CN109492758A - Data forecasting procedure and system - Google Patents
Data forecasting procedure and system Download PDFInfo
- Publication number
- CN109492758A CN109492758A CN201811441091.7A CN201811441091A CN109492758A CN 109492758 A CN109492758 A CN 109492758A CN 201811441091 A CN201811441091 A CN 201811441091A CN 109492758 A CN109492758 A CN 109492758A
- Authority
- CN
- China
- Prior art keywords
- data
- extension
- observational
- preset
- model
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Environmental & Geological Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Ecology (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Biodiversity & Conservation Biology (AREA)
- Data Mining & Analysis (AREA)
- Environmental Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Atmospheric Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of data forecasting procedure and systems, are related to the technical field of weather forecast, comprising: obtain the corresponding observational data in place to be measured, carry out Data Assimilation using preset assimilation model based on the observational data, obtain initial fields；Using preset extension model using the initial fields as input carry out data extension, the extension observational data of subsequent time is obtained；It is integrated using the extension observational data as input carry out data using preset integrated model, obtain integrated data, by carrying out Data Assimilation using preset assimilation model based on the observational data, obtain initial fields, establish the higher initial fields of precision, by using preset extension model using the initial fields as input carry out data extension, obtain the extension observational data of subsequent time, it improves and extends accuracy, by integrated using the extension observational data as input carry out data using preset integrated model, obtain integrated data, finally obtain the higher integrated data of accuracy.
Description
Technical field
The present invention relates to weather forecast technical fields, more particularly, to a kind of data forecasting procedure and system.
Background technique
With the development of the technology of weather forecast, people constantly pursued more accurate forecast and in the periods more remote
Forecast, currently, traditional forecast system is to take single-mode forecasting system to calculate according to the corresponding observational data in prediction place
Then initial fields input stretch system is observed Spatial Multi-Dimensional and converts to obtain extension data, but conventional method by initial fields out
The initial fields precision not only obtained is low, and forecast system itself error is larger, and will not extend data and integrate.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of data forecasting procedure and system, not only with solution
The initial fields precision arrived is low, and forecast system itself error is larger, and will not extend data and carry out integrated technical problem.
In a first aspect, the embodiment of the invention provides a kind of data forecasting procedures, comprising:
The corresponding observational data in place to be measured is obtained, data is carried out using preset assimilation model based on the observational data
Assimilation, obtains initial fields；
Using preset extension model using the initial fields as input carry out data extension, the extension of subsequent time is obtained
Observational data；
It is integrated using the extension observational data as input carry out data using preset integrated model, obtain integrated money
Material.
With reference to first aspect, the embodiment of the invention provides the first possible embodiments of first aspect, wherein institute
It states and Data Assimilation is carried out using preset assimilation model based on the observational data, obtaining initial fields includes:
The observational data is transformed into the corresponding observation space of the observational data according to Observation Operators, obtains observation number
According to；
Data Assimilation is carried out using the observation data as input using preset assimilation model, obtains initial fields.
With reference to first aspect, the embodiment of the invention provides second of possible embodiment of first aspect, wherein root
The observational data is transformed into the corresponding observation space of the observational data according to Observation Operators, before obtaining observation data, also
Include:
The data that default observation condition is not met in the observational data are rejected, to the observational data rejected after operating
Carry out deviation revision.
With reference to first aspect, the embodiment of the invention provides the third possible embodiment of first aspect, wherein institute
It states using preset extension model and obtains the extension observation money of subsequent time using the initial fields as input carry out data extension
Material, comprising:
Initial fields are divided into N number of rank according to image attributes, obtain ranked data corresponding to current time each rank,
Wherein N is more than or equal to 1；
Using preset extension model using the ranked data of each rank as input carry out data extension, obtain down
Extension observational data corresponding to one moment each rank；
Extension data corresponding to N number of rank is synthesized, the extension observational data of subsequent time is reconstructed.
With reference to first aspect, the embodiment of the invention provides the 4th kind of possible embodiment of first aspect, wherein institute
It states and initial fields is divided into N number of rank according to image attributes, obtain ranked data corresponding to current time each rank, comprising:
Obtain initial fields；
The initial fields are subjected to Wavelet image decomposition, obtain ranked data corresponding to current time each rank.
