CN114023399A - Air particulate matter analysis early warning method and device based on artificial intelligence - Google Patents

Air particulate matter analysis early warning method and device based on artificial intelligence Download PDF

Info

Publication number
CN114023399A
CN114023399A CN202111344877.9A CN202111344877A CN114023399A CN 114023399 A CN114023399 A CN 114023399A CN 202111344877 A CN202111344877 A CN 202111344877A CN 114023399 A CN114023399 A CN 114023399A
Authority
CN
China
Prior art keywords
particulate matter
air particulate
detection model
data
anomaly detection
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.)
Pending
Application number
CN202111344877.9A
Other languages
Chinese (zh)
Inventor
刘畅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Chengan Technology Group Co ltd
Original Assignee
Shanghai Chengan Technology Group Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shanghai Chengan Technology Group Co ltd filed Critical Shanghai Chengan Technology Group Co ltd
Priority to CN202111344877.9A priority Critical patent/CN114023399A/en
Publication of CN114023399A publication Critical patent/CN114023399A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • 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
    • G06N3/045—Combinations of networks
    • 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
    • G06N3/047—Probabilistic or stochastic networks
    • 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
    • G06N3/048—Activation functions
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/02—Alarms for ensuring the safety of persons
    • G08B21/12—Alarms for ensuring the safety of persons responsive to undesired emission of substances, e.g. pollution alarms
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G08B21/24—Reminder alarms, e.g. anti-loss alarms
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/20—Identification of molecular entities, parts thereof or of chemical compositions

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Human Resources & Organizations (AREA)
  • Tourism & Hospitality (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Emergency Management (AREA)
  • Chemical & Material Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Marketing (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Environmental & Geological Engineering (AREA)
  • Toxicology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Educational Administration (AREA)
  • Primary Health Care (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Game Theory and Decision Science (AREA)

Abstract

The invention discloses an air particulate matter analysis early warning method and device based on artificial intelligence. The method comprises the following steps: acquiring air particulate data of a target station; preprocessing air particulate matter data and establishing an air particulate matter data set; establishing an anomaly detection model based on a gated cycle unit neural network of an attention mechanism, and training the anomaly detection model according to an air particulate matter data set to obtain a trained anomaly detection model; determining the predicted value of each air particulate matter through the trained anomaly detection model, and calculating the difference value between the predicted value and the actual value of each air particulate matter; and judging whether the air particles are abnormal according to the difference, and if so, giving an early warning prompt. The technical scheme of this embodiment can effectively utilize monitoring data, and automatic identification leads to the reason of atmospheric pollution incident to report an emergency and ask for help or increased vigilance to the very first time deals with, improves the efficiency of dealing with, provides the treatment mode that reduces the manpower, increases efficiency for atmospheric environment's optimization.

