CN111754042A - Atmospheric pollutant concentration prediction method and device based on Gaussian regression - Google Patents
Atmospheric pollutant concentration prediction method and device based on Gaussian regression Download PDFInfo
- Publication number
- CN111754042A CN111754042A CN202010601670.4A CN202010601670A CN111754042A CN 111754042 A CN111754042 A CN 111754042A CN 202010601670 A CN202010601670 A CN 202010601670A CN 111754042 A CN111754042 A CN 111754042A
- Authority
- CN
- China
- Prior art keywords
- data
- time period
- training samples
- prediction
- training
- 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
Links
- 239000003344 environmental pollutant Substances 0.000 title claims abstract description 106
- 231100000719 pollutant Toxicity 0.000 title claims abstract description 106
- 238000000034 method Methods 0.000 title claims abstract description 84
- 238000012549 training Methods 0.000 claims abstract description 105
- 230000007613 environmental effect Effects 0.000 claims abstract description 57
- 230000008569 process Effects 0.000 claims abstract description 26
- 230000006870 function Effects 0.000 claims description 55
- 239000011159 matrix material Substances 0.000 claims description 24
- RAHZWNYVWXNFOC-UHFFFAOYSA-N Sulphur dioxide Chemical compound O=S=O RAHZWNYVWXNFOC-UHFFFAOYSA-N 0.000 claims description 12
- JCXJVPUVTGWSNB-UHFFFAOYSA-N nitrogen dioxide Inorganic materials O=[N]=O JCXJVPUVTGWSNB-UHFFFAOYSA-N 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- MGWGWNFMUOTEHG-UHFFFAOYSA-N 4-(3,5-dimethylphenyl)-1,3-thiazol-2-amine Chemical compound CC1=CC(C)=CC(C=2N=C(N)SC=2)=C1 MGWGWNFMUOTEHG-UHFFFAOYSA-N 0.000 claims description 4
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 claims description 4
- CBENFWSGALASAD-UHFFFAOYSA-N Ozone Chemical compound [O-][O+]=O CBENFWSGALASAD-UHFFFAOYSA-N 0.000 claims description 4
- 229910002091 carbon monoxide Inorganic materials 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 3
- 230000000875 corresponding effect Effects 0.000 description 61
- 238000000354 decomposition reaction Methods 0.000 description 12
- 239000000356 contaminant Substances 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 230000001932 seasonal effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000003915 air pollution Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000009022 nonlinear effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000005295 random walk Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The application provides a method and a device for predicting concentration of an atmospheric pollutant based on Gaussian regression. The method comprises the following steps: acquiring first environment data within a first preset historical time period from the current moment; the environmental data comprises a plurality of pollutant concentration data and meteorological data; obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period; training the Gaussian process regression model by using a plurality of training samples to obtain a prediction model; and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model. The method and the device can improve the accuracy of prediction of the concentration of the atmospheric pollutants in the future time period.
Description
Technical Field
The application relates to the technical field of atmospheric detection, in particular to a method and a device for predicting concentration of atmospheric pollutants based on Gaussian regression.
Background
In recent years, with the increasing of the social and economic level, the emission of pollutants produced and living by people is increasing, the influence on the environment is increasing, and the air pollution is an important part of the pollutants. Common atmospheric pollutants include PM2.5, PM10, SO2、NO2CO and O3The pollutants are generally called atmospheric six-parameter pollutants and are recorded in control stations of various countries. In order to avoid serious atmospheric pollution events, the change of the atmospheric six-parameter concentration needs to be predicted in a local future period of time so as to take measures in advance for prevention and control.
In the prior art, methods for predicting the concentration of the atmospheric pollutants include a prediction method based on an autoregressive moving average model (ARMA) and a prediction method based on a differential integration moving average autoregressive model (ARIMA). Both of the above two model prediction methods cannot simultaneously use data of a long historical period for modeling analysis, and the time influence length of pollutant change may be as long as several hundred hours, so that the above two methods are not accurate enough in predicting the atmospheric pollutant concentration in a future period.
Disclosure of Invention
An object of the embodiments of the present application is to provide a method and an apparatus for predicting an atmospheric pollutant concentration based on gaussian regression, so as to improve accuracy of predicting the atmospheric pollutant concentration in a future time period.
In a first aspect, an embodiment of the present application provides a method for predicting an atmospheric pollutant concentration based on gaussian regression, including: acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period; training a Gaussian process regression model by using the training samples to obtain a prediction model; and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
According to the method and the device, the second environmental data in the second preset historical time period are analyzed by using the Gaussian regression model, and the accuracy of prediction of the concentration of the atmospheric pollutants in the future time period can be improved as the Gaussian regression model is trained in hundreds of historical time periods.
Further, the training a gaussian process regression model by using the plurality of training samples to obtain a prediction model includes: constructing a Gaussian kernel function, wherein parameters in the Gaussian kernel function are initial values; and optimizing parameters in the Gaussian kernel function by using the training samples to obtain the prediction model.
According to the method and the device, the Gaussian process regression algorithm is used for regression of the nonlinear relation between the future concentration of the pollutants and the historical concentration and the weather, and the change of the future pollutants can be predicted more accurately.
Further, the optimizing the parameters in the gaussian kernel function by using the plurality of training samples to obtain the prediction model includes: performing the following iterative learning on the parameters in the Gaussian kernel function by using a plurality of training samples until the distance between the obtained prediction data and the concentration data of the various pollutants corresponding to the second time period is less than a preset value; wherein the step of iterating comprises: substituting first environment data corresponding to a first time period in a training sample into the Gaussian kernel function to obtain a first covariance matrix corresponding to the first environment data corresponding to the first time period; sampling according to the first covariance matrix to obtain prediction data; and optimizing parameters in the Gaussian kernel function according to the prediction data and the concentration data of the various pollutants corresponding to the second time period in the training sample.
