CN114298389A - Ozone concentration forecasting method and device - Google Patents
Ozone concentration forecasting method and device Download PDFInfo
- Publication number
- CN114298389A CN114298389A CN202111583015.1A CN202111583015A CN114298389A CN 114298389 A CN114298389 A CN 114298389A CN 202111583015 A CN202111583015 A CN 202111583015A CN 114298389 A CN114298389 A CN 114298389A
- Authority
- CN
- China
- Prior art keywords
- data
- ozone concentration
- time
- forecasting
- ozone
- 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
- CBENFWSGALASAD-UHFFFAOYSA-N Ozone Chemical compound [O-][O+]=O CBENFWSGALASAD-UHFFFAOYSA-N 0.000 title claims abstract description 148
- 238000013277 forecasting method Methods 0.000 title claims abstract description 17
- 238000004088 simulation Methods 0.000 claims abstract description 47
- 238000000034 method Methods 0.000 claims abstract description 45
- 230000007246 mechanism Effects 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims description 33
- 239000003344 environmental pollutant Substances 0.000 claims description 20
- 231100000719 pollutant Toxicity 0.000 claims description 20
- 238000013528 artificial neural network Methods 0.000 claims description 19
- 230000015654 memory Effects 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 230000005855 radiation Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 abstract description 11
- 230000002265 prevention Effects 0.000 abstract description 5
- 238000012950 reanalysis Methods 0.000 abstract description 4
- 239000000126 substance Substances 0.000 abstract description 2
- 238000004590 computer program Methods 0.000 description 12
- 238000004891 communication Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 239000000356 contaminant Substances 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 238000011426 transformation method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000010267 cellular communication Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an ozone concentration forecasting method and device, and relates to the field of atmospheric chemical data processing. The method comprises the following steps: acquiring observation data and numerical simulation data of the time of start-up; constructing characteristic data based on the observation data, the numerical simulation data and the characteristic engineering of the time of the start-up; inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time. According to the method, single-time observation data is combined with meteorological reanalysis data forecast data, a plurality of characteristics are constructed by using characteristic engineering, the importance of the characteristics is evaluated by using an attention mechanism, different weights are given, and the accuracy of ozone forecast on multiple time scales is improved. Moreover, the hourly ozone concentration is forecasted based on multi-source data based on observation data, so that the ozone forecasting precision can be improved, the ozone forecasting efficiency is improved, and more powerful support is provided for scientific prevention and control of ozone pollution.
Description
Technical Field
The invention relates to the field of atmospheric chemical data processing, in particular to an ozone concentration forecasting method and device.
Background
The ozone prediction problem based on deep learning can be described as predicting the ozone concentration at a future time based on the existing multi-source data mainly based on observation data.
At this stage, many scholars apply machine learning and deep learning algorithms to ozone prediction, and methods may include machine learning methods as well as deep learning methods. Performing characteristic engineering on multiple basis of original data by using a machine learning method such as an extreme gradient lifting tree, support vector machine regression and a neural network, and performing ozone concentration prediction by using data obtained after multi-source characteristic construction as input; in the deep learning method such as a recurrent neural network, an attention mechanism and the like, observation data of a plurality of times are input in the using process, each time comprises a plurality of input elements, and the algorithm extracts the importance of different times but cannot evaluate the importance of the elements in a single time. At present, the two methods have the problem of large dimensionality of input data, so that the calculated amount is large, and the prediction efficiency is low.
Disclosure of Invention
According to an aspect of the present invention, there is provided an ozone concentration forecasting method including:
acquiring observation data and numerical simulation data of the time of start-up;
constructing characteristic data based on the observation data, the numerical simulation data and the characteristic engineering of the time of the start-up;
inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
According to another aspect of the present invention, there is provided an ozone concentration predicting device, including:
the acquisition module is used for acquiring observation data and numerical simulation data of the time of start-up;
the construction module is used for constructing characteristic data based on the observation data, the numerical simulation data and the characteristic engineering of the time of the start-up;
and the forecasting module is used for inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
According to another aspect of the present invention, there is provided an electronic apparatus including:
a processor; and
a memory for storing a program, wherein the program is stored in the memory,
wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the method of any of the above-described ozone concentration forecasting methods.
According to another aspect of the present invention, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of the above ozone concentration forecasting methods.
One or more technical schemes provided in the embodiment of the application can improve the accuracy of ozone prediction and the precision of ozone prediction on multiple time scales.
Drawings
Further details, features and advantages of the invention are disclosed in the following description of exemplary embodiments with reference to the accompanying drawings, in which:
FIG. 1 shows a flow chart of an ozone concentration prediction method according to an exemplary embodiment of the present invention;
FIG. 2 shows a flow chart of an ozone concentration prediction method according to an exemplary embodiment of the present invention;
FIG. 3 shows a flow chart of an ozone concentration prediction method according to an exemplary embodiment of the present invention;
FIG. 4 shows a schematic block diagram of an ozone concentration forecasting arrangement according to an exemplary embodiment of the present invention;
FIG. 5 illustrates a block diagram of an exemplary electronic device that can be used to implement an embodiment of the invention.