Description

Air particulate matter analysis early warning method and device based on artificial intelligence
Technical Field
The embodiment of the invention relates to the technical field of air particulate matter analysis and early warning, in particular to an air particulate matter analysis and early warning method and device based on artificial intelligence.
Background
With the continuous promotion of the urbanization process and the steady improvement of the economic development level in China, urban environmental problems such as air, water and soil pollution are increasingly prominent, people pay more and more attention to the environmental problems, the requirements on the environmental quality are higher and more, and the environmental problems cannot be met by simple manpower monitoring, management and control and management. With the development and the rise of the artificial intelligence technology, the exploration research and the innovative application of the exploration research artificial intelligence technology in the field of environmental management become a new development trend in the field of environmental protection based on the application of information technologies such as the internet of things and big data, and the method has important significance for monitoring and evaluation of regional pollution conditions, large-area joint defense joint control, disposal of environmental pollution events and the like.
At present stage, mainly carry out environmental pollution's detection and management through modes such as digital environmental protection and wisdom environmental protection, but the ageing and intelligent level of above-mentioned two kinds of modes remain to be promoted, also can't be to the reason that causes environmental pollution carrying out the analysis, and this patent is consequently come.
Disclosure of Invention
The invention provides an air particulate matter analysis early warning method and device based on artificial intelligence, which can realize full coverage of pollution monitoring management on the basis of effectively monitoring the atmospheric environmental pollution condition by a solution formed by combining the Internet, the artificial intelligence technology and environmental informatization, and promote the high-efficiency treatment and decision-making scientification of work such as environmental quality supervision, pollution prevention and control, ecological environment protection and the like.
In a first aspect, an embodiment of the present invention provides an air particulate matter analysis and early warning method based on artificial intelligence, including:
acquiring air particulate data of a target station;
preprocessing the air particulate matter data, and establishing an air particulate matter data set;
establishing an anomaly detection model based on a gated cycle unit neural network of an attention mechanism, and training the anomaly detection model according to the air particulate matter data set to obtain a trained anomaly detection model;
and determining a difference value between the predicted value and the actual value of each air particulate matter through the trained anomaly detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt.
In a second aspect, an embodiment of the present invention further provides an air particulate matter analysis and early warning device based on artificial intelligence, including:
the data acquisition module is used for acquiring air particulate matter data of a target station;
the data set establishing module is used for preprocessing the air particulate matter data and establishing an air particulate matter data set;
the model training module is used for establishing an anomaly detection model based on a neural network of a gated circulation unit of an attention mechanism and training the anomaly detection model according to the air particulate matter data set to obtain a trained anomaly detection model;
and the abnormity early warning module is used for determining a difference value between the predicted value and the actual value of each air particulate matter through the trained abnormity detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt.
The embodiment of the invention establishes a multivariable input abnormity detection model by using a gate control circulation unit neural network of an attention mechanism, provides the deviation degree data of the real-time air particle pollutant concentration data relative to the estimated value, and judges whether the alarm is needed or not by combining the fluctuation rule of historical time-sharing data. And classifying the pollution reasons by using the data triggering the alarm through a full-connection neural network to obtain the speculative values of the pollution reasons (such as different pollution types of motor vehicle emission, road dust, catering oil fume, burning smoke dust and the like) so as to respond at the first time, improve the disposal efficiency and provide a management mode of reducing personnel and improving efficiency for the optimization of the atmospheric environment.
Drawings
Fig. 1 is a flowchart of an air particulate analysis early warning based on artificial intelligence according to an embodiment of the present invention;
FIG. 2 is a general block diagram of an artificial intelligence air particle analysis early warning provided by an embodiment of the present invention;
FIG. 3 is a graph fitting the probability distribution of the difference values provided by the embodiment of the present invention;
fig. 4 is a schematic diagram for determining a predicted value of a target station according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Examples
In this embodiment, the particle pollutant data (including PM) acquired by the ground air pollution monitoring point every 15 minutes may be acquired0.3、PM0.5、PM1、PM2.5、PM5、PM10、CO、NO2、O3、SO2The data) and the temperature and humidity data, the wind speed data are subjected to preprocessing processes such as random forest filling missing values and data standardization, and then the attention-GRU model is used for training and predicting the concentration of a certain particle pollutant in the next 15 minutes.
Fig. 1 is a flowchart of an air particulate analysis early warning based on artificial intelligence according to an embodiment of the present invention, which specifically includes the following steps:
and S110, acquiring air particulate matter data of the target station.
Wherein, the air particulate matter data in the embodiment is obtained from the ground atmospheric pollution monitoring station, and can include PM0.3、PM0.5、PM1、PM2.5、PM5、PM10、CO、NO2、O3、SO2And so on. In addition, temperature and humidity data, wind speed data and the like can be obtained from a ground atmospheric pollution monitoring station.