According to the method and the device, parameters in the Gaussian kernel function are optimized by using historical data, so that an accurate covariance matrix can be obtained according to the Gaussian kernel function, and the accurate atmospheric pollutant concentration in the future time can be obtained.
Further, the Gaussian kernel function is k ═ RBF × periodic × C, wherein,
σ1、l1、σ2、l2、p、σb、σ3and c are parameters of the Gaussian kernel function; t is taAnd tbIs any two indexes; k is the first environmental data x in two training samples over the indexes a and baAnd xbThe covariance of (a).
Further, the analyzing the second environmental data by using the prediction model to obtain pollutant concentration data within a future preset time period output by the prediction model includes: acquiring a Gaussian kernel function corresponding to the prediction model, and determining second covariance matrixes corresponding to the training samples according to the training samples and the Gaussian kernel function; calculating to obtain a mean value and a covariance corresponding to the second environmental data by using a Bayesian formula according to the second covariance matrix; and acquiring pollutant concentration data in a future preset time period corresponding to the second environmental data according to the mean value and the covariance matrix.
According to the embodiment of the application, the historical data are analyzed through the Gaussian regression model, and the nonlinear action relation between different pollutant concentrations and environment variables can be accurately captured.
Further, the obtaining, according to the second covariance matrix, a mean and a covariance corresponding to the second environmental data by using a bayesian formula includes: according to
Calculating to obtain a mean value corresponding to the second environment data; according to
Calculating to obtain a covariance corresponding to the second environment data; wherein the second covariance matrix corresponding to the training sample is
X1Is the training sample; mu.s1And mu2Is the mean of the training samples.
According to the embodiment of the application, the mean value and the covariance corresponding to the second environment data can be accurately obtained through the formula, and then random sampling can be carried out according to the mean value and the covariance to obtain the concentration of the atmospheric pollutants in the future preset time period.
Further, the plurality of pollutant concentration data includes a plurality of items of PM2.5, PM10, sulfur dioxide, carbon monoxide, nitrogen dioxide, and ozone, and the meteorological data includes: at least one of air quality index, weather, wind speed, wind direction, and temperature and relative humidity.
In a second aspect, an embodiment of the present application provides an apparatus for predicting a concentration of an atmospheric pollutant, including: the historical data acquisition module is used for acquiring first environmental data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; a sample construction module, configured to obtain a plurality of training samples from the first environmental data according to a preset time window, where the training samples include environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and an earliest time in the first time period is earlier than an earliest time in the second time period; the model training module is used for training the Gaussian process regression model by using the plurality of training samples to obtain a prediction model; and the prediction module is used for acquiring second environment data within a second preset historical time period from the current moment, analyzing the second environment data by using the prediction model and acquiring pollutant concentration data within a future preset time period output by the prediction model.
In a third aspect, an embodiment of the present application provides an electronic device, including: the system comprises a processor, a memory and a bus, wherein the processor and the memory are communicated with each other through the bus; the memory stores program instructions executable by the processor, the processor being capable of performing the method of the first aspect when invoked by the program instructions.
In a fourth aspect, an embodiment of the present application provides a non-transitory computer-readable storage medium, including: the non-transitory computer readable storage medium stores computer instructions that cause the computer to perform the method of the first aspect.
Additional features and advantages of the present application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the embodiments of the present application. The objectives and other advantages of the application may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
FIG. 1 is a flow chart of STL algorithm calculation provided by the comparison scheme;
FIG. 2 is a diagram of STL decomposition effect provided by the comparison scheme;
FIG. 3 is a graph of STL prediction effect provided by the comparison scheme;
fig. 4 is a schematic flowchart of a method for predicting the concentration of an atmospheric pollutant according to an embodiment of the present disclosure;
FIG. 5 is a flow chart of a sample construction provided by an embodiment of the present application;
fig. 6 is a schematic flow chart of gaussian kernel function parameter optimization provided in the embodiment of the present application;
FIG. 7(a) is a prior distribution plot provided by an embodiment of the present application;
FIG. 7(b) is a posterior distribution chart provided by an embodiment of the present application;
FIG. 8 is a comparison chart of predicted results provided in the examples of the present application;
FIG. 9(a) is a graph illustrating the mean absolute error of predictions for various contaminant concentrations provided by an embodiment of the present application;
FIG. 9(b) is a schematic root mean square error of the predictions for each contaminant concentration provided in the examples of the present application;
FIG. 9(c) is a graph illustrating the mean absolute percentage error for predictions of concentrations of various contaminants provided by an example of the present application;
FIG. 10 is a schematic structural diagram of a prediction device according to an embodiment of the present application;
fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
Prior to the present application, methods for online prediction of atmospheric pollutants included ARMA model-based prediction methods, ARIMA model-based prediction methods, and seasonal trend decomposition model (STL) based prediction methods.
The auto-regressive Moving Average Model (ARMA) is an important method for researching time series, and is formed by mixing an auto-regressive Model (for short, an AR Model) and a Moving Average Model (for short, an MA Model) on the basis.
Autoregressive model AR:
AR model uses Yt-1,...,Yt-pTo predict YtWherein c is a constant term;trandom error values assumed to have a mean equal to 0 and a standard deviation equal to σ; σ is independent of time t. The use of the AR model is premised on the presence of auto-correlation of the Y sequence if YtAnd Yt-iIf the autocorrelation coefficient is less than 0.5, the method is not suitable for use, otherwise, the prediction precision is low.
Moving average model MA:
MA model considers YtRandom bias part of values that cannot be regressed using AR model
And can be obtained by historical random bias regression.
This way the AR model and MA model are combined to obtain the following ARMA model:
in practical application, it is necessary to determine the data stationarity (each statistical parameter of the time series is not related to time), then determine the model parameters p and q by the autocorrelation graph and the partial autocorrelation graph, finally substitute the model data into the determined parameter ARMA model to regress c, β in the above formulaiAnd ωiAnd (4) parameters.