Detailed Description
Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and the embodiments of the present invention are illustrative only and are not intended to limit the scope of the present invention.
It should be understood that the various steps recited in the method embodiments of the present invention may be performed in a different order and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the invention is not limited in this respect.
The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions for other terms will be given in the following description. It should be noted that the terms "first", "second", and the like in the present invention are only used for distinguishing different devices, modules or units, and are not used for limiting the order or interdependence relationship of the functions performed by the devices, modules or units.
It is noted that references to "a", "an", and "the" modifications in the present invention are intended to be illustrative rather than limiting, and that those skilled in the art will recognize that reference to "one or more" unless the context clearly dictates otherwise.
The names of messages or information exchanged between devices in the embodiments of the present invention are for illustrative purposes only, and are not intended to limit the scope of the messages or information.
The embodiment of the invention provides an ozone concentration forecasting method, which can be realized by electronic equipment, wherein the electronic equipment can be a terminal or a server. As shown in the flow chart of the ozone concentration forecasting method in fig. 1, the processing flow of the method may include the following steps:
101, acquiring observation data and numerical simulation data of a start-up time;
102, constructing characteristic data based on observation data of the time of start-up, numerical simulation data and characteristic engineering;
the feature engineering refers to preprocessing data such as standardization and the like, and based on the existing data sets and features, new features are constructed by using subject background knowledge and mathematical modes and are used as the input of a model.
And 103, inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
Optionally, the observation data of the time-to-announce includes atmospheric pollutant data of the time-to-announce and meteorological observation data;
wherein the atmospheric pollutant data comprises NO2Concentration and ozone concentration;
the meteorological observation data comprises temperature, relative humidity, weather and wind speed.
Optionally, the numerical simulation data, the site interpolation data including numerical forecast data of the time of onset.
Optionally, the station interpolation data of the numerical forecast data of the starting time includes:
the system comprises first preset height temperature data, first preset height humidity data, second preset height wind speed data, boundary layer height data and ground short wave radiation data.
Optionally, the constructing the feature data based on the observation data of the time of the start-up, the numerical simulation data, and the feature engineering includes:
carrying out feature extraction on observation data, numerical simulation data and time feature data of the time of the start of report to obtain pollutant concentration feature data and meteorological element feature data;
and fusing the observation data of the time of the start of report, the numerical simulation data, the pollutant concentration characteristic data and the meteorological element characteristic data to generate a one-dimensional characteristic sequence.
Optionally, the contaminant concentration profile data comprises:
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
the difference between the ozone concentration at the time t and that before 24 hours;
reporting the maximum concentration average value, the maximum value and the minimum value of ozone within 8 hours in one week before the starting time t;
the time t of the start of the report is the maximum concentration difference of ozone within 8 hours before 24 hours.
Optionally, the meteorological element characteristic data includes:
the temperature difference value between the forecast time t and 24 hours ago, the temperature difference value between the forecast time t and 12 hours ago, the air pressure difference value between the forecast time t and 24 hours ago, the air pressure difference value between the forecast time t and 12 hours ago and the air pressure difference value between the forecast time t and 6 hours ago.
Optionally, the training process of the ozone concentration forecasting model comprises:
acquiring a training sample, wherein the training sample comprises sample input and a corresponding sample true value, the sample input is a characteristic sequence constructed by sample observation data and sample numerical simulation data through characteristic engineering, and the sample true value is corresponding ozone concentration observation data of the sample input;
inputting the training sample into an initial ozone concentration forecasting model to obtain a predicted value corresponding to the training sample;
and training initial parameters in the initial ozone concentration forecasting model based on the sample truth value, the predicted value and the cross entropy loss function.
Optionally, the ozone concentration prediction model is a Bi-directional long and short memory neural network Attention model, and the Bi-directional long and short memory neural network Attention model comprises an input layer, an LSTM layer, an Attention mechanism layer and a fully connected neural network layer;
inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time, wherein the ozone concentration forecasting data comprises the following steps:
inputting the characteristic data into an input layer of the Bi-LSTM-Attention model, and inputting the characteristic data into an LSTM layer through the input layer;
through an LSTM layer, giving implicit data and output data to each feature in the feature data, calculating weights of different features, and inputting an obtained weight matrix to an attention mechanism layer;
calculating the weight matrix and different characteristics through an attention mechanism layer, and inputting a calculation result into a full-connection neural network layer;
and the fully connected neural network layer determines ozone concentration forecast data of the time of the start-up through the calculation result.