With reference to first aspect, the embodiment of the invention provides the 5th kind of possible embodiments of first aspect, wherein phase
It answers, it is described that extension data corresponding to N number of rank is synthesized, reconstruct the extension observational data of subsequent time, comprising:
Extension data corresponding to N number of rank is synthesized using wavelet inverse transformation, reconstructs the extension of subsequent time
Observational data.
With reference to first aspect, the embodiment of the invention provides the 6th kind of possible embodiments of first aspect, wherein three
A model includes: deep neural network.
With reference to first aspect, the embodiment of the invention provides the 7th kind of possible embodiment of first aspect, wherein institute
The weight coefficient for stating deep neural network is to determine that the weight coefficient makes according to preset tactful neural network output result
The network state Maximum Value of the deep neural network.
With reference to first aspect, the embodiment of the invention provides the 8th kind of possible embodiment of first aspect, wherein institute
Stating tactful neural network includes: value assessment network and control neural network, and the value assessment network is for assessing the depth
The value of the network state of neural network is spent, value that the control neural network export based on the value assessment network and in advance
If adjustable strategies adjust the output result of the tactful neural network.
Second aspect, the embodiment of the present invention also provide a kind of data forecast system, comprising: Data Assimilation module, data are prolonged
Stretch module and data integration module；
The Data Assimilation module is for obtaining the corresponding observational data in place to be measured, using preset Optimized model with institute
Observational data is stated as input and carries out Data Assimilation, obtains initial fields；
The data extension of module is prolonged using the preset Optimized model using the initial fields as input carry out data
It stretches, obtains the extension observational data of subsequent time；
The data integration module is carried out using the preset Optimized model using the extension observational data as input
Data is integrated, obtains integrated data.
Include: to obtain the corresponding observational data in place to be measured in the method for the embodiment of the present invention, is based on the observational data
Data Assimilation is carried out using preset assimilation model, obtains initial fields；Using preset extension model using the initial fields as
Carry out data extension is inputted, the extension observational data of subsequent time is obtained；It is observed using preset integrated model with the extension
Data is integrated as input carry out data, obtains integrated data, by utilizing preset assimilation model based on the observational data
Data Assimilation is carried out, initial fields is obtained, establishes the higher initial fields of precision, by utilizing preset extension model with described initial
Field obtains the extension observational data of subsequent time as input carry out data extension, improves and extends accuracy, default by utilizing
Integrated model it is integrated using the extension observational data as input carry out data, obtain integrated data, finally obtain accuracy
Higher integrated data.
Other features and advantages of the present invention will illustrate in the following description, also, partly become from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention are in specification, claims
And specifically noted structure is achieved and obtained in attached drawing.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, is also possible to obtain other drawings based on these drawings.
Fig. 1 is data forecasting procedure flow chart provided in an embodiment of the present invention；
Fig. 2 is step S101 implementation flow chart provided in an embodiment of the present invention；
Fig. 3 is step S102 implementation flow chart provided in an embodiment of the present invention；
Fig. 4 is data forecast system module diagram provided in an embodiment of the present invention.
Icon:
01- Data Assimilation module；02- data extension of module；03- data integration module.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention
Technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
Embodiment one:
According to embodiments of the present invention, a kind of data forecasting procedure embodiment is provided, it should be noted that in the stream of attached drawing
The step of journey illustrates can execute in a computer system such as a set of computer executable instructions, although also, flowing
Logical order is shown in journey figure, but in some cases, it can be to be different from shown or described by sequence execution herein
The step of.