In pollutant concentration prediction, after a batch of pollutant concentration historical samples are obtained, the ARMA model can be used for regressing a time value Y to be predicted in the historical samplestAnd the historical value Yt-iAnd random perturbationt、t-iAnd finally predicting the Yt value at the next moment, wherein the prediction result is expressed as:
Yt＝ARMA(Yt-1,...,Yt-p)
differential integration moving average autoregressive model ARIMA
The ARMA model is suitable on the premise that a time sequence meets time sequence stationarity, when the time sequence sample does not meet the stationarity, the time sequence sample can be differentiated to enable the differentiated sequence to meet the stationarity, and finally the ARMA model is used for modeling the differentiated stationary sequence, so that the model is called ARIMA (p, d, q), wherein parameters p and q are consistent with the ARMA model, parameter d represents a differential order, and generally d is 0, 1 and 2. When d is 0, the time series is not subjected to difference, and when d is 1 or 2, the time series is subjected to the following first-order or second-order difference:
d＝1:ΔYt＝Yt+1-Yt
d＝2:Δ2Yt＝Yt+2-2Yt+1+Yt
the prediction steps of the ARIMA model on the pollutant concentration are similar to those of the ARMA model, and the predicted value expression is as follows: y ist＝ARIMA(Yt-1,...,Yt-p)。
Seasonal trend decomposition model STL
A Seasonal-Trend Decomposition (STL) algorithm is a common algorithm in time sequence Decomposition, and decomposes data Yt at a certain moment into a Trend component, a periodic component and a remainder Based on LOESS, and realizes time sequence Decomposition by taking robust local weighted regression as a smoothing method, wherein Y is Yt＝Tt+St+Rt,t＝1,...,N；
Loess (local weighted regression) adopted by the algorithm is local polynomial regression fitting, is a common method for smoothing a two-dimensional scatter diagram, and combines the simplicity of the traditional linear regression and the flexibility of the nonlinear regression. When a response variable value is estimated, a data subset is taken from the vicinity of a predictive variable of the response variable value, then linear regression or quadratic regression is carried out on the data subset, a weighted least square method is adopted during regression, namely the weight of a value closer to an estimation point is larger, and finally the value of the response variable is estimated by using an obtained local regression model. The whole fitting curve is obtained by performing point-by-point operation by the method.
The STL is robust to outliers and can only handle decomposition in the addition mode, requiring conversion to the addition mode for processing and then inverse transformation for the multiplication mode.
The calculation flow of the STL algorithm is shown in figure 1, the STL decomposition effect is shown in figure 2, and the STL decomposition effect comprises an original sequence, a trend item, a seasonal period item and a residual item from the last time to the next time.
The prediction effect using the STL decomposition results in combination with random walk sampling is shown in fig. 3, where the prediction value expression is: y ist＝STL(Yt-1,...,Yt-p)。
The ARMA model and the ARIMA model are based on the premise that the time sequence is stable or the difference is stable, and are difficult to ensure under the condition of less data volume; in addition, both methods try to regress the linear correlation relationship between the future value and the historical record, in the pollutant concentration record, the pollutant concentration changes are mutually coupled and nonlinearly correlated, and the ARMA model and the ARIMA model are no longer suitable for the situation, so that the long-time change of the future pollutant concentration cannot be accurately predicted. The STL decomposition algorithm requires manual setting of corresponding frequency band or period parameters for decomposition, and is greatly influenced by subjective factors. Therefore, the method has low accuracy in predicting the concentration of the atmospheric pollutants in the future preset time period.
Based on the above technical problem, the embodiments of the present application provide a method for predicting the concentration of an atmospheric pollutant based on gaussian regression. It can be understood that the prediction method for the concentration of the atmospheric pollutant provided by the embodiment of the present application can be applied to a terminal device (also referred to as an electronic device) and a server; the terminal device may be a smart phone, a tablet computer, a Personal Digital Assistant (PDA), or the like; the server may specifically be an application server, and may also be a Web server.
For convenience of understanding, in the technical solution provided in the embodiment of the present application, an application scenario of the prediction method for the concentration of the atmospheric pollutant provided in the embodiment of the present application is described below by taking a terminal device as an execution subject.
Fig. 4 is a schematic flow chart of a method for predicting the concentration of an atmospheric pollutant provided in an embodiment of the present application, and as shown in fig. 4, the method mainly includes two parts, namely, on-line training of a model and prediction by using the trained model. The specific process comprises the following steps:
step 401: acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data includes a plurality of pollutant concentration data and meteorological data.
In a specific implementation process, the current time may be a real current time, or may be a historical time. The concrete determination can be carried out according to the actual situation. For example: now 10 am at 4/8/2019, assuming that the user wants to know the concentration of atmospheric pollutants in a certain future time period after 10 am at 4/8/2019, the current time is the real current time; assuming that the user wants to know the concentration of the atmospheric pollutants in a future time period after 11 am of 4/7/2019, the current time is a historical time, i.e., 11 am of 4/7/2019.
The first preset historical time period may be 360 hours or 400 hours, and the specific time length may be preset according to an actual situation.
The first environmental data includes a plurality of pollutant concentration data and meteorological data, wherein the plurality of pollutant concentration data includes a plurality of PM2.5, PM10, sulfur dioxide, carbon monoxide, nitrogen dioxide, and ozone, the meteorological data including: at least one of air quality index, weather, wind speed, wind direction, and temperature and relative humidity. It is understood that the first environment data may be obtained from a corresponding monitoring station, or may be obtained from other places, for example, a weather station, the internet, and the like.
Step 402: obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period.
In a specific implementation process, the preset time window may be 72 hours, or may be other more or less hours, which may be determined according to actual situations. The preset time window is equal to a union of the first time period and the second time period. And the earliest time in the first time period is earlier than the earliest time in the second time period. For example: first environment data 360 hours before the current time is acquired, the 360 hours are sequenced from morning to evening, and a time sequence 0,2, 3. Assuming that the first time period is 49 hours, the second time period is 24 hours, and the time corresponding to the 48 th time series is taken as the current time, then 0-47 are the historical time periods corresponding to the 48 th time series, and the historical time periods and the current time periods are combined into the first time period, and 49-72 are the second time periods. When a training sample is obtained, the first environmental data corresponding to the first time period is used as input data of the training sample, and the concentration data of multiple pollutants corresponding to the second time period is used as a label.