In the embodiment of the invention, observation data and numerical simulation data of the time of start-up are obtained; constructing characteristic data based on observation data of the time of the start-up, numerical simulation data and characteristic engineering; inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time. Therefore, single-time observation data is combined with GFS (Global forecast System) meteorological reanalysis data forecast data, a plurality of features are constructed on the basis of initial input by using feature engineering skills, feature importance is evaluated in a training process by using an attention mechanism, different weights are given, and accuracy of ozone forecast on multiple time scales is improved. Moreover, the hourly ozone concentration is forecasted based on multi-source data based on observation data, so that the ozone forecasting precision can be improved, the ozone forecasting efficiency is improved, and more powerful support is provided for scientific prevention and control of ozone pollution.
The embodiment of the invention provides an ozone concentration forecasting method, which can be realized by electronic equipment, wherein the electronic equipment can be a terminal or a server. As shown in the flowchart of the ozone concentration forecasting method shown in fig. 2, the ozone concentration forecasting model in the embodiment of the present invention may be a Bi-LSTM-orientation (Bi-directional Long Short-Term Memory orientation, bidirectional Long Short-Term Memory network Attention mechanism) model, the Bi-LSTM-orientation model may include an input layer, an LSTM layer, an Attention mechanism layer, and a fully-connected neural network layer, and the processing flow of the method may include the following steps:
In a possible embodiment, when a user wants to obtain a predicted value of ozone concentration at a certain time (which may be referred to as a time-to-alarm), it is necessary to collect pollutant observation data of a single site, meteorological observation data of a site nearest to the pollutant observation site, and numerical simulation data, and in order to obtain a more accurate predicted value, the collected data has a time length of at least 1 year, and the observation data and the numerical simulation data of the time-to-alarm are obtained by collating the data.
Optionally, the observation data of the time-to-alarm may include atmospheric pollutant data of the time-to-alarm and meteorological observation data.
Wherein air pollutionThe object data may include NO2Concentration and ozone concentration; the meteorological observation data comprises temperature, relative humidity, weather and wind speed.
Optionally, the numerical simulation data, the site interpolation data including numerical forecast data of the time of onset.
Alternatively, the site interpolation data of the numerical forecast data of the attack time may include: the system comprises first preset height temperature data, first preset height humidity data, second preset height wind speed data, boundary layer height data and ground short wave radiation data.
Wherein, the first preset height can be 2 meters, and the second preset height can be 10 meters.
In one possible embodiment, the site interpolation data of the numerical prediction data may be derived from two sources: one is the forecast data output by the atmospheric numerical mode, such as WRF (Weather Research and Forecasting); a GFS data is provided for public numerical forecasting products, such as global numerical weather forecasting systems, that incorporate global computer models and variational analysis. Of course, the data source may also include other ways, which are not limited by the embodiment of the present invention.
Wherein the temporal characteristic data may include: the month of the hour to be forecasted, the information in the month (on the first day of the month), the information in the week (on the second day of the week), and the information in the hour (data forecasted on the second hour of 24 hours a day).
Wherein the contaminant concentration profile data may include: reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t; reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t; the difference between the ozone concentration at the time t and that before 24 hours; reporting the maximum concentration average value, the maximum value and the minimum value of ozone within 8 hours in one week before the starting time t; the time t of the start of the report is the maximum concentration difference of ozone within 8 hours before 24 hours.
Wherein, meteorological element characteristic data can include:
the temperature difference value between the forecast time t and 24 hours ago, the temperature difference value between the forecast time t and 12 hours ago, the air pressure difference value between the forecast time t and 24 hours ago, the air pressure difference value between the forecast time t and 12 hours ago and the air pressure difference value between the forecast time t and 6 hours ago.
In a possible implementation manner, after the observation data and the numerical simulation data of the attack time are acquired in step 201, feature extraction is performed on the observation data and the numerical simulation data of the attack time through the time feature, so as to obtain corresponding pollutant concentration feature data and meteorological element feature data. For example, the observation data of the alarm time obtained in step 201 may include observation data of a plurality of times within one week before the alarm time, and then, by using the time characteristics, observation data of a specific time, such as an average value, a maximum value, a minimum value, and the like of the ozone concentration of one week before the alarm time t, is extracted from the obtained observation data of a plurality of times. Through the mode, the data of the preset time can be conveniently and quickly extracted, and the hourly ozone concentration can be forecasted through the time characteristics accurate to the hour, so that the ozone forecasting precision can be improved, the ozone forecasting efficiency is improved, and more powerful support is provided for scientific prevention and control of ozone pollution.
And step 203, fusing the observation data of the time of start-up, the numerical simulation data, the pollutant concentration characteristic data and the meteorological element characteristic data to generate a one-dimensional characteristic sequence.
In a feasible implementation manner, the observation data, the numerical simulation data, the pollutant concentration characteristic data, and the meteorological element characteristic data of the time of arrival at the beginning obtained in the above steps may be spliced into a one-dimensional characteristic sequence according to a preset sequence, and the splicing sequence may be preset by a user, which is not described herein again in the embodiments of the present invention.