Fig. 1 is data forecast system method according to an embodiment of the present invention, as shown in Figure 1, this method comprises the following steps:
Step S101 obtains the corresponding observational data in place to be measured, utilizes preset assimilation mould based on the observational data
Type carries out Data Assimilation, obtains initial fields；
Wherein, at model, three models include: deep neural network for the assimilation model, extension model sum aggregate.It is described
The weight coefficient of deep neural network is to determine that the weight coefficient makes institute according to preset tactful neural network output result
State the network state Maximum Value of deep neural network.Wherein the tactful neural network includes: value assessment network and control
Neural network, the value assessment network are used to assess the value of the network state of the deep neural network, the control mind
The value and default adjustable strategies exported through network based on the value assessment network adjusts the output of the tactful neural network
As a result.Tactful neural network is depth CNN neural network model, provides the adjustable strategies of valence deep neural network, provides depth
The value estimations value of neural network adjustment evaluates the value of deep neural network adjustment.Tactful neural network is mainly
By the value of value assessment network query function, the weight coefficient of percentage regulation neural network is removed.It is when being worth increase, then positive to adjust
The system of whole deep neural network moves the value of the value assessment network of the extension value of deep neural network toward direction is increased
It is dynamic.If value reduces, the weight coefficient of reversed percentage regulation neural network makes the value of the extension value of deep neural network
The value for assessing network is mobile toward opposite direction.Control neural network removes the weight coefficient of adjustment neural network, the depth nerve made
The value of network develops toward incremental direction, when the Maximum Value of value assessment network, then terminates network.The control nerve net
The default adjustable strategies of network are that RNN clustered control network is sent, and the RNN clustered control network is used to comment based on the value
The value of network output is estimated to the control neural network output adjustment strategy.RNN clustered control network provides a series of essence
Dimension parameter is spent, provides the policing parameter of adjustable strategies to tactful neural network, RNN clustered control network model is deep using cluster
RNN neural network model is spent, there is forward and reverse to adjust strategy, effectively estimated according to the value that tactful neural network model provides
Evaluation, the adjustment direction of controlling depth neural network weight and bias.The embodiment of the present invention passes through training sample by setting
Model after training, respectively in Typical data forecast system, assimilation step, data extend step and integrated step is changed
Into, depending on the composition of specific deep neural network can be according to actual conditions, the embodiment of the present invention only provides a kind of feasible pattern,
It does not limit this.
In embodiments of the present invention, the observational data includes the ambient field and image of observation place, and needs to image
It is pre-processed, obtains the corresponding picture information in place to be predicted and 0,1 binaryzation data.Read control parameter file, setting
Non-mode amount is converted mode amount by the operating parameters such as the assimilation time window of mode, be input in preset assimilation model into
Row Data Assimilation improves the precision of the initial fields of generation.As shown in Fig. 2, it is based on step S101, it is described to be based on the observational data
Data Assimilation is carried out using preset assimilation model, obtaining initial fields includes:
The observational data is transformed into the corresponding observation space of the observational data according to Observation Operators by step S201,
Obtain observation data；
Step S202 carries out Data Assimilation using the observation data as input using preset assimilation model, obtains just
Beginning field.
In embodiments of the present invention, the weather station observational data and moonscope data for including to observational data turn
Change, atmospheric outline is transformed into the corresponding observation space of observational data according to Observation Operators, obtains the corresponding observation of observational data
Conversion data.Wherein, Observation Operators use radiative transmission mode when converting to moonscope data.For picture information,
The corresponding image observation operator of picture information is established, picture information is converted, according to image observation operator by atmospheric outline
It is transformed into the corresponding observation space of picture information, obtains the corresponding picture conversion data of picture information.For binaryzation data, build
The vertical corresponding binaryzation Observation Operators of binaryzation data, are transformed into binaryzation data for atmospheric outline according to binaryzation Observation Operators
Corresponding observation space obtains the corresponding binaryzation conversion data of binaryzation data.Wherein according to Observation Operators by the observation
Data is transformed into the corresponding observation space of the observational data, obtains observing before data, further includes:
The data that default observation condition is not met in the observational data are rejected, to the observational data rejected after operating
Carry out deviation revision.
In embodiments of the present invention, the purpose of revision is to carry out data sieve to picture information and binaryzation data
Choosing.Above-mentioned default observation condition is that data format meets observation format, and climate state is without departing from preset climate state range,
Also to meet horizontal and vertical continuity requirement, and be located within assimilation time window.It is not met in rejecting observational data default
When the data of observation condition, Data selection is carried out to original weather station observational data and moonscope data, when filtering out assimilation
Between Shock absorption effect in window, the method step includes: climate state extreme value range validity check, from observation data
It is middle to reject the abnormal point deviateed other than Climatological pre-determined distance；Format and logical check reject in observational data and illegally observe lattice
Formula data reject missing point present in observational data；Horizontal and vertical continuity check will not be met continuous in observational data
Property require point deletion；The not data in assimilation time window in observational data is rejected in time consistency inspection, specifically used
Image processing algorithm can be according to actual conditions depending on, which is not limited by the present invention.
