, estimating each candidate pollution source region, estimating the position, the type and the intensity of the pollution source by using a maximum likelihood method or a least square method according to the sub-region track and the monitoring data, and outputting pollution source information and a trend analysis result.
According to one aspect of the application, step S23b is further:
calculating the information entropy of the pollution trend index by using the shannon information entropy;
according to the change rule of the information entropy, determining an optimal threshold value so that the information entropy reaches the maximum or minimum;
and screening the sub-regions according to the optimal threshold or proportion, and selecting the sub-regions with the comprehensive value exceeding the threshold or ranked in the top N bits as candidate pollution source regions.
According to one aspect of the present application, step S22 is further:
s22a, constructing a nonlinear function set and a generalized additive model, analyzing distribution characteristics and change rules of pollutant monitoring data according to the mapping relation of pollutant data and nonlinear functions, and selecting a corresponding nonlinear function as a basis function of the generalized additive model;
step S22b, estimating parameters of a generalized additive model by using maximum likelihood estimation, wherein the parameters comprise coefficients of a basis function and smoothing parameters;
step S22c, fitting the pollutant concentration by using a generalized additive model, calculating residual errors between the fitting value and the actual value, and summing squares of the residual errors to obtain an estimated value of variance; and calculating the change rate of the pollutant concentration along with time, and taking the change rate and the variance as pollution trend indexes.
According to another aspect of the present application, an atmospheric pollution tracing system based on a neural network includes:
at least one processor; and a memory communicatively coupled to at least one of the processors; the memory stores instructions executable by the processor, where the instructions are configured to be executed by the processor to implement the neural network-based atmospheric pollution tracing method according to any one of the above technical solutions.
The beneficial effects are that: the application has higher efficiency of tracing the atmospheric pollution and greatly improves the accuracy. The relevant advantages will be described below in connection with the detailed description.
Drawings
Fig. 1 is a flow chart of the present application.
Fig. 2 is a flowchart of step S1 of the present application.
Fig. 3 is a flow chart of step S2 of the present application.
Fig. 4 is a flowchart of step S3 of the present application.
Detailed Description
As shown in fig. 1, the following examples are provided.
S1, reading pollution monitoring data of a mobile detection station and preprocessing the pollution monitoring data to obtain standardized monitoring data, wherein the pollution monitoring data comprises monitored dye concentration, monitoring time and monitoring position coordinates;
step S11: and (5) quality inspection. The sub-step is mainly to test the validity and integrity of the mobile monitoring data, reject invalid or missing data, and ensure the quality of the data. The specific implementation process is as follows:
first, the mobile monitoring data is checked for format and range, ensuring that the data meets the expected type and range of values, e.g., contaminant concentration should be non-negative, time stamp should be valid date and time format, etc.
Secondly, integrity checking is performed on the mobile monitoring data, and whether a missing value or a null value exists is detected, for example, a certain mobile monitoring station does not upload data in a certain time period, a certain pollutant index does not record, and the like.
Finally, the mobile monitoring data is checked for validity to detect the presence of unreasonable or unreliable data, e.g., a certain mobile monitoring station spans a significant distance in a short time, or a certain contaminant concentration exceeds a normal level, etc.
For invalid or missing data detected, strategies such as deletion, replacement or neglect can be adopted for processing, and the method depends on the importance and influence degree of the data.
Step S12: and (5) outlier rejection. The substep mainly detects and eliminates abnormal values of the mobile monitoring data, eliminates noise and interference of the data and improves the accuracy of the data. The specific implementation process is as follows:
firstly, carrying out statistical analysis on mobile monitoring data, and calculating basic statistics such as mean, variance, maximum value, minimum value, median, quartile and the like of each index so as to know the distribution characteristics of the data.
And secondly, detecting abnormal values of the mobile monitoring data, and selecting a proper abnormal value detection method and a proper abnormal value detection threshold according to different indexes and scenes, wherein a box line graph method, a Z-score method, an isolated forest method and the like can be adopted.
And finally, removing abnormal values from the mobile monitoring data, deleting or replacing the abnormal values according to the detection result, or marking and analyzing the abnormal values as special cases.
Step S13: spatial interpolation. The sub-step is mainly to perform spatial interpolation on the mobile monitoring data, fill up the data gap in the space and increase the density and continuity of the data. The specific implementation process is as follows:
first, the mobile monitoring data is analyzed in spatial distribution, and an interpolation area and an interpolation point in space are determined, for example, the space may be divided into grids according to a Geographic Information System (GIS), and a grid center is selected as the interpolation point.
And secondly, performing spatial interpolation calculation on the mobile monitoring data, and selecting a proper spatial interpolation method and parameters according to different indexes and scenes, wherein the methods such as an inverse distance weighting method, a kriging method, a radial basis function method and the like can be adopted.
Finally, the mobile monitoring data is subjected to spatial interpolation evaluation, and the interpolation accuracy and reliability are evaluated by comparing and analyzing the interpolation result and the actual situation, for example, methods such as mean square error, correlation coefficient, cross verification and the like can be adopted.