And step 204, training an initial ozone concentration forecasting model.
In a possible implementation manner, an initial ozone concentration prediction model is first constructed, parameters in the initial ozone concentration prediction model are initial parameters, then a training sample is obtained to train the initial model, and the training process may include the following steps 2041-:
step 2041, obtaining training samples.
The training sample comprises sample input and a corresponding sample true value, the sample input is sample observation data, sample numerical simulation data and a characteristic sequence constructed through the characteristic engineering of the step 202, and the sample true value is corresponding ozone concentration observation data of the sample input.
In a feasible implementation manner, the manner of obtaining the training sample may be to obtain observation data, numerical simulation data, and ozone concentration data of a plurality of sample times, and process the observation data and the numerical simulation data of the plurality of sample times with reference to the step 201 and the step 203 to obtain sample observation data and sample numerical simulation data corresponding to the sample times and a feature sequence constructed through feature engineering, and the specific processing process is not described herein again.
Step 2042, inputting the training sample into an initial ozone concentration forecasting model to obtain a predicted value corresponding to the training sample.
And 2043, training initial parameters in the initial ozone concentration forecasting model based on the sample truth value, the predicted value and the cross entropy loss function.
In a feasible implementation manner, based on a sample truth value, a predicted value and a preset cross entropy loss function, training initial parameters in an initial ozone concentration prediction model is performed until the loss function converges or the iteration times reach preset times, the training is stopped, and the current model is used as the trained ozone concentration prediction model.
It should be noted that the loss function used in the embodiment of the present invention is a cross-entropy loss function, which is only one available loss function, and other loss functions may also be used according to a user requirement, which is not limited in the embodiment of the present invention. In addition, the model training method in the embodiment of the present invention may refer to a currently common model training method, which is not described herein again.
And step 205, inputting the characteristic data into the trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
In a possible embodiment, the one-dimensional feature data obtained in the step 203 is input into a trained Bi-LSTM-orientation model, and the ozone concentration prediction data of the start time is obtained through the model, as shown in fig. 3, the process of obtaining the ozone concentration prediction data through the Bi-LSTM-orientation model may include the following steps 2061-:
step 2051, inputting the one-dimensional feature data obtained in the step 203 to an Input layer of the trained Bi-LSTM-orientation model, and inputting the feature data to an LSTM layer through the Input layer.
In a possible implementation manner, in order to satisfy the attention mechanism, dimension conversion is performed on the one-dimensional feature data, and the one-dimensional feature data is converted into three-dimensional feature data. The specific transformation method can be a transformation method commonly used in the prior art, and details thereof are not described in the embodiment of the present invention.
And step 2052, giving implicit data and output data to each feature in the feature data through the LSTM layer, calculating weights of different features, and inputting the obtained weight matrix to the attention mechanism layer.
In a feasible implementation manner, a bidirectional LSTM is adopted, example features in a feature sequence are input into an LSTM layer one by one, for each feature in the sequence, the LSTM gives a value of an implicit layer and an output value, weight influence of different sequence features is calculated through a memory mechanism of a neuron itself, a weight matrix corresponding to the sequence features is obtained, and the weight matrix is output to an attention mechanism attention layer.
And step 2053, calculating the weight matrix and different characteristics through the attention mechanism layer, and inputting the calculation result into the fully-connected neural network layer.
In one possible embodiment, the attention mechanism layer multiplies the implicit value and the output value of the feature, and the function of the attention mechanism layer is to evaluate the importance of the feature and further give different weights.
And step 2054, the fully-connected neural network layer determines ozone concentration forecast data of the time of the start-up according to the calculation result.
In a possible implementation mode, the multiplication result is processed by using a softmax function, each feature is allocated with a percentage after the softmax function, the sum of all feature importance is one hundred percent, and the ozone concentration forecast data of the time of the start is determined according to each feature and the corresponding feature importance. Different methods for determining the ozone concentration can be designed according to softmax, for example, a maximum value can be determined in the feature importance, and the ozone concentration of the feature corresponding to the maximum value is determined as ozone concentration forecast data of the starting time; or multiplying the concentration corresponding to the characteristic by the corresponding characteristic importance, adding all the obtained products, and determining the obtained numerical value as ozone concentration forecast data of the starting time. The embodiment of the present invention is not limited thereto.
In the embodiment of the invention, observation data and numerical simulation data of the time of start-up are obtained; constructing characteristic data based on observation data of the time of the start-up, numerical simulation data and characteristic engineering; inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time. Therefore, single-time observation data is combined with GFS meteorological reanalysis data forecast data, a plurality of features are constructed on the basis of initial input by using feature engineering skills, feature importance is evaluated in the training process by using an attention mechanism, different weights are given, and accuracy of ozone forecast on multiple time scales is improved. Moreover, the hourly ozone concentration is forecasted based on multi-source data based on observation data, so that the ozone forecasting precision can be improved, the ozone forecasting efficiency is improved, and more powerful support is provided for scientific prevention and control of ozone pollution.