Step S102 is obtained next using preset extension model using the initial fields as input carry out data extension
The extension observational data at moment；
In embodiments of the present invention, using preset extension model, the weight system of model can specifically be extended by configuring
System, which extends, obtains the extension data under residual error minimum, is based ultimately upon the extension observation chart that extension data reconstructs subsequent time
As data, improves and extend accuracy.Based on step S102, as shown in figure 3, being made using preset extension model with the initial fields
To input carry out data extension, the extension observational data of subsequent time is obtained, comprising:
Initial fields are divided into N number of rank according to image attributes, obtained corresponding to current time each rank by step S301
Ranked data, wherein N is more than or equal to 1；
In embodiments of the present invention, original observed image data may include: that satellite data, Radar Data and automatic Weather Station are seen
Survey data etc..Image attributes can be down to image resolution ratio or graphical rule etc..
Step S302 is prolonged using preset extension model using the ranked data of each rank as input carry out data
It stretches, obtains extension observational data corresponding to each rank of subsequent time；
In embodiments of the present invention, the model that extends is using the weight coefficient of preset tactful neural network output as defeated
Enter, the weight coefficient makes the network state Maximum Value for extending model, in embodiments of the present invention, the network state
Value is after the error update of deep neural network, and the state of deep neural network can change, and value network can be according to depth
The state of degree neural network is evaluated, and to a value estimations value of the state of deep neural network out, is equivalent to depth
The negative feedback mechanism of neural network, in addition, perhaps this state is positive, be may is that when system meeting generating state changes
It is reversed, can be evaluated by value network come.If it is forward direction, note that intensified learning is adjusted according to direction.
Otherwise note that intensified learning according to reversely having adjusted.The adjustable strategies of the network state value, most with final output model
For the purpose of the figure of merit, such as in numerical weather forecasting field, the image obtained in forecast data including various ways is inputted, by big
Gas neural network model exports the highest integrated data of accuracy, it may be said that the network state valence of the atmosphere neural network model
Value is maximum；Extend the state estimation that model is used to save the classification image document of each rank.Extend model and passes through a depth
A neural network for spending neural network classification (a certain rank) image, by certain strategy come percentage regulation neural network
Weight coefficient, the best estimate of (a certain rank) image is classified, as classification (certain level-one under a certain resolution ratio
Clear image not).Deep neural network describes the state of atmosphere using full connection type, is adjusted by weight and bias
Weight, extension obtain the image under a certain rank.
Step S303 synthesizes extension data corresponding to N number of rank, reconstructs the extension observation money of subsequent time
Material.
The embodiment of the present invention is worked as by the way that original observed image data is divided into N number of rank according to image attributes first
Then classification image document corresponding to preceding moment each rank utilizes institute of the preset deep neural network based on each rank
It states classification image document to be extended, obtains extension data corresponding to each rank of subsequent time, it finally can be to N number of rank
Corresponding extension data carries out image synthesis, reconstructs the extension observed image data of subsequent time.Extended model can be with
It is atmospheric depth neural network, such as: the embodiment of the present invention can be using deep neural network based on described in each rank points
Grade image document is extended, and extension data corresponding to each rank of subsequent time is obtained, due to the deep neural network
Using the weight coefficient of preset tactful neural network output as input, the weight coefficient can make the deep neural network
Network state Maximum Value, the embodiment of the present invention can using make deep neural network network state Maximum Value weight
Coefficient configurable deep neural network, and provided using extending to obtain the extension under residual error minimum with the deep neural network postponed
Material is based ultimately upon the extension observed image data that extension data reconstructs subsequent time, improves and extends accuracy.Based on step
S301, the embodiment of the invention also provides a kind of feasible patterns, and initial fields are divided into N number of rank according to image attributes, are worked as
Ranked data corresponding to preceding moment each rank, comprising: obtain initial fields；The initial fields are subjected to Wavelet image decomposition,
Obtain ranked data corresponding to current time each rank.
Wherein, Wavelet image decomposition method can provide the small echo classification image of different resolution or different scale, will divide
The small echo of different resolution or different scale after solution is classified image, the different stage image after decomposing as Wavelet image.