Referring to fig. 2, fig. 2 is a general block diagram of an artificial intelligent air particulate matter analysis and early warning system provided by an embodiment of the present invention, which includes a ground atmospheric pollution monitoring station network, an online monitoring data storage server, and a data processing server. And the data processing server performs analysis early warning on the data acquired through the ground atmospheric pollution monitoring station network, and sends the analysis early warning result to a large screen, an APP and other display terminals through the online monitoring data access server for displaying.
And S120, preprocessing the air particulate matter data, and establishing an air particulate matter data set.
If the sample data has missing values, much information in the original data can be lost, and certain influence can be generated on the distribution of the data. This may lead to unreliable results for the sample data when fitting through the model. However, if the integrity and the accuracy of the information are ensured only by one step and the data with missing values are discarded, the training data may be too few to cause the under-fitting phenomenon.
Therefore, in this embodiment, the acquired air particulate matter data needs to be preprocessed, specifically, a random forest filling method is used, missing values in the data are used as target variables, and characteristic values of known other variables are substituted into a random forest fitting model to estimate missing values, so that filling is performed according to the estimated values.
In addition, in the embodiment, a minimum-maximum normalization method is used for performing data normalization processing on the air particulate matter data, and the normalization principle of the method is that a linear mapping operation is performed on sample data by using a conversion function, so that a result after the linear mapping falls within a [0,1] interval, and subsequent processing is facilitated. When the standardized sample data is used for modeling and predicting by using the neural network, the convergence rate of the model can be greatly improved, and meanwhile, the prediction precision and the learning efficiency of the model are improved.
Further, an air particulate matter data set is established based on the data after the preprocessing. In this embodiment, 80% of the whole data set is selected as the training set for training the model parameters, and the other 20% is selected as the test set for evaluating the generalization ability of the model. Considering the limited amount of data available, the K-Fold cross validation method is used in the present embodiment to evaluate the predictive performance of the model, and the validation set is not divided separately.
S130, establishing an anomaly detection model based on a neural network of a gated circulation unit of an attention mechanism, and training the anomaly detection model according to the air particulate matter data set to obtain the trained anomaly detection model.
In the anomaly detection model in this embodiment, two layers of GRUs and one layer of Attention are stacked to perform feature learning on sequence data. Two GRU networks are connected by a Dropout layer to reduce the parameter amount during training and prevent overfitting, and the Dropout layer parameter is 0.5.
Referring to the experimental results of the visual test of different numbers of neurons, the first layer GRU is set to 256 neurons and the second layer GRU is set to 64 neurons in this embodiment according to the experimental setting of the optimal prediction result. The second layer GRU outputs 64 learned features and proceeds to the next Attention layer for further learning. The Attention layer is followed by a fully connected layer, which is set according to the type of air particles to be currently predicted. If only PM is currently predicted in this embodiment10The result of the model output at this time is a one-dimensional vector, so this layer is set as a neuron. Referring to fig. 3 in detail, fig. 3 is a network structure diagram of the Attention-GRU model constructed in the present embodiment.
The GRU layer is added into the neural network model, so that historically important information can be selected and left, unimportant information can be left, and the action of adding the Attention layer is to highlight the influence of key information in an input sequence on an output result.
The loss function of the model uses a mean square error loss function, GPU hardware is used for training, the initial learning rate is reasonably adjusted, the learning rate is adjusted once every 5 epochs, an Adam algorithm is used for training on a training set, a test set is not used during training, and training is stopped when the training accuracy reaches a certain degree.
In practical applications, the pollutant concentration tends to have a certain regionality, and the change trends of the pollutant concentration values of adjacent sites along with time are similar, that is, the pollutant concentration data has spatial correlation. Therefore, in the embodiment, the data of the nearby sites is used for assisting the prediction of the target site so as to improve the model performance and generalization capability.
Specifically, if other stations exist within a certain distance from the target station, the attention-based gated loop unit neural network respectively establishes an abnormality detection model for the target station and the other stations, and the target station and the other stations are respectively trained according to the air particulate matter data sets corresponding to the target station and the other stations, so as to obtain an abnormality detection model corresponding to the trained target station and an abnormality detection model corresponding to the other stations.
And then, putting the predicted value of each air particulate matter determined by the abnormality detection model corresponding to the target station and the predicted value of each air particulate matter determined by the abnormality detection model corresponding to other stations into a convolutional neural network to obtain the final predicted value of each air particulate matter of the target station.
With reference to figure 4 for an exemplary illustration,
selecting a site within a range of 1km around a target site to be predicted, and if no other site exists within the range of 1km around the target site, not performing the step; there are 4 nearby stations around the target station in fig. 4. The above-mentioned Attention-GRU model is used for prediction respectively, and then values of a target station and nearby stations are sent into a convolutional neural network, wherein the Convolutional Neural Network (CNN) has unique advantages in extracting spatial correlation of data. Because there are few input nodes, no pooling layer is provided in the CNN to simplify the feature map, and only the convolutional layer is used to extract the spatial information of the input feature map.