Step S14: and (5) time synchronization. The sub-step is mainly to time synchronize the mobile monitoring data, so that the data of different mobile monitoring stations at the same time can be corresponding, and the subsequent analysis and comparison are convenient. The specific implementation process is as follows:
first, the time distribution of the mobile monitoring data is analyzed, and a synchronization interval and a synchronization point in time are determined, for example, the time may be divided into equal-length time periods according to the sampling frequency and the time range of the data, and the midpoint of the time periods is selected as the synchronization point.
Secondly, calculating time synchronization of the mobile monitoring data, and selecting a proper time synchronization method and parameters according to different indexes and scenes, wherein the method can be a linear interpolation method, a moving average method, a Kalman filtering method and the like.
Finally, the mobile monitoring data is subjected to time synchronization evaluation, and the accuracy and the reliability of synchronization are evaluated by comparing and analyzing according to the synchronization result and the actual situation, for example, methods such as mean square error, correlation coefficient, cross verification and the like can be adopted.
S2, mining a time-space evolution trend of the pollutants based on the standardized monitoring data to obtain a trend analysis result;
step S21: and dividing subareas. The sub-step is mainly to divide mobile monitoring data into a plurality of sub-areas according to the motion trail of the mobile monitoring station and the position information of the mobile monitoring station, wherein each sub-area contains monitoring data of one or more mobile monitoring stations in a certain time. The specific implementation process is as follows:
firstly, track segmentation is carried out on the motion track of a mobile monitoring station, continuous track points are divided into a plurality of track segments according to a certain time interval or distance interval, and each track segment comprises one or a plurality of track points.
And secondly, carrying out track clustering on each track segment, and dividing the track segments with similar motion characteristics into a category by using a clustering algorithm, wherein each category represents a sub-region.
And finally, carrying out track merging on each sub-region, connecting track segments belonging to the same category to form a complete sub-region track, and extracting all monitoring data in the sub-region.
One possible trajectory segmentation and clustering algorithm is a Density-based spatio-temporal trajectory clustering algorithm (Density-basedpatio-TemporalTrajectoryClustering, DBSTC) that can segment and cluster trajectories according to the spatio-temporal distance between the trajectory points and a Density threshold. The specific process is as follows:
and 1, calculating the space-time distance between the track points. The method mainly comprises the steps of calculating the distance between every two track points in space and time by utilizing formulas such as Euclidean distance or Manhattan distance, and storing the distance as a distance matrix.
And 2, determining a density threshold value. The method mainly comprises the steps of determining a proper density threshold according to the distribution characteristics of data, and judging whether track points belong to the same cluster or not. In general, methods based on information entropy or based on kernel density estimation, etc., may be employed to automatically determine the optimal density threshold.
And 3, carrying out track segmentation and clustering. The method mainly uses the concept of density connectivity to divide track points which are less than a density threshold and are continuous in time into track segments, and divide track segments with similar motion characteristics into track clusters. In particular, algorithms based on DBSCAN or OPTICS, etc., may be employed to implement track segmentation and clustering.
The method can effectively process the track data with different lengths, different shapes and different densities; an appropriate density threshold may be adaptively determined; the spatiotemporal and motion characteristics of the trajectory data may be preserved. The method solves the problem of dividing the mobile monitoring data into a plurality of sub-areas according to the motion trail of the mobile monitoring stations, wherein each sub-area contains monitoring data of one or more mobile monitoring stations in a certain time.
Step S22: trend analysis. The substep is mainly to perform trend analysis on the monitoring data in each subarea, and calculate the change rate and variance of the pollutant concentration along with time as pollution trend indexes. The specific implementation process is as follows:
firstly, carrying out time series analysis on the monitoring data in each sub-area, fitting the relation between the pollutant concentration and time by using a time series model, and calculating the change rate of the pollutant concentration along with time as one of pollution trend indexes.
And secondly, performing analysis of variance on the monitoring data in each sub-area, calculating the fluctuation degree of the pollutant concentration in time by using an analysis of variance model, and calculating the variance of the pollutant concentration in time as one of pollution trend indexes.
And finally, comprehensively evaluating the monitoring data in each subarea, and carrying out weighted summation on the change rate and variance of the pollutant concentration along with time by utilizing a comprehensive evaluation model to obtain the comprehensive value of the pollution trend index.
In one embodiment, the above is implemented using a time series model, ARMA, which is an auto regressive moving average model that predicts future data from historical data and calculates the prediction error as a rate of change.
Step 1: the stationarity and white noise performance were checked. This step is mainly to check whether the time series meets the basic assumption of the ARMA model, i.e. stationarity and white noise. Generally, whether the time sequence is stable or not can be judged by adopting methods such as unit root test or stability test; an autocorrelation function or a partial autocorrelation function may be used to determine whether the time series is white noise. If the time series does not meet the stationarity or white noise, a differential or transform method may be used to make it meet the ARMA model assumption.
Step 2: the model order is determined. This step is mainly to determine the autoregressive order p and the moving average order q of the ARMA model, i.e. the ARMA (p, q) model. Generally, information criteria or residual analysis methods may be used to select the optimal model order. The information criterion is to evaluate the quality of the model according to the fitting degree and complexity of the model, and the common information criterion is AIC, BIC, HQ; the residual analysis is to evaluate the rationality of the model according to whether the residual of the model is white noise, and the common residual analysis includes Ljung-Box test, normal test and the like.