Correspondingly, it is based on step S303, and it is described that extension data corresponding to N number of rank is synthesized, it reconstructs next
The extension observational data at moment, comprising: extension data corresponding to N number of rank is synthesized using wavelet inverse transformation, is reconstructed
The extension observational data of subsequent time out.
In this step, can use the inverse transformations such as wavelet inverse transformation mode to extension data corresponding to N number of rank into
The synthesis of row image, reconstructs the extension observed image data of subsequent time, the extension observed image data of subsequent time can be with structure
At a clear image.
Step S103, it is integrated using the extension observational data as input carry out data using preset integrated model, it obtains
To integrated data.
In embodiments of the present invention, it is input in preset integrated model by the way that observational data will be extended, instead of traditional
Single integrated approach, a variety of extensions observation money is integrated, and pass through the multi-mode weight proportion function of integrated model, final
To the higher integrated data of accuracy.
Embodiment two:
The embodiment of the invention also provides a kind of data forecast system, data forecast system is mainly used for executing of the invention real
Data forecasting procedure provided by an above content is applied, specific Jie is done to data forecast system provided in an embodiment of the present invention below
It continues.
Fig. 4 is a kind of schematic diagram of data forecast system according to an embodiment of the present invention, as shown in figure 4, the data is forecast
System specifically includes that Data Assimilation module 01, data extension of module 02 and data integration module 03；
The Data Assimilation module 01 for obtaining the corresponding observational data in place to be measured, using preset Optimized model with
The observational data carries out Data Assimilation as input, obtains initial fields；
The data extension of module 02 is using the preset Optimized model using the initial fields as input carry out data
Extend, obtains the extension observational data of subsequent time；
The data integration module 03 using the preset Optimized model using the extension observational data as input into
Row data is integrated, obtains integrated data.
The technical effect and preceding method embodiment phase of system provided by the embodiment of the present invention, realization principle and generation
Together, to briefly describe, system embodiment part does not refer to place, can refer to corresponding contents in preceding method embodiment.
In addition, in the description of the embodiment of the present invention unless specifically defined or limited otherwise, term " installation ", " phase
Even ", " connection " shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or be integrally connected；It can
To be mechanical connection, it is also possible to be electrically connected；It can be directly connected, can also can be indirectly connected through an intermediary
Connection inside two elements.For the ordinary skill in the art, above-mentioned term can be understood at this with concrete condition
Concrete meaning in invention.
In the description of the present invention, it should be noted that term " center ", "upper", "lower", "left", "right", "vertical",
The orientation or positional relationship of the instructions such as "horizontal", "inner", "outside" be based on the orientation or positional relationship shown in the drawings, merely to
Convenient for description the present invention and simplify description, rather than the device or element of indication or suggestion meaning must have a particular orientation,
It is constructed and operated in a specific orientation, therefore is not considered as limiting the invention.In addition, term " first ", " second ",
" third " is used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance.
In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, it can be with
It realizes by another way.System embodiment described above is only schematical, for example, the division of the unit,
Only a kind of logical function partition, there may be another division manner in actual implementation, in another example, multiple units or components can
To combine or be desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or beg for
The mutual coupling, direct-coupling or communication connection of opinion can be through some communication interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in the executable non-volatile computer-readable storage medium of a processor.Based on this understanding, of the invention
Technical solution substantially the part of the part that contributes to existing technology or the technical solution can be with software in other words
The form of product embodies, which is stored in a storage medium, including some instructions use so that
One computer equipment (can be personal computer, server or the network equipment etc.) executes each embodiment institute of the present invention
State all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-
Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can be with
Store the medium of program code.
Finally, it should be noted that embodiment described above, only a specific embodiment of the invention, to illustrate the present invention
Technical solution, rather than its limitations, scope of protection of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair
It is bright to be described in detail, those skilled in the art should understand that: anyone skilled in the art
In the technical scope disclosed by the present invention, it can still modify to technical solution documented by previous embodiment or can be light
It is readily conceivable that variation or equivalent replacement of some of the technical features；And these modifications, variation or replacement, do not make
The essence of corresponding technical solution is detached from the spirit and scope of technical solution of the embodiment of the present invention, should all cover in protection of the invention
Within the scope of.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. a kind of data forecasting procedure characterized by comprising
The corresponding observational data in place to be measured is obtained, it is same using preset assimilation model progress data based on the observational data
Change, obtains initial fields；
Using preset extension model using the initial fields as input carry out data extension, the extension observation of subsequent time is obtained
Data；
It is integrated using the extension observational data as input carry out data using preset integrated model, obtain integrated data.