Since the input data is a one-dimensional array, a 1-dimensional convolutional neural network (Conv1D) is used in this embodiment. Using three convolution layers, wherein the sizes of convolution kernels are 3 x 3, and the sizes of nonlinear activation functions are tanh; the first layer of convolutional layer is provided with 4 convolutional kernels, the second layer of convolutional layer is provided with 8 convolutional kernels, and the third layer of convolutional layer is designed to be 8 convolutional kernels. And then passing through a Flatten layer and a fully-connected layer containing 10 neurons, wherein the last fully-connected layer is used as a numerical value with the output of 1 x 1, namely the final predicted value of the target site. The Conv1D model training method in this example is substantially the same as the previous attention-GRU model.
S140, determining a difference value between the predicted value and the actual value of each air particulate matter through the trained anomaly detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt.
Proceed to predict PM10Whether there is an abnormality is exemplified by obtaining a trained abnormality detection model, and then PM10PM after 15 minutes is sequentially obtained on the data set by using a trained Attention-GRU model10The predicted concentration value pm10_ pred is calculated as a difference pm10_ residual from the actual value pm10_ test.
The probability distribution histogram data is calculated for all differences pm10_ residual, and the probability distribution for pm10_ residual is fitted with a hyperbolic distribution function, see fig. 3, in which the probability density of the value x is combined with the hyperbolic distribution
Figure BDA0003353600960000051
Is in direct proportion. Fitting the values of the parameters alpha, beta, delta and mu to obtain fPM10(x; α, β, δ, μ) for a new pm10_ residual value, a probability density value p can be calculated using the probability distribution functionPM10When the probability density value p isPM10Is less than a certain threshold p &PM10In time, the probability of the PM10_ residual value corresponding to the predicted value is smaller, the difference between the actual value and the regular predicted value is considered to be larger, and accordingly the abnormal PM is judged10And (4) data.
Further, after judging that the air particulate matter is abnormal, the method further comprises the following steps:
extracting corresponding characteristic quantities from the air particulate matter concentration data and meteorological data;
and outputting the event type of the air particulate matter abnormity through the fully-connected neural network according to the characteristic quantity.
Use of PM by the above procedure0.3、PM0.5、PM1、PM2.5、PM5Training various particle pollutant data to obtain a corresponding Attention-GRU model, and adding PM10The Attention-GRU of (g) has 6 models and fPM0.3、fPM0.5、fPM1、fPM2.5、fPM5、fPM10Probability distribution function of six differences. When PM10If it is determined to be abnormal, based on fPM0.3、fPM0.5、fPM1、fPM2.5、fPM5Calculating a probability density value pPM0.3、pPM0.5、pPM1、pPM2.5、pPM5. According to PM0.3、PM0.5、PM1、PM2.5、PM5And PM10Difference and probability density of CO, NO2、O3、SO2The pollutant concentration data, temperature TEMP, relative humidity RH and wind speed are three meteorological data, and 19 data are taken as the characteristic quantity of 1 × 19 particle pollutants. And then, the computer can analyze the pollution reason by using the extracted characteristic quantity through a machine learning method.
In this embodiment, a fully-connected neural network is used, the obtained 1 × 19 characteristic quantity is used as an input layer of the neural network, two hidden layers are then set, the number of neurons included in the hidden layers is 128, and the output layer gives the types of events causing particulate pollution, such as motor vehicle emission, road dust, windrow dust, cooking fume, burning smoke and the like, corresponding to the number of classification results.
The air particulate matter analysis early warning method based on artificial intelligence can effectively utilize monitoring data, automatically identify the reason of the air pollution event and give an alarm on the basis of the existing air monitoring station, so that government departments can deal with the air particulate matter in the first time, the disposal efficiency is improved, and a treatment mode of reducing manpower and increasing efficiency is provided for the optimization of the atmospheric environment. The embodiment of the invention also provides an air particulate matter analysis and early warning device based on artificial intelligence, which comprises:
the data acquisition module is used for acquiring air particulate matter data of a target station;
the data set establishing module is used for preprocessing the air particulate matter data and establishing an air particulate matter data set;
the model training module is used for establishing an anomaly detection model based on a neural network of a gated circulation unit of an attention mechanism and training the anomaly detection model according to the air particulate matter data set to obtain a trained anomaly detection model;
the anomaly detection model comprises two layers of GRUs and one layer of Attention, and the two layers of GRU networks are connected by a Dropout layer; wherein the first layer GRU is provided as 256 neurons and the second layer GRU is provided as 64 neurons.
And the abnormity early warning module is used for determining a difference value between the predicted value and the actual value of each air particulate matter through the trained abnormity detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt. The device also comprises an analysis module, a data processing module and a data processing module, wherein the analysis module is used for extracting corresponding characteristic quantities from the air particles and meteorological data;
and outputting the event type of the air particulate matter abnormity through the fully-connected neural network according to the characteristic quantity.
Specifically, the data set establishing module is specifically configured to: filling the air particulate matter data based on a random forest filling method;
and carrying out data standardization processing on the air particulate matter data by a minimum and maximum value normalization method.
The abnormity early warning module is also used for: calculating a probability distribution histogram of the difference values;
fitting the probability distribution histogram of the difference value by adopting a hyperbolic distribution function to obtain a probability distribution function of the difference value;
and determining a difference value between the predicted value and the actual value of the current air particulate matter, and if the probability density value corresponding to the difference value is smaller than a set threshold value in the probability distribution function, determining that the air particulate matter is abnormal.
The air particulate matter analysis and early warning device based on artificial intelligence provided by the embodiment of the invention can execute the air particulate matter analysis and early warning method based on artificial intelligence provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (8)