The embodiment of the invention provides an ozone concentration forecasting device, which is used for realizing the ozone concentration forecasting method. As shown in the schematic block diagram of the ozone concentration forecasting apparatus shown in fig. 4, the ozone concentration forecasting apparatus 400 includes an obtaining module 410, a constructing module 420, and a forecasting module 430, wherein:
an obtaining module 410, configured to obtain observation data of a start-up time and numerical simulation data;
a construction module 420, configured to construct feature data based on the observation data of the start-up time, the numerical simulation data, and the feature engineering;
and the forecasting module 430 is configured to input the feature data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
Optionally, the observation data of the time-to-announce includes atmospheric pollutant data of the time-to-announce and meteorological observation data;
wherein the atmospheric pollutant data comprises NO2Concentration and ozone concentration;
the meteorological observation data comprises temperature, relative humidity, weather and wind speed.
Optionally, the numerical simulation data, the site interpolation data including numerical forecast data of the time of onset.
Optionally, the station interpolation data of the numerical forecast data of the starting time includes:
the system comprises first preset height temperature data, first preset height humidity data, second preset height wind speed data, boundary layer height data and ground short wave radiation data.
Optionally, a building module 420, configured to:
carrying out feature extraction on observation data, numerical simulation data and time feature data of the time of the start of report to obtain pollutant concentration feature data and meteorological element feature data;
and fusing the observation data of the time of the start of report, the numerical simulation data, the pollutant concentration characteristic data and the meteorological element characteristic data to generate a one-dimensional characteristic sequence.
Optionally, the contaminant concentration profile data comprises:
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
the difference between the ozone concentration at the time t and that before 24 hours;
reporting the maximum concentration average value, the maximum value and the minimum value of ozone within 8 hours in one week before the starting time t;
the time t of the start of the report is the maximum concentration difference of ozone within 8 hours before 24 hours.
Optionally, the meteorological element characteristic data includes:
the temperature difference value between the forecast time t and 24 hours ago, the temperature difference value between the forecast time t and 12 hours ago, the air pressure difference value between the forecast time t and 24 hours ago, the air pressure difference value between the forecast time t and 12 hours ago and the air pressure difference value between the forecast time t and 6 hours ago.
Optionally, a building module 420, configured to:
acquiring a training sample, wherein the training sample comprises sample input and a corresponding sample true value, the sample input comprises sample observation data, sample numerical simulation data and a characteristic sequence constructed through characteristic engineering, and the sample true value is corresponding ozone concentration observation data of the sample input;
inputting the training sample into an initial ozone concentration forecasting model to obtain a predicted value corresponding to the training sample;
and training initial parameters in the initial ozone concentration forecasting model based on the sample truth value, the predicted value and the cross entropy loss function.
Optionally, the ozone concentration prediction model is a Bi-directional long and short memory neural network Attention model, and the Bi-directional long and short memory neural network Attention model comprises an input layer, an LSTM layer, an Attention mechanism layer and a fully connected neural network layer;
a forecasting module 430 to:
inputting the characteristic data into an input layer of the Bi-LSTM-Attention model, and inputting the characteristic data into an LSTM layer through the input layer;
through an LSTM layer, giving implicit data and output data to each feature in the feature data, calculating weights of different features, and inputting an obtained weight matrix to an attention mechanism layer;
calculating the weight matrix and different characteristics through an attention mechanism layer, and inputting a calculation result into a full-connection neural network layer;
and the fully connected neural network layer determines ozone concentration forecast data of the time of the start-up through the calculation result.
In the embodiment of the invention, observation data and numerical simulation data of the time of start-up are obtained; constructing characteristic data based on observation data of the time of the start-up, numerical simulation data and characteristic engineering; inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time. Therefore, single-time observation data is combined with meteorological reanalysis data forecast data, a plurality of characteristics are constructed on the basis of initial input by using characteristic engineering skills, the importance of the characteristics is evaluated in the training process by using an attention mechanism, different weights are given, and the accuracy of ozone forecast on multiple time scales is improved. Moreover, the hourly ozone concentration is forecasted based on multi-source data based on observation data, so that the ozone forecasting precision can be improved, the ozone forecasting efficiency is improved, and more powerful support is provided for scientific prevention and control of ozone pollution.
An exemplary embodiment of the present invention also provides an electronic device including: at least one processor. And a memory communicatively coupled to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program, when executed by the at least one processor, is for causing the electronic device to perform a method according to an embodiment of the invention.
Exemplary embodiments of the present invention also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is operable to cause the computer to perform a method according to an embodiment of the present invention.