2. data forecasting procedure according to claim 1, which is characterized in that described utilized based on the observational data is preset
Assimilation model carry out Data Assimilation, obtaining initial fields includes:
The observational data is transformed into the corresponding observation space of the observational data according to Observation Operators, obtains observation data；
Data Assimilation is carried out using the observation data as input using preset assimilation model, obtains initial fields.
3. data forecasting procedure according to claim 2, which is characterized in that turned the observational data according to Observation Operators
Change observation space corresponding to the observational data, obtain observing before data, further includes:
The data for not meeting default observation condition in the observational data are rejected, the observational data rejected after operating is carried out
Deviation revision.
4. data forecasting procedure according to claim 1, which is characterized in that described to utilize preset extension model with described
Initial fields obtain the extension observational data of subsequent time as input carry out data extension, comprising:
Initial fields are divided into N number of rank according to image attributes, obtain ranked data corresponding to current time each rank, wherein
N is more than or equal to 1；
Using preset extension model using the ranked data of each rank as input carry out data extension, lower a period of time is obtained
Carve extension observational data corresponding to each rank；
Extension data corresponding to N number of rank is synthesized, the extension observational data of subsequent time is reconstructed.
5. data forecasting procedure according to claim 4, which is characterized in that described to be divided into initial fields according to image attributes
N number of rank obtains ranked data corresponding to current time each rank, comprising:
Obtain initial fields；
The initial fields are subjected to Wavelet image decomposition, obtain ranked data corresponding to current time each rank.
6. data forecasting procedure according to claim 5, which is characterized in that correspondingly, described to corresponding to N number of rank
Extension data is synthesized, and the extension observational data of subsequent time is reconstructed, comprising:
Extension data corresponding to N number of rank is synthesized using wavelet inverse transformation, reconstructs the extension observation of subsequent time
Data.
7. data forecasting procedure according to claim 1, which is characterized in that three models include: deep neural network.
8. data forecasting procedure according to claim 7, which is characterized in that the weight coefficient of the deep neural network is
It is determined according to preset tactful neural network output result, the weight coefficient makes the network state of the deep neural network
Maximum Value.
9. data forecasting procedure according to claim 8, which is characterized in that the strategy neural network includes: that value is commented
Estimate network and control neural network, the value assessment network is used to assess the valence of the network state of the deep neural network
Value, value and the adjustment of default the adjustable strategies strategy that the control neural network export based on the value assessment network are refreshing
Output result through network.
10. a kind of data forecast system characterized by comprising Data Assimilation module, data extension of module and data integrate mould
Block；
The Data Assimilation module is for obtaining the corresponding observational data in place to be measured, using preset Optimized model with the sight
Survey data carries out Data Assimilation as input, obtains initial fields；
The data extension of module, using the initial fields as input carry out data extension, is obtained using the preset Optimized model
To the extension observational data of subsequent time；
The data integration module is using the preset Optimized model using the extension observational data as input carry out data
It is integrated, obtain integrated data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811441091.7A CN109492758A (en) | 2018-11-28 | 2018-11-28 | Data forecasting procedure and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811441091.7A CN109492758A (en) | 2018-11-28 | 2018-11-28 | Data forecasting procedure and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109492758A true CN109492758A (en) | 2019-03-19 |
Family
ID=65698552