Step 3: estimating model parameters and predicting future data. The method mainly comprises the steps of estimating parameters of an ARMA model by using methods such as maximum likelihood estimation or least square estimation, predicting future data by using the estimated parameters and historical data, and calculating a prediction error as a change rate.
The method aims at the problems existing in the prior art: how to perform time series analysis on the monitoring data in each sub-area, fitting the relationship between the pollutant concentration and time by using a time series model, and calculating the change rate of the pollutant concentration along with time. The method can effectively describe the dynamic characteristics and the random characteristics of the time sequence; future data can be predicted by using the historical data; the trend of the time series can be reflected by the prediction error.
Another possible embodiment is realized by constructing the analysis of variance model as a generalized additive model. A generalized additive model GAM (GeneralizedAdditiveModel) that fits the relationship of contaminant concentration to time according to a nonlinear function and calculates the sum of squares of residuals as the variance.
The specific process is as follows:
and 1, selecting a nonlinear function. The step mainly comprises selecting proper nonlinear function as the basis function of GAM model, such as spline function, polynomial function, logarithmic function, etc., according to the distribution characteristics and change rule of data.
And 2, estimating model parameters. The method mainly utilizes methods such as maximum likelihood estimation or least square estimation to estimate parameters of the GAM model, including coefficients of a basis function, smoothing parameters and the like.
And 3, calculating a residual square sum. The method mainly comprises the steps of fitting pollutant concentration by using a GAM model, calculating residual errors between a fitting value and an actual value, and summing squares of the residual errors to obtain an estimated value of variance.
The method and the device can effectively process nonlinear and non-stable data, can utilize different types of basis functions to perform flexible fitting, and can reflect fluctuation degree of the data through residual errors. The embodiment aims at how to perform analysis of variance on the monitoring data in each subarea in the prior art, calculates the fluctuation degree of the pollutant concentration in time by using an analysis of variance model, and calculates the variance of the pollutant concentration in time.
One possible comprehensive evaluation model is analytic hierarchy process (AnalyticHierarchyProcess, AHP), which can determine the rate of change and the weight of the variance based on expert scoring and consistency testing, and perform weighted summation.
And 1, constructing a hierarchical structure. The method mainly comprises the step of decomposing the comprehensive evaluation problem into a plurality of layers, including a target layer, a criterion layer and a scheme layer. The target layer is the total target of the comprehensive evaluation problem, namely pollution trend index, the criterion layer is the factor influencing the comprehensive evaluation result, namely the change rate and the variance, and the scheme layer is the object needing to carry out comprehensive evaluation comparison, namely each subarea.
And 2, constructing a judgment matrix. The step is mainly to construct a judgment matrix according to expert scoring or objective data to reflect the relative importance or priority among different layers or among elements in the same layer. Generally, a 1-9 scale or other quantization method may be used to represent the element values in the decision matrix.
And 3, calculating weight vectors and consistency ratios. The step mainly comprises the steps of calculating the maximum eigenvalue of the judgment matrix and the corresponding eigenvector by using an eigenvalue method or other methods, and normalizing the eigenvector to be used as a weight vector. Meanwhile, the consistency index or other methods are utilized to calculate and judge the consistency proportion of the matrix, and whether the consistency requirement is met is checked. If the consistency requirement is not met, the judgment matrix can be corrected until the consistency requirement is met.
And 4, carrying out weighted summation and sequencing. The method mainly comprises the steps of carrying out weighted summation by utilizing a weight vector and the change rate and variance of each subarea to obtain the comprehensive value of the pollution trend index of each subarea, and sequencing according to the comprehensive value to obtain the comprehensive evaluation result.
The method has the advantages that the comprehensive evaluation problems of multiple criteria and multiple schemes can be effectively processed, expert knowledge and objective data can be fully utilized, and the rationality and the credibility of the evaluation result can be ensured through consistency test.
The method can solve the technical problems of comprehensively evaluating the monitoring data in each subarea, and weighting and summing the change rate and variance of the pollutant concentration along with time by utilizing a comprehensive evaluation model to obtain the comprehensive value of the pollution trend index.
Step S23: and (5) regional screening. The sub-step is mainly to rank and screen each sub-region according to pollution trend indexes, select the sub-region which is ranked at the front and exceeds a threshold value as a candidate pollution source region, and estimate the position and the intensity of the pollution source. The specific implementation process is as follows:
firstly, the pollution trend indexes of each sub-region are ranked according to the comprehensive value from high to low, and a ranking list of the sub-regions is obtained.
And secondly, screening pollution trend indexes of each sub-region, and selecting sub-regions with comprehensive values exceeding the threshold value or ranked in the first few bits as candidate pollution source regions according to the threshold value or the preset proportion.
And finally, estimating each candidate pollution source region, and estimating the pollution source position and intensity by using a maximum likelihood method or a least square method and the like according to the sub-region track and the monitoring data.
One possible method for determining the threshold or the proportion is a method based on information entropy, which can calculate the information entropy of the pollution trend index according to the distribution characteristics of the pollution trend index, and determine the optimal threshold or the proportion according to the information entropy.