1. The utility model provides an air particulate matter analysis early warning method based on artificial intelligence which characterized in that includes:
acquiring air particulate data of a target station;
preprocessing the air particulate matter data, and establishing an air particulate matter data set;
establishing an anomaly detection model based on a gated cycle unit neural network of an attention mechanism, and training the anomaly detection model according to the air particulate matter data set to obtain a trained anomaly detection model;
and determining a difference value between the predicted value and the actual value of each air particulate matter through the trained anomaly detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt.
2. The method of claim 1, after determining that the air particulate matter is abnormal, further comprising:
extracting corresponding characteristic quantities from the air particles and meteorological data;
and outputting the event type of the air particulate matter abnormity through the fully-connected neural network according to the characteristic quantity.
3. The method of claim 1, wherein pre-processing the air particulate data comprises:
filling the air particulate matter data based on a random forest filling method;
and carrying out data standardization processing on the air particulate matter data by a minimum and maximum value normalization method.
4. The method of claim 1, wherein determining whether each airborne particle is abnormal based on the difference comprises:
calculating a probability distribution histogram of the difference values;
fitting the probability distribution histogram of the difference value by adopting a hyperbolic distribution function to obtain a probability distribution function of the difference value;
and determining a difference value between the predicted value and the actual value of the current air particulate matter, and if the probability density value corresponding to the difference value is smaller than a set threshold value in the probability distribution function, determining that the air particulate matter is abnormal.
5. The method of claim 1, wherein the anomaly detection model comprises two layers of GRUs and one layer of Attention, the two layers of GRU networks being connected by a Dropout layer;
wherein the first layer GRU is provided as 256 neurons and the second layer GRU is provided as 64 neurons.
6. The method of claim 1, wherein establishing an anomaly detection model based on an attention-system gated-cyclic unit neural network and training the anomaly detection model according to the air particulate matter dataset to obtain a trained anomaly detection model comprises:
if other stations exist within a certain distance from the target station, respectively establishing an abnormality detection model for the target station and the other stations based on a gated circulation unit neural network of an attention system, and respectively training the target station and the other stations according to the air particulate matter data sets corresponding to the target station and the other stations to obtain the abnormality detection model corresponding to the trained target station and the abnormality detection model corresponding to the other stations.
7. The method of claim 6, wherein determining the predicted value for each airborne particle via the trained anomaly detection model comprises:
and inputting the predicted value of each air particulate matter determined by the abnormality detection model corresponding to the target station and the predicted value of each air particulate matter determined by the abnormality detection model corresponding to other stations into a convolutional neural network to obtain the final predicted value of each air particulate matter of the target station.
8. The utility model provides an air particulate matter analysis early warning device based on artificial intelligence which characterized in that includes:
the data acquisition module is used for acquiring air particulate matter data of a target station;
the data set establishing module is used for preprocessing the air particulate matter data and establishing an air particulate matter data set;
the model training module is used for establishing an anomaly detection model based on a neural network of a gated circulation unit of an attention mechanism and training the anomaly detection model according to the air particulate matter data set to obtain a trained anomaly detection model;
and the abnormity early warning module is used for determining a difference value between the predicted value and the actual value of each air particulate matter through the trained abnormity detection model, judging whether each air particulate matter is abnormal according to the difference value, and if so, giving an early warning prompt.
CN202111344877.9A 2021-11-15 2021-11-15 Air particulate matter analysis early warning method and device based on artificial intelligence Pending CN114023399A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111344877.9A CN114023399A (en) 2021-11-15 2021-11-15 Air particulate matter analysis early warning method and device based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111344877.9A CN114023399A (en) 2021-11-15 2021-11-15 Air particulate matter analysis early warning method and device based on artificial intelligence