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811441091.7A Pending CN109492758A (en) | 2018-11-28 | 2018-11-28 | Data forecasting procedure and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492758A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105976020A (en) * | 2016-04-28 | 2016-09-28 | 华北电力大学 | Network flow prediction method considering wavelet cross-layer correlations |
| CN106371155A (en) * | 2016-08-25 | 2017-02-01 | 华南师范大学 | A weather forecast method and system based on big data and analysis fields |
| CN106597574A (en) * | 2016-12-30 | 2017-04-26 | 重庆邮电大学 | Weather temperature prediction method and device based on time-varying cloud model |
| CN106960253A (en) * | 2017-03-10 | 2017-07-18 | 国网浙江省电力公司经济技术研究院 | A kind of regional middle-term electricity demand forecasting method for considering great social activities influence |
| CN108182492A (en) * | 2017-12-29 | 2018-06-19 | 中科赛诺（北京）科技有限公司 | A kind of Data Assimilation method and device |
| CN108763865A (en) * | 2018-05-21 | 2018-11-06 | 成都信息工程大学 | A kind of integrated learning approach of prediction DNA protein binding sites |
-
2018
- 2018-11-28 CN CN201811441091.7A patent/CN109492758A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105976020A (en) * | 2016-04-28 | 2016-09-28 | 华北电力大学 | Network flow prediction method considering wavelet cross-layer correlations |
| CN106371155A (en) * | 2016-08-25 | 2017-02-01 | 华南师范大学 | A weather forecast method and system based on big data and analysis fields |
| CN106597574A (en) * | 2016-12-30 | 2017-04-26 | 重庆邮电大学 | Weather temperature prediction method and device based on time-varying cloud model |
| CN106960253A (en) * | 2017-03-10 | 2017-07-18 | 国网浙江省电力公司经济技术研究院 | A kind of regional middle-term electricity demand forecasting method for considering great social activities influence |
| CN108182492A (en) * | 2017-12-29 | 2018-06-19 | 中科赛诺（北京）科技有限公司 | A kind of Data Assimilation method and device |
| CN108763865A (en) * | 2018-05-21 | 2018-11-06 | 成都信息工程大学 | A kind of integrated learning approach of prediction DNA protein binding sites |
Non-Patent Citations (1)
| Title |
|---|
| 丁红 等: ""小波变换集遗传算法神经网络的径流预测建模"", 《广西大学学报:自然科学版》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE112020003136T5 (en) | Method for generating a lane change decision model, method and device for lane change decision of an unmanned vehicle | |
| Roodposhti et al. | Towards automatic calibration of neighbourhood influence in cellular automata land-use models | |
| DE112020003498T5 (en) | GENERATION OF TRAINING AND VALIDATION DATA FOR MACHINE LEARNING | |
| JP6859577B2 (en) | Learning methods, learning programs, learning devices and learning systems | |
| Diniz et al. | Mapping future changes in livelihood security and environmental sustainability based on perceptions of small farmers in the Brazilian Amazon | |
| Sharma | Designing and modeling fuzzy control Systems | |
| CN109711401A (en) | A kind of Method for text detection in natural scene image based on Faster Rcnn | |
| Paletta et al. | Convolutional neural networks applied to sky images for short-term solar irradiance forecasting | |
| Li et al. | Faster algorithm and sharper analysis for constrained Markov decision process | |
| Baležentis et al. | MULTIMOORA-IFN: A MCDM METHOD BASED ON INTUITIONISTIC FUZZY NUMBER FOR PERFORMANCE MANAGEMENT. | |
| CN107808167A (en) | A kind of method that complete convolutional network based on deformable segment carries out target detection | |
| DE102021207269A1 (en) | METHOD AND SYSTEM FOR LEARNING PERTURBATION QUANTITIES IN MACHINE LEARNING | |
| Chen et al. | Optimal structural policies for ambiguity and risk averse inventory and pricing models | |
| CN108073978A (en) | A kind of constructive method of the ultra-deep learning model of artificial intelligence | |
| WO2023057362A1 (en) | Structure detection for optimizing the use of resources in physical systems | |
| CN109492758A (en) | Data forecasting procedure and system | |
| JP7287490B2 (en) | LEARNING DEVICE, LEARNING METHOD, AND PROGRAM | |
| DE102021200348A1 (en) | COMPUTER-IMPLEMENTED METHOD OF TRAINING A COMPUTER VISION MODEL | |
| CN116324523A (en) | Turbulence prediction system and turbulence prediction method | |
| CN108073985A (en) | A kind of importing ultra-deep study method for voice recognition of artificial intelligence | |
| CN114255392B (en) | Nitrogen dioxide concentration prediction system based on satellite hyperspectral remote sensing and artificial intelligence | |
| Stephanopoulos et al. | Multi-scale aspects in model-predictive control | |
| DE112015005501T5 (en) | Aging profile forming machine for physical systems | |
| CN109492757A (en) | Data integrates method and device | |
| Martínez-Arellano et al. | Characterisation of large changes in wind power for the day-ahead market using a fuzzy logic approach |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190319 |