And 1, calculating information entropy. The method mainly comprises the steps of calculating the information entropy of the pollution trend index by using shannon information entropy or other information entropy formulas, and reflecting the uncertainty or confusion degree of the distribution of the information entropy. Generally, the larger the information entropy, the more uniform or random the distribution, and the smaller the information entropy, the more concentrated or ordered the distribution.
And 2, determining an optimal threshold value or proportion. The method mainly comprises the steps of determining an optimal threshold value or proportion according to the change rule of the information entropy, so that the information entropy reaches the maximum or minimum. In general, methods such as maximum likelihood estimation or least squares estimation can be used to solve for the optimal threshold or scale.
And 3, carrying out regional screening. The method mainly comprises the steps of screening sub-areas according to an optimal threshold value or proportion, and selecting sub-areas with comprehensive values exceeding the threshold value or ranked in the first few bits as candidate pollution source areas. The embodiment can effectively process data of different types and distributions; objective and scientific decisions can be made by utilizing the principle of information theory, and the distribution characteristics and change rules of the data can be reflected by information entropy. The method and the device can solve the problem of how to rank and screen each sub-region according to pollution trend indexes and select the sub-region which is ranked at the front and exceeds the threshold value as the candidate pollution source region in the prior art.
S3, reading multi-source pollutant data, constructing a pollution source correction module, taking multi-source pollutants as input, obtaining a predicted result of the pollutants, comparing the predicted result with a simulation result, and correcting the simulation result;
step S31: and (5) diffusion simulation. The sub-step mainly uses CAMX-OSAT+PSAT mode, selects proper diffusion model and parameters according to different types of pollution sources such as point source, surface source, line source and the like, and simulates and predicts the candidate pollution source area to obtain the pollutant concentration distribution in space. The specific implementation process is as follows:
first, input parameters of the CAMX-osat+psat mode, such as the discharge source position, the discharge amount, the discharge height, the discharge rate, etc., are set according to information of the position, the intensity, the type, etc., of the candidate pollution source area.
Next, the CAMX-osat+psat mode was run, using euler chemistry model (eulerian chemistry-TransportModel, CTM) to simulate the atmospheric pollutant transport and chemical conversion process, and using OSAT (ozone source distribution technology) and PSAT (particulate source distribution technology) tools to resolve the sources of ozone and particulate, and calculate the contribution of each source to the pollutant concentration.
And finally, outputting the result of the CAMX-OSAT+PSAT mode to obtain a pollutant concentration distribution diagram of the candidate pollutant source region in space and a contribution diagram of each source to the pollutant concentration.
Step S32: and (5) environmental correction. The sub-step mainly uses CAMX-OSAT+PSAT mode to correct and revise the diffusion model and parameters according to atmospheric environmental factors such as wind speed, wind direction, temperature, humidity, air pressure, etc., to obtain the pollutant concentration change in time. The specific implementation process is as follows:
firstly, data of atmospheric environmental factors such as wind speed, wind direction, temperature, humidity, air pressure and the like can be obtained from sources such as weather station data, remote sensing image data, geographic information data and the like.
And secondly, correcting and correcting the diffusion model and parameters according to the data of the atmospheric environment factors by utilizing a CAMX-OSAT+PSAT mode, and considering the influence of the atmospheric environment factors on the pollutant transmission and diffusion.
And finally, outputting a result of the CAMX-OSAT+PSAT mode to obtain a pollutant concentration change graph of the candidate pollutant source region in time and an influence graph of atmospheric environmental factors on the pollutant concentration change.
Step S33: and (5) information fusion. The sub-step mainly uses information of other sources, such as fixed monitoring station data, remote sensing image data, geographic information data, meteorological information data and the like to compare and evaluate simulation and prediction results, calculates similarity and error between the simulation and prediction results and actual conditions, sorts and screens candidate pollution source areas according to the similarity and error, selects an optimal pollution source area, and determines information of positions, strength, types and the like of the pollution source areas. The specific implementation process is as follows:
first, information from other sources, such as fixed monitoring station data, remote sensing image data, geographic information data, weather information data, etc., may be obtained from published databases, websites, reports, etc.
And secondly, comparing and evaluating simulation and prediction results by using a multisource information fusion method, and calculating the similarity and error between the simulation and prediction results and actual conditions, wherein methods such as mean square error, correlation coefficient, cross verification and the like can be adopted.
Finally, the candidate pollution source regions are sequenced and screened according to the similarity and the error by utilizing a multi-source information fusion method, the optimal pollution source region is selected, and the information such as the position, the intensity and the type of the pollution source region is determined, for example, a hierarchical analysis method, a fuzzy comprehensive evaluation method, a gray correlation analysis method and the like can be adopted.
And S4, constructing a graph neural network model, and coding and training based on the corrected simulation result to obtain traceable data, wherein the traceable data comprises pollution source identification, contribution of each pollution source, a pollution propagation path and a sensitivity analysis result.