Exemplary embodiments of the present invention also provide a computer program product comprising a computer program, wherein the computer program is operative, when executed by a processor of a computer, to cause the computer to perform a method according to an embodiment of the present invention.
Referring to fig. 5, a block diagram of a structure of an electronic device 500, which may be a server or a client of the present invention, which is an example of a hardware device that may be applied to aspects of the present invention, will now be described. Electronic device is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
As shown in fig. 5, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM)502 or a computer program loaded from a storage unit 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The calculation unit 501, the ROM 502, and the RAM 503 are connected to each other by a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
A number of components in the electronic device 500 are connected to the I/O interface 505, including: an input unit 506, an output unit 507, a storage unit 508, and a communication unit 509. The input unit 506 may be any type of device capable of inputting information to the electronic device 500, and the input unit 506 may receive input numeric or character information and generate key signal inputs related to user settings and/or function controls of the electronic device. Output unit 507 may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, a video/audio output terminal, a vibrator, and/or a printer. The storage unit 508 may include, but is not limited to, a magnetic disk, an optical disk. The communication unit 509 allows the electronic device 500 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and/or chipsets, such as bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and/or the like.
The computing unit 501 may be a variety of general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 501 performs the respective methods and processes described above. For example, in some embodiments, the ozone concentration prediction method described above may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 500 via the ROM 502 and/or the communication unit 509. In some embodiments, the calculation unit 501 may be configured to perform the above-described ozone concentration prediction method by any other suitable means (e.g., by means of firmware).
Program code for implementing the methods of the present invention may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present invention, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
Claims (12)
1. An ozone concentration prediction method, comprising:
acquiring observation data and numerical simulation data of the time of start-up;
constructing characteristic data based on the observation data, the numerical simulation data and the characteristic engineering of the time of the start-up;
inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
2. The ozone concentration forecasting method according to claim 1, wherein the observation data of the attack time includes atmospheric pollutant data of the attack time and meteorological observation data;
wherein the atmospheric pollutant data comprises NO2Concentration and ozone concentration;
the meteorological observation data comprise temperature, relative humidity, weather and wind speed.
3. The method of claim 1, wherein the numerical simulation data includes site interpolation data of numerical prediction data of a time of start.
4. The ozone concentration forecasting method according to claim 3, wherein the site interpolation data of the numerical forecasting data of the attack time includes:
the system comprises first preset height temperature data, first preset height humidity data, second preset height wind speed data, boundary layer height data and ground short wave radiation data.
5. The method of claim 1, wherein the constructing feature data based on the observation data of the attack time, numerical simulation data, and feature engineering comprises:
carrying out feature extraction on the observation data, the numerical simulation data and the time feature data of the time to be reported to obtain pollutant concentration feature data and meteorological element feature data;
and fusing the observation data, the numerical simulation data, the pollutant concentration characteristic data and the meteorological element characteristic data of the time of the start-up to generate a one-dimensional characteristic sequence.
6. The ozone concentration forecasting method of claim 5, wherein the pollutant concentration characteristic data comprises:
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
reporting the average value, the maximum value and the minimum value of the ozone concentration one week before the starting time t;
the difference between the ozone concentration at the time t and that before 24 hours;
reporting the maximum concentration average value, the maximum value and the minimum value of ozone within 8 hours in one week before the starting time t;
the time t of the start of the report is the maximum concentration difference of ozone within 8 hours before 24 hours.
7. The ozone concentration forecasting method according to claim 5, wherein the meteorological element characteristic data includes:
the temperature difference value between the forecast time t and 24 hours ago, the temperature difference value between the forecast time t and 12 hours ago, the air pressure difference value between the forecast time t and 24 hours ago, the air pressure difference value between the forecast time t and 12 hours ago and the air pressure difference value between the forecast time t and 6 hours ago.
8. The ozone concentration forecasting method according to claim 1, wherein the training process of the ozone concentration forecasting model includes:
obtaining a training sample, wherein the training sample comprises a sample input and a corresponding sample true value, the sample input is a characteristic sequence constructed by sample observation data and sample numerical simulation data through characteristic engineering, and the sample true value is ozone concentration observation data corresponding to the sample input;
inputting the training sample into an initial ozone concentration forecasting model to obtain a predicted value corresponding to the training sample;
training initial parameters in the initial ozone concentration forecasting model based on the sample truth values, the predicted values and a cross entropy loss function.