It should be noted that if the factor of weather is included in the first environment data, the first time period corresponding to weather includes data of the historical time period and the future time period, i.e., 0 to 72 are the first time period corresponding to the factor of weather.
Fig. 5 is a flow chart of a sample configuration provided in an embodiment of the present application, where, as shown in fig. 5, the first environmental data includes PM2.5, PM10, and weather, and the pollutant to be predicted includes PM2.5 and PM10, it is understood that the first environmental data may also include other factors listed above, and likewise, the pollutant to be predicted may also include other factors listed above. Selecting the time t as the current time, taking the PM2.5 concentration and the PM10 concentration of 0-47 time series before the time t (including the time t), the weather data of 0-47 time series before the time t (including the time t) and the weather data of 0-23 time series after the time t as input data of a training sample, namely an X sample, and taking the PM2.5 concentration and the PM10 concentration of 0-23 time series after the time t as labels of the training sample, namely a Y sample. X samples and Y samples constitute a training sample.
It will be appreciated that the construction of multiple training samples may be performed by varying the value at time t.
Step 403: and training the Gaussian process regression model by using the plurality of training samples to obtain a prediction model.
In a specific implementation process, a Gaussian process regression model is adopted to carry out regression on the relation between the X sample and the Y sample in the training sample, so that a prediction model is obtained.
Step 404: and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
In a specific implementation process, the current time here is consistent with the current time in step 401, and details are not described here. The second preset historical time period may also be 49 hours, and the current specific value of the second preset historical time period may be set according to the actual situation. The factor type included in the second environment data is consistent with the first environment data, and is not described herein again. And inputting second environment data which is 48 hours before the current time and at the current time into a trained prediction model, and analyzing the second environment data by the prediction model to obtain pollutant concentration data output by the prediction model in a future preset time period. It should be noted that the preset time period in the future may be 24 hours after the current time, or may be other longer or shorter time periods, which may be determined specifically during the training of the model. The obtained contaminant species within the future preset time period is also determined when training the model. I.e. which contaminant concentrations are included in the Y sample, the predictive model is able to output which contaminant concentrations.
According to the method and the device, the second environmental data in the second preset historical time period are analyzed by using the Gaussian regression model, and the accuracy of prediction of the concentration of the atmospheric pollutants in the future time period can be improved as the Gaussian regression model is trained in hundreds of historical time periods.
On the basis of the foregoing embodiment, the training a gaussian process regression model by using the plurality of training samples to obtain a prediction model includes:
constructing a Gaussian kernel function, wherein parameters in the Gaussian kernel function are initial values;
and optimizing parameters in the Gaussian kernel function by using the training samples to obtain the prediction model.
In a specific implementation, the gaussian process is a combination of a series of random variables in an exponential set that obeys a normal distribution. Let random variable X ═ X1,x2,...,xn]TThe subscript t 1.. times.n constitutes a set of indices, making X obey a normal distribution: x to N (μ, Σ), where μ is the mean vector and μ ═ μ2,...,μn]TAnd Σ is a covariance matrix, and can be calculated by a gaussian process kernel function. In general, the mean value is 0, so the emphasis is on the calculation of covariance. Before the covariance is calculated, a gaussian kernel function needs to be obtained, and the gaussian kernel function is obtained as follows:
a gaussian kernel function is initialized in advance, and the gaussian kernel function used in the embodiment of the present application is a product of an RBF kernel, a periodic kernel and a linear kernel, i.e., k ═ RBF × periodic × C, wherein,
σ1、l1、σ2、l2、p、σb、σ3and c are parameters of the Gaussian kernel function, in the initialized Gaussian kernel function, the parameters are also initialized values which can be all 1, and can also be values according to parameters in other pre-training models; t is taAnd tbIs any two indexes; k is the first environmental data x in two training samples over the indexes a and baAnd xbThe covariance of (a).
From the above equation, after obtaining the optimal parameter value in the gaussian kernel function, the covariance of the unobserved data can be obtained. The unobserved data refers to data without a corresponding future prediction result. The second environment data in the embodiment of the present application may be understood as unobserved data.
After the initial gaussian kernel function is obtained, parameters in the gaussian kernel function can be optimized using a plurality of training samples in order to obtain an optimal set of parameters. The specific optimization process is shown in fig. 6, and includes:
step 601: substituting first environment data corresponding to a first time period in a training sample into the Gaussian kernel function to obtain a first covariance matrix corresponding to the first environment data corresponding to the first time period; the first environment data comprises historical samples X constructed in a first time period and future change data real values Y;
step 602: sampling according to the first covariance matrix to obtain prediction data;
step 603: judging whether to stop optimization; and calculating the distance between the prediction data and the concentration of the various pollutants corresponding to the second time period in the training sample, wherein the distance is used for representing the difference between the prediction data and the concentration of the various pollutants corresponding to the second time period, and can be a Euclidean distance, a Manhattan distance and the like. Judging whether the optimization stopping condition is met or not according to the distance obtained by calculation, if so, taking a group of finally obtained parameters as optimal parameters, otherwise, executing the step 603; and stopping optimizing under the condition that the distance between the obtained prediction data and the concentration data of the various pollutants corresponding to the second time period is smaller than a preset value.
Step 604: and optimizing parameters in the Gaussian kernel function according to the prediction data and the concentration data of the various pollutants corresponding to the second time period in the training sample, and executing step 601.
After the parameters in the optimal Gaussian kernel function are obtained, a prediction model can be obtained.
It can be understood that, through experiments, the optimization process of the parameters only needs 10 to 20 seconds, and can be trained on line in real time.