Step S41: the graph neural network encodes and embeds. The substep is mainly to encode and embed the graph data by utilizing a graph neural network, learn the implicit characteristics of nodes and edges, and the relationship and similarity between the nodes. The neural network may employ different architectures and parameters, such as GraphSAGE, GCN, GAT, etc. The specific implementation process is as follows:
firstly, preprocessing graph data, and normalizing or normalizing the attributes of nodes and edges to eliminate the dimension and deviation of the data and improve the comparability and stability of the data. The node data includes: contaminant concentration, monitoring time, monitoring location coordinates, and type of source of contamination. The side data includes contaminant propagation direction, propagation velocity, and propagation distance.
And selecting proper graphic neural network architecture and parameters, selecting the most proper graphic neural network model such as GraphSAGE, GCN, GAT according to different analysis targets and data characteristics, and setting parameters such as the layer number, the activation function, the loss function, the optimizer and the like of the model. The graph structure information of the network architecture includes an adjacency matrix, a degree matrix, and a laplace matrix.
Finally, the graph data is encoded and embedded by utilizing the graph neural network, the attributes of the nodes and the edges are converted into low-dimensional vector representations, namely embedded vectors, through multi-layer information propagation and aggregation, and the embedded vectors are stored to be used as input for subsequent analysis.
Step S42: the graph neural network decodes and predicts. The sub-step mainly utilizes the graph neural network to decode and predict the graph data, so as to realize different analysis targets, such as pollution source identification, traceability contribution analysis, sensitivity analysis and the like. The graph neural network can employ different loss functions and evaluation indexes, such as MSE, MAE, R2 and the like. The specific implementation process is as follows:
first, a suitable decoder or predictor is designed according to different analysis targets, the embedded vector is taken as an input, and required analysis results such as pollution source positions, intensities, types and the like are output.
And secondly, selecting proper loss functions and evaluation indexes, selecting the most proper loss functions and evaluation indexes such as MSE, MAE, R2 and the like according to different analysis targets and output types, and calculating the performances of the model on a training set and a testing set.
And finally, decoding and predicting the graph data by utilizing the graph neural network, outputting a required analysis result, and optimizing and adjusting the model according to the loss function and the evaluation index. Interpretation is performed according to the semantic information of the graph, including pollution source areas, pollutant propagation paths and pollution trend indexes.
In the application, aiming at the existing problems of tracing the source by using only a pollution source correction module, the method mainly comprises the steps that a plurality of pollution source areas can be provided, and different types and intensities are possible, so that the pollution source areas can be effectively identified and distinguished; the propagation path of pollutants in the air can be multiple, and can be influenced by factors such as meteorological conditions, topography, buildings and the like, how to effectively track and analyze the pollutants; uncertainty and variability may exist in the source area and contaminant propagation path, how to effectively evaluate and predict them; the source area of contamination and the path of contaminant travel may be affected by different factors, how to perform effective sensitivity analysis and optimization of them.
The above problems are solved by the characteristics of the graph neural network. Classifying or clustering analysis is carried out on the graph data by utilizing a graph neural network, so that the pollution source area and type are more accurately identified; carrying out regression analysis on the graph data by utilizing a graph neural network so as to estimate the intensity and contribution rate of each pollution source region; generating or reconstructing and analyzing the graph data by utilizing a graph neural network so as to trace the propagation path and range of pollutants in the air; the sensitivity analysis is carried out on the graph data by utilizing the graph neural network, so that the response and influence of the pollution source area and the pollutant propagation path on different factors are estimated.
Combining the analog correction module with the graph neural network, the data output by the analog correction module is one of the input data of the graph neural network, and the following information can be provided: the data fusion and correction result comprises an optimal pollution source area and related information; data comparison and evaluation results, including similarity and error; environmental information, weather information. After analog correction, the data of each point is more accurate, so that more excellent input data is provided for the graph neural network.
The preferred embodiments of the present application have been described in detail above, but the present application is not limited to the specific details of the above embodiments, and various equivalent changes can be made to the technical solution of the present application within the scope of the technical concept of the present application, and all the equivalent changes belong to the protection scope of the present application.
Claims (10)
1. The air pollution tracing method based on the neural network is characterized by comprising the following steps of:
step S1, reading and preprocessing mobile monitoring data of a mobile detection station to obtain standardized mobile monitoring data, wherein the mobile monitoring data at least comprises monitored dye concentration, monitoring time and monitoring position coordinates;
s2, mining a time-space evolution trend of the pollutants based on the mobile monitoring data to obtain a trend analysis result;
s3, reading multi-source pollutant data, constructing a pollution source correction module, taking multi-source pollutants as input, obtaining a predicted result of the pollutants, comparing the predicted result with a simulation result, and correcting the simulation result;
and S4, constructing a graph neural network model, and coding and training based on the corrected simulation result to obtain traceable data, wherein the traceable data comprises pollution source identification, contribution of each pollution source, a pollution propagation path and a sensitivity analysis result.
2. The air pollution tracing method based on the neural network of claim 1, wherein in said step S1, the process of preprocessing the movement monitoring data of the mobile detection station comprises:
step S11, detecting and eliminating abnormal values of the mobile monitoring data, and eliminating noise and interference of the data;
step S12, setting an abnormal value judgment standard, and detecting and removing abnormal values in the mobile monitoring data;
step S13, performing spatial interpolation on the mobile monitoring data to fill in the data gaps in the space;
and S14, performing time synchronization on the mobile monitoring data so that the data of different mobile monitoring stations at the same time are corresponding.