9. The method of claim 1, wherein the ozone concentration prediction model is a Bi-directional long-short memory neural network Attention mechanism Bi-LSTM-Attention model, the Bi-LSTM-Attention model comprising an input layer, an LSTM layer, an Attention mechanism layer, and a fully connected neural network layer;
inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time, wherein the ozone concentration forecasting data comprises the following steps:
inputting the feature data to an input layer of the Bi-LSTM-Attention model, the feature data being input to the LSTM layer through the input layer;
giving implicit data and output data to each feature in the feature data through the LSTM layer, calculating weights of different features, and inputting an obtained weight matrix to the attention mechanism layer;
calculating the weight matrix and different characteristics through the attention mechanism layer, and inputting a calculation result into the fully-connected neural network layer;
and the fully-connected neural network layer determines ozone concentration forecast data of the alarm time according to the calculation result.
10. An ozone concentration predicting apparatus, comprising:
the acquisition module is used for acquiring observation data and numerical simulation data of the time of start-up;
the construction module is used for constructing characteristic data based on the observation data, the numerical simulation data and the characteristic engineering of the time of the start-up;
and the forecasting module is used for inputting the characteristic data into a pre-trained ozone concentration forecasting model to obtain ozone concentration forecasting data of the starting time.
11. An electronic device, comprising:
a processor; and
a memory for storing a program, wherein the program is stored in the memory,
wherein the program comprises instructions which, when executed by the processor, cause the processor to carry out the method according to any one of claims 1-9.
12. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111583015.1A CN114298389A (en) | 2021-12-22 | 2021-12-22 | Ozone concentration forecasting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111583015.1A CN114298389A (en) | 2021-12-22 | 2021-12-22 | Ozone concentration forecasting method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114298389A true CN114298389A (en) | 2022-04-08 |
Family
ID=80970366
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111583015.1A Pending CN114298389A (en) | 2021-12-22 | 2021-12-22 | Ozone concentration forecasting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114298389A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117291319A (en) * | 2023-11-27 | 2023-12-26 | 新禾数字科技(无锡)有限公司 | O based on machine learning 3 Residual prediction method |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109061776A (en) * | 2018-10-08 | 2018-12-21 | 中国气象局广州热带海洋气象研究所（广东省气象科学研究所） | One kind closing on numerical weather forecast system and method in short-term |
| CN109541730A (en) * | 2018-11-23 | 2019-03-29 | 长三角环境气象预报预警中心（上海市环境气象中心） | A kind of method and apparatus of pollutant prediction |
| CN110261547A (en) * | 2019-07-04 | 2019-09-20 | 北京思路创新科技有限公司 | A kind of Urban Air Pollution Methods and equipment |
| CN110334732A (en) * | 2019-05-20 | 2019-10-15 | 北京思路创新科技有限公司 | A kind of Urban Air Pollution Methods and device based on machine learning |
| KR102063358B1 (en) * | 2019-10-22 | 2020-03-02 | (주)한국해양기상기술 | Learning method and testing method for generating high-resolution weather and climate data, and learning device and testing device using the same |
| US20200082041A1 (en) * | 2018-09-06 | 2020-03-12 | Terrafuse, Inc. | Estimating physical parameters of a physical system based on a spatial-temporal emulator |
| CN111612254A (en) * | 2020-05-22 | 2020-09-01 | 中国科学院合肥物质科学研究院 | Road motor vehicle exhaust emission prediction method based on improved attention bidirectional long-short term memory network |
| CN111832814A (en) * | 2020-07-01 | 2020-10-27 | 北京工商大学 | Air pollutant concentration prediction method based on graph attention machine mechanism |
| CN112308281A (en) * | 2019-11-12 | 2021-02-02 | 北京嘉韵楷达气象科技有限公司 | Temperature information prediction method and device |
| CN112766549A (en) * | 2021-01-07 | 2021-05-07 | 清华大学 | Air pollutant concentration forecasting method and device and storage medium |
| CN112862090A (en) * | 2021-01-22 | 2021-05-28 | 南开大学 | Air temperature forecasting method based on deep space-time neural network |
| CN113011106A (en) * | 2021-04-15 | 2021-06-22 | 中国气象局公共气象服务中心(国家预警信息发布中心) | Multi-value mode integrated forecasting method and device |
| CN113657042A (en) * | 2021-09-01 | 2021-11-16 | 中国人民解放军国防科技大学 | Air quality parameter prediction method based on multi-scale attention mechanism |
-
2021
- 2021-12-22 CN CN202111583015.1A patent/CN114298389A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200082041A1 (en) * | 2018-09-06 | 2020-03-12 | Terrafuse, Inc. | Estimating physical parameters of a physical system based on a spatial-temporal emulator |