According to the method and the device, the Gaussian process regression algorithm is used for regression of the nonlinear relation between the future concentration of the pollutants and the historical concentration and the weather, and the change of the future pollutants can be predicted more accurately.
On the basis of the foregoing embodiment, the analyzing the second environmental data by using the prediction model to obtain pollutant concentration data within a future preset time period output by the prediction model includes:
acquiring a Gaussian kernel function corresponding to the prediction model, and determining second covariance matrixes corresponding to the training samples according to the training samples and the Gaussian kernel function;
calculating to obtain a mean value and a covariance corresponding to the second environmental data by using a Bayesian formula according to the second covariance matrix;
and acquiring pollutant concentration data in a future preset time period corresponding to the second environmental data according to the mean value and the covariance matrix.
In a specific implementation process, after a trained prediction model is obtained, a plurality of training samples can be calculated according to a gaussian kernel function in the prediction model to obtain corresponding second covariance matrices, the training samples can be understood as observed variables, and the observed variables are X1＝[x1,...,xm]M is the number of training samples, and the unobserved variable is X2＝[xm+1,...,xn]Then, then
The mean value is:
the second covariance matrix is:
according to Bayes' formula, using X1Information pair X of2The posterior distribution parameters of (a) are adjusted, and the mean and covariance of the posterior distribution on the exponential set of the unobserved samples are calculated. The calculation formula of the mean and the covariance is as follows:
wherein, mu2|1Being the mean of a posterior distribution, sigma2|1Is the covariance of the posterior distribution.
After the mean and covariance are obtained, the mean and covariance are substituted into the prediction model to regress the change of the second environment data over the entire set of indices.
Fig. 7(a) is a prior distribution diagram provided in the embodiment of the present application, fig. 7(b) is a posterior distribution diagram provided in the embodiment of the present application, and black dots in fig. 7(b) are observation samples.
The prediction model can predict the concentrations of six pollutants simultaneously, fig. 8 is a comparison graph of prediction results provided by the embodiment of the present application, historical data is obtained before time w, a dotted line after time w represents a true value of each pollutant concentration, and a solid line represents a predicted value of each pollutant concentration obtained through prediction by the prediction model of the present application. It can be seen that the prediction model can well capture the change of each pollutant in a future period of time.
FIG. 9(a) is a graph illustrating the mean absolute error of predictions for various contaminant concentrations provided by an embodiment of the present application; FIG. 9(b) is a schematic root mean square error of the predictions for each contaminant concentration provided in the examples of the present application; fig. 9(c) is a graph illustrating the mean absolute percentage error of predictions for each contaminant concentration provided in the examples of the present application. It is to be understood that the error map is only calculated after prediction of the concentrations of the pollutants for a future period of time according to a certain period of historical data, and is only used for quantification of the performance of the prediction model provided by the embodiment of the present application.
In summary, the prediction model can be obtained to predict the errors of six pollutants in the future period as shown in the following table:
| PM10 | PM2.5 | CO | O3 | NO2 | SO2 | |
| MAE | 23 | 15 | 0.38 | 16 | 5.2 | 2.1 |
| |
30 | 21 | 0.5 | 21 | 6.9 | 2.7 |
| SMAPE | 0.77 | 0.92 | 0.49 | 0.41 | 0.76 | 0.57 |
according to the table, the prediction method provided by the embodiment of the application can accurately predict the concentration of each pollutant in a future period of time.
Fig. 10 is a schematic structural diagram of a prediction apparatus according to an embodiment of the present application, where the prediction apparatus may be a module, a program segment, or code on an electronic device. It should be understood that the apparatus corresponds to the above-mentioned embodiment of the method of fig. 4, and can perform various steps related to the embodiment of the method of fig. 4, and the specific functions of the apparatus can be referred to the description above, and the detailed description is appropriately omitted here to avoid redundancy. The device includes: a historical data acquisition module 1001, a sample construction module 1002, a model training module 1003, and a prediction module 1004, wherein:
the historical data acquisition module 1001 is configured to acquire first environmental data within a first preset historical time period from a current time; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; the sample construction module 1002 is configured to obtain a plurality of training samples from the first environmental data according to a preset time window, where the training samples include environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and an earliest time in the first time period is earlier than an earliest time in the second time period; the model training module 1003 is configured to train a gaussian process regression model by using the plurality of training samples to obtain a prediction model; the prediction module 1004 is configured to obtain second environmental data within a second preset historical time period from the current time, analyze the second environmental data by using the prediction model, and obtain pollutant concentration data within a future preset time period output by the prediction model.
On the basis of the foregoing embodiment, the model training module 1003 is specifically configured to:
constructing a Gaussian kernel function, wherein parameters in the Gaussian kernel function are initial values;
and optimizing parameters in the Gaussian kernel function by using the training samples to obtain the prediction model.
On the basis of the foregoing embodiment, the model training module 1003 is specifically configured to:
performing the following iterative learning on the parameters in the Gaussian kernel function by using a plurality of training samples until the distance between the obtained prediction data and the concentration data of the various pollutants corresponding to the second time period is less than a preset value; wherein the step of iterating comprises:
substituting first environment data corresponding to a first time period in a training sample into the Gaussian kernel function to obtain a first covariance matrix corresponding to the first environment data corresponding to the first time period;
sampling according to the first covariance matrix to obtain prediction data;
and optimizing parameters in the Gaussian kernel function according to the prediction data and the concentration data of the various pollutants corresponding to the second time period in the training sample.
On the basis of the above embodiment, the gaussian kernel function is:
k＝RBF×periodic×C；
wherein the content of the first and second substances,
σ1、l1、σ2、l2、p、σb、σ3and c are parameters of the Gaussian kernel function; t is taAnd tbIs any two indexes; k is the first environmental data x in two training samples over the indexes a and baAnd xbThe covariance of (a).