3. The method for tracing atmospheric pollution according to claim 2, wherein the step S2 is further:
step S21, dividing mobile monitoring data into a plurality of sub-areas according to the position information of the mobile monitoring stations, wherein each sub-area contains monitoring data of at least one mobile monitoring station in a preset time, and forming a time sequence of the monitoring data;
step S22, carrying out trend analysis on the monitoring data in each subarea based on a time sequence analysis method, and calculating the change rate and variance of the pollutant concentration along with time to be used as pollution trend indexes;
and S23, ranking and screening each sub-region according to the pollution trend index, selecting the sub-region which is ranked at the front and exceeds the threshold value as a candidate pollution source region, estimating the position, the type and the intensity of the pollution source, and outputting pollution source information and a trend analysis result.
4. The method for tracing atmospheric pollution according to claim 3, wherein said step S3 further comprises:
step S31, acquiring data of multi-source pollutants, including fixed monitoring station data, remote sensing image data, geographic information data and meteorological information data;
s32, constructing a pollution source correction module, wherein the pollution source correction module comprises a CAMX-OSAT-PSAT module; performing environmental correction on the pollution source correction module based on pre-stored environmental information;
and S33, taking multi-source pollutants as input, obtaining a predicted result of the pollutants, comparing the predicted result with a simulation result, comparing and evaluating the simulation result and the predicted result, calculating the similarity and error between the simulated result and the actual situation, sequencing and screening candidate pollution source areas according to the similarity and error, selecting an optimal pollution source area, and determining the position, the intensity and the type of the pollution source area.
5. The method for tracing atmospheric pollution based on a neural network according to claim 4, wherein said step S4 further comprises:
step S41, coding and embedding graph data by utilizing a graph neural network, and learning hidden characteristics of nodes and edges, and relations and similarities among the nodes;
and step S42, decoding and predicting the graph data by utilizing the graph neural network to realize different analysis targets.
6. The method for tracing atmospheric pollution according to claim 3, wherein the step S22 is further:
s22a, constructing an autoregressive moving average model, and judging whether the time sequence is stable or not by adopting a unit root test or a stationarity test; judging whether the time sequence is white noise or not by adopting an autocorrelation function or a partial autocorrelation function; if the time sequence does not meet the stationarity or white noise property, adopting a differential method to enable the time sequence to meet the assumption of an ARMA model;
step S22b, selecting an optimal model order by adopting an information criterion method or a residual analysis method, wherein the information criterion method comprises AIC, BIC and HQ; the residual analysis method comprises Ljung-Box test and normalization test;
step S22c, estimating parameters of the ARMA model by using a maximum likelihood estimation method or a least square estimation method, predicting future values of the monitoring data by using the estimated parameters and historical data, and calculating a prediction error as a change rate and variance to be used as a pollution trend index.
7. The method for tracing atmospheric pollution according to claim 3, wherein the step S23 is further:
step S23a, ranking according to the comprehensive value from high to low, and sequencing pollution trend indexes of each sub-area to obtain a ranking list of the sub-areas;
step S23b, screening pollution trend indexes of each sub-area, and selecting sub-areas with comprehensive values exceeding a threshold value or ranked in the top N bits as candidate pollution source areas according to a preset threshold value or proportion; n is a natural number;
and step S23c, estimating each candidate pollution source region, estimating the position, the type and the intensity of the pollution source by using a maximum likelihood method or a least square method according to the sub-region track and the monitoring data, and outputting pollution source information and a trend analysis result.
8. The method for tracing atmospheric pollution based on a neural network of claim 7, wherein step S23b further comprises:
calculating the information entropy of the pollution trend index by using the shannon information entropy;
according to the change rule of the information entropy, determining an optimal threshold value so that the information entropy reaches the maximum or minimum;
and screening the sub-regions according to the optimal threshold or proportion, and selecting the sub-regions with the comprehensive value exceeding the threshold or ranked in the top N bits as candidate pollution source regions.
9. The method for tracing atmospheric pollution according to claim 3, wherein step S22 further comprises:
s22a, constructing a nonlinear function set and a generalized additive model, analyzing distribution characteristics and change rules of pollutant monitoring data according to the mapping relation of pollutant data and nonlinear functions, and selecting a corresponding nonlinear function as a basis function of the generalized additive model;
step S22b, estimating parameters of a generalized additive model by using maximum likelihood estimation, wherein the parameters comprise coefficients of a basis function and smoothing parameters;
step S22c, fitting the pollutant concentration by using a generalized additive model, calculating residual errors between the fitting value and the actual value, and summing squares of the residual errors to obtain an estimated value of variance; and calculating the change rate of the pollutant concentration along with time, and taking the change rate and the variance as pollution trend indexes.