Publications (1)

Publication Number Publication Date
CN114023399A true CN114023399A (en) 2022-02-08

Family

ID=80063879

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111344877.9A Pending CN114023399A (en) 2021-11-15 2021-11-15 Air particulate matter analysis early warning method and device based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN114023399A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114662611A (en) * 2022-04-07 2022-06-24 中科三清科技有限公司 Method and device for restoring particulate component data, electronic equipment and storage medium
CN116182949A (en) * 2023-02-23 2023-05-30 中国人民解放军91977部队 Marine environment water quality monitoring system and method
CN116295604A (en) * 2023-01-04 2023-06-23 中铁十一局集团有限公司 Intelligent dust real-time monitoring and control system

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114662611A (en) * 2022-04-07 2022-06-24 中科三清科技有限公司 Method and device for restoring particulate component data, electronic equipment and storage medium
CN116295604A (en) * 2023-01-04 2023-06-23 中铁十一局集团有限公司 Intelligent dust real-time monitoring and control system
CN116295604B (en) * 2023-01-04 2024-02-06 中铁十一局集团有限公司 Intelligent dust real-time monitoring and control system
CN116182949A (en) * 2023-02-23 2023-05-30 中国人民解放军91977部队 Marine environment water quality monitoring system and method
CN116182949B (en) * 2023-02-23 2024-03-19 中国人民解放军91977部队 Marine environment water quality monitoring system and method

Similar Documents

Publication Publication Date Title
Janarthanan et al. A deep learning approach for prediction of air quality index in a metropolitan city
CN114023399A (en) Air particulate matter analysis early warning method and device based on artificial intelligence
CN102098180B (en) Network security situational awareness method
CN116625438B (en) Gas pipe network safety on-line monitoring system and method thereof
CN109800875A (en) Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine
CN111210128B (en) Wetland early warning method based on artificial intelligence and random self-adaptive threshold
Kosek et al. Ensemble regression model-based anomaly detection for cyber-physical intrusion detection in smart grids
CN110636066B (en) Network security threat situation assessment method based on unsupervised generative reasoning
CN112966714A (en) Edge time sequence data anomaly detection and network programmable control method
CN113408659A (en) Building energy consumption integrated analysis method based on data mining
CN115358155A (en) Power big data abnormity early warning method, device, equipment and readable storage medium
CN116383773A (en) Data center energy efficiency anomaly detection method, system and medium based on self-adaptive prediction interval
CN115470850A (en) Water quality abnormal event recognition early warning method based on pipe network water quality time-space data
CN117370919B (en) Remote monitoring system for sewage treatment equipment
CN117077870B (en) Water resource digital management method based on artificial intelligence
CN115964503B (en) Safety risk prediction method and system based on community equipment facilities
Gai et al. Research on water quality spatiotemporal forecasting model based on ST-BIGRU-SVR neural network
Li et al. Information granularity with the self-emergence mechanism for event detection in WSN-based tunnel health monitoring
CN114579962A (en) AI safety attack and defense test method
Xue et al. A Fire Detection and Assessment Method based on YOLOv8
KR20220138250A (en) A method and an electronic device for inferring occurrence of highly-concentrated fine dust
CN112032567A (en) Buried gas pipeline leakage risk prediction system
Hu et al. The early warning model of dust concentration in smart construction sites based on long short term memory network
CN117421684B (en) Abnormal data monitoring and analyzing method based on data mining and neural network
CN111832832B (en) District self-inspection system based on thing networking

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