On the basis of the foregoing embodiment, the prediction module 1004 is specifically configured to:
acquiring a Gaussian kernel function corresponding to the prediction model, and determining second covariance matrixes corresponding to the training samples according to the training samples and the Gaussian kernel function;
calculating to obtain a mean value and a covariance corresponding to the second environmental data by using a Bayesian formula according to the second covariance matrix;
and acquiring pollutant concentration data in a future preset time period corresponding to the second environmental data according to the mean value and the covariance matrix.
On the basis of the foregoing embodiment, the prediction module 1004 is specifically configured to:
according to
Calculating to obtain a mean value corresponding to the second environment data;
according to
Calculating to obtain a covariance corresponding to the second environment data;
wherein the second covariance matrix corresponding to the training sample is
X1Is the training sample; mu.s1And mu2Is the mean of the training samples.
On the basis of the above embodiment, the plurality of pollutant concentration data includes a plurality of items of PM2.5, PM10, sulfur dioxide, carbon monoxide, nitrogen dioxide and ozone, and the meteorological data includes: at least one of air quality index, weather, wind speed, wind direction, and temperature and relative humidity.
Fig. 11 is a schematic structural diagram of an entity of an electronic device provided in an embodiment of the present application, and as shown in fig. 11, the electronic device includes: a processor (processor)1101, a memory (memory)1102, and a bus 1103; wherein the content of the first and second substances,
the processor 1101 and the memory 1102 communicate with each other via the bus 1103;
the processor 1101 is configured to call the program instructions in the memory 1102 to perform the methods provided by the above-mentioned method embodiments, for example, including: acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period; training a Gaussian process regression model by using the training samples to obtain a prediction model; and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
The processor 1101 may be an integrated circuit chip having signal processing capabilities. The processor 1101 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. Which may implement or perform the various methods, steps, and logic blocks disclosed in the embodiments of the present application. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The Memory 1102 may include, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read Only Memory (PROM), Erasable Read Only Memory (EPROM), Electrically Erasable Read Only Memory (EEPROM), and the like.
The present embodiment discloses a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the method provided by the above-mentioned method embodiments, for example, comprising: acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period; training a Gaussian process regression model by using the training samples to obtain a prediction model; and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
The present embodiments provide a non-transitory computer-readable storage medium storing computer instructions that cause the computer to perform the methods provided by the above method embodiments, for example, including: acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data; obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period; training a Gaussian process regression model by using the training samples to obtain a prediction model; and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
In addition, units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
Furthermore, the functional modules in the embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (10)
1. A prediction method of atmospheric pollutant concentration based on Gaussian regression is characterized by comprising the following steps:
acquiring first environment data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data;
obtaining a plurality of training samples from the first environmental data according to a preset time window, wherein the training samples comprise environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and the earliest time in the first time period is earlier than the earliest time in the second time period;
training a Gaussian process regression model by using the training samples to obtain a prediction model;
and acquiring second environment data within a second preset historical time period from the current moment, and analyzing the second environment data by using the prediction model to acquire pollutant concentration data within a future preset time period output by the prediction model.
2. The method of claim 1, wherein training a gaussian process regression model with the plurality of training samples to obtain a prediction model comprises:
constructing a Gaussian kernel function, wherein parameters in the Gaussian kernel function are initial values;
and optimizing parameters in the Gaussian kernel function by using the training samples to obtain the prediction model.
3. The method of claim 2, wherein the optimizing the parameters in the gaussian kernel function using the plurality of training samples to obtain the prediction model comprises:
performing the following iterative learning on the parameters in the Gaussian kernel function by using a plurality of training samples until the distance between the obtained prediction data and the concentration data of the various pollutants corresponding to the second time period is less than a preset value; wherein the step of iterating comprises:
substituting first environment data corresponding to a first time period in a training sample into the Gaussian kernel function to obtain a first covariance matrix corresponding to the first environment data corresponding to the first time period;
sampling according to the first covariance matrix to obtain prediction data;
and optimizing parameters in the Gaussian kernel function according to the prediction data and the concentration data of the various pollutants corresponding to the second time period in the training sample.
4. The method of claim 2, wherein the gaussian kernel function is:
k＝RBF×periodic×C；
wherein the content of the first and second substances,
σ1、l1、σ2、l2、p、σb、σ3and c are parameters of the Gaussian kernel function; t is taAnd tbIs any two indexes; k is the first environmental data x in two training samples over the indexes a and baAnd xbThe covariance of (a).
5. The method of claim 1, wherein analyzing the second environmental data using the predictive model to obtain pollutant concentration data within a predetermined time period in the future from the output of the predictive model comprises:
acquiring a Gaussian kernel function corresponding to the prediction model, and determining second covariance matrixes corresponding to the training samples according to the training samples and the Gaussian kernel function;
calculating to obtain a mean value and a covariance corresponding to the second environmental data by using a Bayesian formula according to the second covariance matrix;
and acquiring pollutant concentration data in a future preset time period corresponding to the second environmental data according to the mean value and the covariance matrix.
6. The method according to claim 5, wherein the calculating the mean and covariance corresponding to the second environment data according to the second covariance matrix by using a Bayesian formula comprises:
according to
Calculating to obtain a mean value corresponding to the second environment data;
according to
Calculating to obtain a covariance corresponding to the second environment data;
wherein the second covariance matrix corresponding to the training sample is
X1Is the training sample; mu.s1And mu2Is the mean of the training samples.
7. The method of any one of claims 1-6, wherein the plurality of pollutant concentration data includes a plurality of PM2.5, PM10, sulfur dioxide, carbon monoxide, nitrogen dioxide, and ozone, and the meteorological data includes: at least one of air quality index, weather, wind speed, wind direction, and temperature and relative humidity.