10. An atmospheric pollution traceability system based on a neural network, comprising:
at least one processor; and a memory communicatively coupled to at least one of the processors; the memory stores instructions executable by the processor, the instructions for being executed by the processor to implement the neural network-based atmospheric pollution tracing method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311126298.6A CN116881671B (en) | 2023-09-04 | 2023-09-04 | Atmospheric pollution tracing method and system based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311126298.6A CN116881671B (en) | 2023-09-04 | 2023-09-04 | Atmospheric pollution tracing method and system based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116881671A true CN116881671A (en) | 2023-10-13 |
| CN116881671B CN116881671B (en) | 2023-11-17 |
Family
ID=88262420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311126298.6A Active CN116881671B (en) | 2023-09-04 | 2023-09-04 | Atmospheric pollution tracing method and system based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116881671B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117074627A (en) * | 2023-10-16 | 2023-11-17 | 三科智能(山东)集团有限公司 | Medical laboratory air quality monitoring system based on artificial intelligence |
| CN117373556A (en) * | 2023-12-04 | 2024-01-09 | 山东清控生态环境产业发展有限公司 | Tracing instrument and system based on multidimensional data |
| CN117473398A (en) * | 2023-12-26 | 2024-01-30 | 四川国蓝中天环境科技集团有限公司 | Urban dust pollution source classification method based on slag transport vehicle activity |
| CN117689399A (en) * | 2024-01-31 | 2024-03-12 | 四川省生态环境科学研究院 | Surface water pollution source tracing method and system for monitoring abnormal increase of index |
| CN117849512A (en) * | 2024-01-19 | 2024-04-09 | 深圳市轩宇软件开发有限公司 | Motorcycle electrical system fault detection system |
| CN117972397A (en) * | 2024-03-28 | 2024-05-03 | 四川省地质矿产勘查开发局成都综合岩矿测试中心（国土资源部成都矿产资源监督检测中心） | Atmospheric dry-wet sedimentation model simulation optimization monitoring method and medium based on big data |
| CN118070234A (en) * | 2024-04-22 | 2024-05-24 | 河北工程大学 | Water quality state fusion sensing and prediction tracing method |
| CN118169339A (en) * | 2024-05-16 | 2024-06-11 | 北京市生态环境监测中心 | Surrounding pollution source tracing method, device, medium and equipment based on multi-source data |
| CN118566442A (en) * | 2024-07-30 | 2024-08-30 | 洛阳敦鑫建筑工程有限公司 | Atmospheric environment pollutant concentration detection method and system |
| CN118656768A (en) * | 2024-08-21 | 2024-09-17 | 江苏电力信息技术有限公司 | Atmospheric pollutant emission data processing method based on time sequence diagram neural network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108733893A (en) * | 2018-04-25 | 2018-11-02 | 同济大学 | The public building burst pollution of coupling depth learning method is traced to the source |
| KR102256243B1 (en) * | 2020-12-21 | 2021-05-27 | 한국건설기술연구원 | Deep learning-based system for time-dependent assessment of concrete deterioration |
| CN114860816A (en) * | 2022-07-11 | 2022-08-05 | 广东盈峰科技有限公司 | Water pollution intelligent early warning system integration method, device and related equipment |
| CN115269675A (en) * | 2022-06-15 | 2022-11-01 | 浙江工业大学 | Time-space analysis method for atmospheric pollution in industrial park |
| CN115358332A (en) * | 2022-08-25 | 2022-11-18 | 浙江工业大学 | Atmospheric pollution tracing method for multi-source data |
| US20230194755A1 (en) * | 2021-12-15 | 2023-06-22 | Zhejiang University Of Technology | Data-driven rapid traceability method for air pollutants in small-scale regionals |
-
2023
- 2023-09-04 CN CN202311126298.6A patent/CN116881671B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108733893A (en) * | 2018-04-25 | 2018-11-02 | 同济大学 | The public building burst pollution of coupling depth learning method is traced to the source |
| KR102256243B1 (en) * | 2020-12-21 | 2021-05-27 | 한국건설기술연구원 | Deep learning-based system for time-dependent assessment of concrete deterioration |
| US20230194755A1 (en) * | 2021-12-15 | 2023-06-22 | Zhejiang University Of Technology | Data-driven rapid traceability method for air pollutants in small-scale regionals |
| CN115269675A (en) * | 2022-06-15 | 2022-11-01 | 浙江工业大学 | Time-space analysis method for atmospheric pollution in industrial park |
| CN114860816A (en) * | 2022-07-11 | 2022-08-05 | 广东盈峰科技有限公司 | Water pollution intelligent early warning system integration method, device and related equipment |
| CN115358332A (en) * | 2022-08-25 | 2022-11-18 | 浙江工业大学 | Atmospheric pollution tracing method for multi-source data |
Non-Patent Citations (1)
| Title |
|---|
| 程麟钧;: "我国大气环境监测数据共享技术现状、问题及对策", 中国环境监测, no. 06 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117074627B (en) * | 2023-10-16 | 2024-01-09 | 三科智能(山东)集团有限公司 | Medical laboratory air quality monitoring system based on artificial intelligence |
| CN117074627A (en) * | 2023-10-16 | 2023-11-17 | 三科智能(山东)集团有限公司 | Medical laboratory air quality monitoring system based on artificial intelligence |
| CN117373556A (en) * | 2023-12-04 | 2024-01-09 | 山东清控生态环境产业发展有限公司 | Tracing instrument and system based on multidimensional data |