| CN109061776A (en) * | 2018-10-08 | 2018-12-21 | 中国气象局广州热带海洋气象研究所（广东省气象科学研究所） | One kind closing on numerical weather forecast system and method in short-term |
| CN109541730A (en) * | 2018-11-23 | 2019-03-29 | 长三角环境气象预报预警中心（上海市环境气象中心） | A kind of method and apparatus of pollutant prediction |
| CN110334732A (en) * | 2019-05-20 | 2019-10-15 | 北京思路创新科技有限公司 | A kind of Urban Air Pollution Methods and device based on machine learning |
| CN110261547A (en) * | 2019-07-04 | 2019-09-20 | 北京思路创新科技有限公司 | A kind of Urban Air Pollution Methods and equipment |
| KR102063358B1 (en) * | 2019-10-22 | 2020-03-02 | (주)한국해양기상기술 | Learning method and testing method for generating high-resolution weather and climate data, and learning device and testing device using the same |
| CN112308281A (en) * | 2019-11-12 | 2021-02-02 | 北京嘉韵楷达气象科技有限公司 | Temperature information prediction method and device |
| CN111612254A (en) * | 2020-05-22 | 2020-09-01 | 中国科学院合肥物质科学研究院 | Road motor vehicle exhaust emission prediction method based on improved attention bidirectional long-short term memory network |
| CN111832814A (en) * | 2020-07-01 | 2020-10-27 | 北京工商大学 | Air pollutant concentration prediction method based on graph attention machine mechanism |
| CN112766549A (en) * | 2021-01-07 | 2021-05-07 | 清华大学 | Air pollutant concentration forecasting method and device and storage medium |
| CN112862090A (en) * | 2021-01-22 | 2021-05-28 | 南开大学 | Air temperature forecasting method based on deep space-time neural network |
| CN113011106A (en) * | 2021-04-15 | 2021-06-22 | 中国气象局公共气象服务中心(国家预警信息发布中心) | Multi-value mode integrated forecasting method and device |
| CN113657042A (en) * | 2021-09-01 | 2021-11-16 | 中国人民解放军国防科技大学 | Air quality parameter prediction method based on multi-scale attention mechanism |
Non-Patent Citations (3)
| Title |
|---|
| SEASKY_STEVEN: ""文本分类实战—Bi-LSTM+Attention模型", 《HTTPS://BLOG.CSDN.NET/SEASKY_STEVEN/ARTICLE/DETAILS/104629389》 * |
| 陈玉涛: "基于深度学习的空气质量预测研究", 《中国优秀博硕士学位论文全文数据库（硕士） 工程科技I辑》 * |
| 高敬鹏: "《深度学习 卷积神经网络技术与实践》", 30 June 2020 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117291319A (en) * | 2023-11-27 | 2023-12-26 | 新禾数字科技(无锡)有限公司 | O based on machine learning 3 Residual prediction method |
| CN117291319B (en) * | 2023-11-27 | 2024-02-20 | 新禾数字科技(无锡)有限公司 | O based on machine learning 3 Residual prediction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110766142A (en) | Model generation method and device | |
| CN112489677A (en) | Voice endpoint detection method, device, equipment and medium based on neural network | |
| JP7446359B2 (en) | Traffic data prediction method, traffic data prediction device, electronic equipment, storage medium, computer program product and computer program | |
| CN113888381A (en) | Pollutant concentration forecasting method and device | |
| CN113554280B (en) | Training method, device, equipment and storage medium of power grid system scheduling model | |
| CN116684330A (en) | Traffic prediction method, device, equipment and storage medium based on artificial intelligence | |
| CN114298389A (en) | Ozone concentration forecasting method and device | |
| CN114972877A (en) | Image classification model training method and device and electronic equipment | |
| CN112801357B (en) | Solar radiation quantity prediction method, device, equipment and storage medium | |
| CN116756522B (en) | Probability forecasting method and device, storage medium and electronic equipment | |
| CN113506023A (en) | Working behavior data analysis method, device, equipment and storage medium | |
| CN117252564A (en) | Resource scheduling method, device, computer equipment and storage medium | |
| CN115453661B (en) | Weather forecasting method, weather forecasting device, weather forecasting equipment and storage medium | |
| CN116777646A (en) | Artificial intelligence-based risk identification method, apparatus, device and storage medium | |
| CN116843483A (en) | Vehicle insurance claim settlement method, device, computer equipment and storage medium | |
| CN114282658B (en) | Method, device and medium for analyzing and predicting flow sequence | |
| CN115952921A (en) | Photovoltaic energy power prediction method and device, electronic equipment and storage medium | |
| CN115986729A (en) | Solving method, device and equipment based on source network load storage data driving model | |
| JP2024507602A (en) | Data processing methods and methods for training predictive models | |
| CN114580740A (en) | Bus load prediction method and device, electronic equipment and storage medium | |
| CN113851226A (en) | Infectious disease epidemic situation prediction method and device and electronic equipment | |
| CN114051057A (en) | Method and device for determining queuing time of cloud equipment, electronic equipment and medium | |
| CN113704314A (en) | Data analysis method and device, electronic equipment and storage medium | |
| CN114067805A (en) | Method and device for training voiceprint recognition model and voiceprint recognition | |
| CN113361621A (en) | Method and apparatus for training a model |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220408 |
|
| RJ01 | Rejection of invention patent application after publication |