8. An apparatus for predicting the concentration of an atmospheric pollutant, comprising:
the historical data acquisition module is used for acquiring first environmental data within a first preset historical time period from the current moment; wherein the environmental data comprises a plurality of pollutant concentration data and meteorological data;
a sample construction module, configured to obtain a plurality of training samples from the first environmental data according to a preset time window, where the training samples include environmental data corresponding to a first time period and a plurality of pollutant concentration data corresponding to a second time period, and an earliest time in the first time period is earlier than an earliest time in the second time period;
the model training module is used for training the Gaussian process regression model by using the plurality of training samples to obtain a prediction model;
and the prediction module is used for acquiring second environment data within a second preset historical time period from the current moment, analyzing the second environment data by using the prediction model and acquiring pollutant concentration data within a future preset time period output by the prediction model.
9. An electronic device, comprising: a processor, a memory, and a bus, wherein,
the processor and the memory are communicated with each other through the bus;
the memory stores program instructions executable by the processor, the processor invoking the program instructions to perform the method of any one of claims 1-7.
10. A non-transitory computer-readable storage medium storing computer instructions which, when executed by a computer, cause the computer to perform the method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010601670.4A CN111754042A (en) | 2020-06-28 | 2020-06-28 | Atmospheric pollutant concentration prediction method and device based on Gaussian regression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010601670.4A CN111754042A (en) | 2020-06-28 | 2020-06-28 | Atmospheric pollutant concentration prediction method and device based on Gaussian regression |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111754042A true CN111754042A (en) | 2020-10-09 |
Family
ID=72677747
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010601670.4A Pending CN111754042A (en) | 2020-06-28 | 2020-06-28 | Atmospheric pollutant concentration prediction method and device based on Gaussian regression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111754042A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112288192A (en) * | 2020-11-23 | 2021-01-29 | 深圳市维度数据科技股份有限公司 | Environment-friendly monitoring and early warning method and system |
| CN112540159A (en) * | 2020-11-27 | 2021-03-23 | 中广核工程有限公司 | Nuclear power plant atmospheric diffusion prediction correction method, system, medium and electronic equipment |
| CN112561191A (en) * | 2020-12-22 | 2021-03-26 | 北京百度网讯科技有限公司 | Prediction model training method, prediction method, device, apparatus, program, and medium |
| CN112580898A (en) * | 2020-12-30 | 2021-03-30 | 成都佳华物链云科技有限公司 | Pollution prediction method and pollution prediction model training method |
| CN112598178A (en) * | 2020-12-23 | 2021-04-02 | 北方民族大学 | Atmospheric pollutant concentration prediction method based on Bayesian framework |
| CN112836179A (en) * | 2021-02-07 | 2021-05-25 | 北京中关村水木医疗科技有限公司 | Method and device for simulating medical instrument transportation environment |
| CN113255137A (en) * | 2021-05-31 | 2021-08-13 | 中铁第一勘察设计院集团有限公司 | Target object strain data processing method and device and storage medium |
| CN113380340A (en) * | 2021-06-21 | 2021-09-10 | 深圳开云智能有限公司 | Training method and device of fly ash concentration prediction model and computer equipment |
| CN113627529A (en) * | 2021-08-11 | 2021-11-09 | 成都佳华物链云科技有限公司 | Air quality prediction method, device, electronic equipment and storage medium |
| CN114488801A (en) * | 2022-01-18 | 2022-05-13 | 无锡安起科技有限公司 | Bridge crane model prediction control method based on data driving |
| CN116153425A (en) * | 2023-02-24 | 2023-05-23 | 北京和利时工业软件有限公司 | Ammonia-carbon ratio soft measurement method, device, equipment and medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108053071A (en) * | 2017-12-21 | 2018-05-18 | 宇星科技发展(深圳)有限公司 | Regional air pollutant concentration Forecasting Methodology, terminal and readable storage medium storing program for executing |
| CN108108836A (en) * | 2017-12-15 | 2018-06-01 | 清华大学 | A kind of ozone concentration distribution forecasting method and system based on space-time deep learning |
| CN108776717A (en) * | 2018-04-25 | 2018-11-09 | 香港中文大学（深圳） | Kernel and data predication method, device, equipment and storage medium |
| CN110070224A (en) * | 2019-04-20 | 2019-07-30 | 北京工业大学 | A kind of Air Quality Forecast method based on multi-step recursive prediction |
| CN110263479A (en) * | 2019-06-28 | 2019-09-20 | 浙江航天恒嘉数据科技有限公司 | A kind of air pollution agent concentration spatial and temporal distributions prediction technique and system |
| CN110333556A (en) * | 2019-06-03 | 2019-10-15 | 深圳中兴网信科技有限公司 | Air Quality Forecast method, apparatus, computer equipment and readable storage medium storing program for executing |
| CN111222685A (en) * | 2019-11-21 | 2020-06-02 | 江苏方天电力技术有限公司 | Boiler wide-load NOx emission concentration prediction method based on model migration |
-
2020
- 2020-06-28 CN CN202010601670.4A patent/CN111754042A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108108836A (en) * | 2017-12-15 | 2018-06-01 | 清华大学 | A kind of ozone concentration distribution forecasting method and system based on space-time deep learning |
| CN108053071A (en) * | 2017-12-21 | 2018-05-18 | 宇星科技发展(深圳)有限公司 | Regional air pollutant concentration Forecasting Methodology, terminal and readable storage medium storing program for executing |
| CN108776717A (en) * | 2018-04-25 | 2018-11-09 | 香港中文大学（深圳） | Kernel and data predication method, device, equipment and storage medium |
| CN110070224A (en) * | 2019-04-20 | 2019-07-30 | 北京工业大学 | A kind of Air Quality Forecast method based on multi-step recursive prediction |