| CN117373556B (en) * | 2023-12-04 | 2024-02-13 | 山东清控生态环境产业发展有限公司 | Tracing instrument and system based on multidimensional data |
| CN117473398A (en) * | 2023-12-26 | 2024-01-30 | 四川国蓝中天环境科技集团有限公司 | Urban dust pollution source classification method based on slag transport vehicle activity |
| CN117473398B (en) * | 2023-12-26 | 2024-03-19 | 四川国蓝中天环境科技集团有限公司 | Urban dust pollution source classification method based on slag transport vehicle activity |
| CN117849512A (en) * | 2024-01-19 | 2024-04-09 | 深圳市轩宇软件开发有限公司 | Motorcycle electrical system fault detection system |
| CN117689399B (en) * | 2024-01-31 | 2024-04-12 | 四川省生态环境科学研究院 | Surface water pollution source tracing method and system for monitoring abnormal increase of index |
| CN117689399A (en) * | 2024-01-31 | 2024-03-12 | 四川省生态环境科学研究院 | Surface water pollution source tracing method and system for monitoring abnormal increase of index |
| CN117972397A (en) * | 2024-03-28 | 2024-05-03 | 四川省地质矿产勘查开发局成都综合岩矿测试中心（国土资源部成都矿产资源监督检测中心） | Atmospheric dry-wet sedimentation model simulation optimization monitoring method and medium based on big data |
| CN117972397B (en) * | 2024-03-28 | 2024-05-31 | 四川省地质矿产勘查开发局成都综合岩矿测试中心（国土资源部成都矿产资源监督检测中心） | Atmospheric dry-wet sedimentation model simulation optimization monitoring method and medium based on big data |
| CN118070234A (en) * | 2024-04-22 | 2024-05-24 | 河北工程大学 | Water quality state fusion sensing and prediction tracing method |
| CN118169339A (en) * | 2024-05-16 | 2024-06-11 | 北京市生态环境监测中心 | Surrounding pollution source tracing method, device, medium and equipment based on multi-source data |
| CN118169339B (en) * | 2024-05-16 | 2024-07-23 | 北京市生态环境监测中心 | Surrounding pollution source tracing method, device, medium and equipment based on multi-source data |
| CN118566442A (en) * | 2024-07-30 | 2024-08-30 | 洛阳敦鑫建筑工程有限公司 | Atmospheric environment pollutant concentration detection method and system |
| CN118566442B (en) * | 2024-07-30 | 2024-10-25 | 洛阳敦鑫建筑工程有限公司 | Atmospheric environment pollutant concentration detection method and system |
| CN118656768A (en) * | 2024-08-21 | 2024-09-17 | 江苏电力信息技术有限公司 | Atmospheric pollutant emission data processing method based on time sequence diagram neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116881671B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116881671B (en) | Atmospheric pollution tracing method and system based on neural network | |
| Farmani et al. | Pipe failure prediction in water distribution systems considering static and dynamic factors | |
| CN117116382B (en) | Water quality space-time prediction method and system for water-bearing lake under influence of diversion engineering | |
| CN110738355A (en) | urban waterlogging prediction method based on neural network | |
| CN113496314B (en) | Method for predicting road traffic flow by neural network model | |
| CN105488317A (en) | Air quality prediction system and method | |
| CN118245822B (en) | Similarity set forecast optimization method, device, equipment and medium | |
| CN113836808A (en) | PM2.5 deep learning prediction method based on heavy pollution feature constraint | |
| CN118095112B (en) | Marine floating facility environment influence analysis and quantification method | |
| CN117371571A (en) | Regional air quality prediction model based on multi-scale dynamic synchronous diagram mechanism | |
| CN117556197A (en) | Typhoon vortex initialization method based on artificial intelligence | |
| CN117725448A (en) | Cluster analysis method for meteorological navigation signal characteristics | |
| CN117933497B (en) | TSA-ARIMA-CNN-based enterprise carbon emission prediction method | |
| CN113688506B (en) | Potential atmospheric pollution source identification method based on multi-dimensional data such as micro-station and the like | |
| CN112734129B (en) | Air pollution space-time trend prediction method based on unsupervised restrictive optimization | |
| CN117370813A (en) | Atmospheric pollution deep learning prediction method based on K line pattern matching algorithm | |
| CN113965618B (en) | Abnormal track detection method based on fuzzy theory | |
| CN115062708A (en) | Abnormal node detection method based on track deviation point embedding and depth clustering | |
| Hong et al. | Drainage network flow anomaly classification based on XGBoost | |
| CN118072873B (en) | Air quality prediction method | |
| CN117271959B (en) | Uncertainty evaluation method and equipment for PM2.5 concentration prediction result | |
| Choudhary et al. | Does PM 10 influence the prediction of PM 2.5? | |
| CN118427536B (en) | Smart city-oriented data index management method and system | |
| KR20240050754A (en) | Method for estimating concentration of particulate matters using ai model | |
| CN118296568A (en) | Interpretable track prediction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20231013 Assignee: Shandong Heneng Technology Co.,Ltd. Assignor: Shandong Zhiming Environmental Protection Engineering Co.,Ltd. Contract record no.: X2023370000026 Denomination of invention: A Neural network-based Method and System for Tracing Air Pollution Sources Granted publication date: 20231117 License type: Common License Record date: 20231229 |