| CN110333556A (en) * | 2019-06-03 | 2019-10-15 | 深圳中兴网信科技有限公司 | Air Quality Forecast method, apparatus, computer equipment and readable storage medium storing program for executing |
| CN110263479A (en) * | 2019-06-28 | 2019-09-20 | 浙江航天恒嘉数据科技有限公司 | A kind of air pollution agent concentration spatial and temporal distributions prediction technique and system |
| CN111222685A (en) * | 2019-11-21 | 2020-06-02 | 江苏方天电力技术有限公司 | Boiler wide-load NOx emission concentration prediction method based on model migration |
Non-Patent Citations (4)
| Title |
|---|
| 常和玲: "高斯过程混合模型在时间序列预测中的应用研究", no. 2, pages 140 - 385 * |
| 庞蕊: "高斯过程回归在气象数据方面的应用研究", no. 1, pages 009 - 44 * |
| 康军,等: "高斯过程回归短时交通流预测方法", vol. 15, no. 4, pages 52 - 54 * |
| 彭喜元, 等: "《数据驱动的故障预测》", vol. 978, 31 March 2016, 哈尔滨工业大学出版社, pages: 170 - 173 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112288192A (en) * | 2020-11-23 | 2021-01-29 | 深圳市维度数据科技股份有限公司 | Environment-friendly monitoring and early warning method and system |
| CN112540159A (en) * | 2020-11-27 | 2021-03-23 | 中广核工程有限公司 | Nuclear power plant atmospheric diffusion prediction correction method, system, medium and electronic equipment |
| CN112561191A (en) * | 2020-12-22 | 2021-03-26 | 北京百度网讯科技有限公司 | Prediction model training method, prediction method, device, apparatus, program, and medium |
| CN112561191B (en) * | 2020-12-22 | 2024-02-27 | 北京百度网讯科技有限公司 | Prediction model training method, prediction device, prediction apparatus, prediction program, and program |
| CN112598178B (en) * | 2020-12-23 | 2022-03-29 | 北方民族大学 | Atmospheric pollutant concentration prediction method based on Bayesian framework |
| CN112598178A (en) * | 2020-12-23 | 2021-04-02 | 北方民族大学 | Atmospheric pollutant concentration prediction method based on Bayesian framework |
| CN112580898A (en) * | 2020-12-30 | 2021-03-30 | 成都佳华物链云科技有限公司 | Pollution prediction method and pollution prediction model training method |
| CN112580898B (en) * | 2020-12-30 | 2023-11-03 | 成都佳华物链云科技有限公司 | Pollution prediction method and pollution prediction model training method |
| CN112836179B (en) * | 2021-02-07 | 2023-07-14 | 北京中关村水木医疗科技有限公司 | Method and device for simulating medical instrument transportation environment |
| CN112836179A (en) * | 2021-02-07 | 2021-05-25 | 北京中关村水木医疗科技有限公司 | Method and device for simulating medical instrument transportation environment |
| CN113255137A (en) * | 2021-05-31 | 2021-08-13 | 中铁第一勘察设计院集团有限公司 | Target object strain data processing method and device and storage medium |
| CN113380340A (en) * | 2021-06-21 | 2021-09-10 | 深圳开云智能有限公司 | Training method and device of fly ash concentration prediction model and computer equipment |
| CN113627529A (en) * | 2021-08-11 | 2021-11-09 | 成都佳华物链云科技有限公司 | Air quality prediction method, device, electronic equipment and storage medium |
| CN114488801A (en) * | 2022-01-18 | 2022-05-13 | 无锡安起科技有限公司 | Bridge crane model prediction control method based on data driving |
| CN114488801B (en) * | 2022-01-18 | 2023-12-22 | 无锡安起科技有限公司 | Bridge crane model prediction control method based on data driving |
| CN116153425A (en) * | 2023-02-24 | 2023-05-23 | 北京和利时工业软件有限公司 | Ammonia-carbon ratio soft measurement method, device, equipment and medium |
| CN116153425B (en) * | 2023-02-24 | 2024-04-02 | 北京和利时工业软件有限公司 | Ammonia-carbon ratio soft measurement method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111754042A (en) | Atmospheric pollutant concentration prediction method and device based on Gaussian regression | |
| CN109816221B (en) | Project risk decision method, apparatus, computer device and storage medium | |
| CN107153874B (en) | Water quality prediction method and system | |
| CN111832222B (en) | Pollutant concentration prediction model training method, pollutant concentration prediction method and pollutant concentration prediction device | |
| Guo et al. | Output-relevant variational autoencoder for just-in-time soft sensor modeling with missing data | |
| CN111814956A (en) | Multi-task learning air quality prediction method based on multi-dimensional secondary feature extraction | |
| CN115587666A (en) | Load prediction method and system based on seasonal trend decomposition and hybrid neural network | |
| Xiao et al. | An adaptive decomposition and ensemble model for short-term air pollutant concentration forecast using ICEEMDAN-ICA | |
| Xiong et al. | Soft sensor modeling with a selective updating strategy for Gaussian process regression based on probabilistic principle component analysis | |
| CN115983087A (en) | Method for detecting time sequence data abnormity by combining attention mechanism and LSTM and terminal | |
| CN116028315A (en) | Operation early warning method, device, medium and electronic equipment | |
| CN114564345A (en) | Server abnormity detection method, device, equipment and storage medium | |
| Zhong et al. | Online prediction of noisy time series: Dynamic adaptive sparse kernel recursive least squares from sparse and adaptive tracking perspective | |
| CN112949947A (en) | Power failure early warning method and system based on convolution long-term and short-term memory network | |
| Nieto | Forecasting with univariate TAR models | |
| CN114970745B (en) | Intelligent security and environment big data system of Internet of things | |
| Khennou et al. | Forecasting covid-19 spreading in canada using deep learning | |
| CN116665798A (en) | Air pollution trend early warning method and related device | |
| CN116502672A (en) | Neural network quantitative deployment method, system, equipment and medium | |
| CN115099488A (en) | CNN-AGU-based air quality index prediction method and device | |
| Allen | Advanced statistical post-processing of ensemble weather forecasts | |
| CN112529268A (en) | Medium-short term load prediction method and device based on manifold learning | |
| CN116467645B (en) | Pollution source collection monitoring system | |
| CN117671507B (en) | River water quality prediction method combining meteorological data | |
| CN109800923B (en) | Short-term power combination prediction method for distributed wind power generation |
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 |