CN107038279A - The Forecasting Methodology and device of a kind of turbulence signal - Google Patents
The Forecasting Methodology and device of a kind of turbulence signal Download PDFInfo
- Publication number
- CN107038279A CN107038279A CN201710134437.8A CN201710134437A CN107038279A CN 107038279 A CN107038279 A CN 107038279A CN 201710134437 A CN201710134437 A CN 201710134437A CN 107038279 A CN107038279 A CN 107038279A
- Authority
- CN
- China
- Prior art keywords
- data
- markov
- model
- data result
- result
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
- G06F18/295—Markov models or related models, e.g. semi-Markov models; Markov random fields; Networks embedding Markov models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of Forecasting Methodology of turbulence signal and device, methods described includes：Turbulence signal is predicted using gray model, the first data result is obtained；Markov model is built, first data result is modified using the Markov model, the second data result is obtained；Second data result is modified using physical constraint condition, the 3rd data result is obtained, the 3rd data result meets physics law corresponding with the physical constraint condition.
Description
Technical field
The present invention relates to the Forecasting Methodology and dress in hydrodynamics data analysis technique field, more particularly to a kind of turbulence signal
Put.
Background technology
Turbulent flow is the common liquid form of nature and engineer applied field.Gas, liquid and gas-liquid-solid three-phase are complicated
Turbulent flow under flowing has very strong unsteady and stochastic behaviour.Turbulent flow is modeled to predict instantaneous flow and statistics
Characteristic is the important content that hydrodynamics is studied over nearly more than 100 years.Turbulent Flow Field can carry out table by a variety of physical quantitys
Levy, the flow field Basic Physical Properties such as speed, pressure, temperature and the momentum defined by these physical quantitys, kinetic energy, vorticity and
The mechanism amount such as stress.For turbulent flow sign physical quantity carry out timing variations prediction can catch turbulent flow interval it is existing
As, laminar flow turn twist coherent structure and laminar flow/turbulent boundary space-time characterisation in phenomenon, turbulent flow etc., with very important
Theory significance and use value.
Gray prediction is a kind of conventional Forecasting Methodology, and it can be to changing within the specific limits, relevant with the time
Gray system (the not only system containing Given information but also containing uncertain information) is predicted.Gray prediction thinking is：To original
Data are handled, and generation has stronger regular data sequence, corresponding Differential Equation Model is then set up, according to generation
Given data sequence solves the differential equation, obtains generating the matched curve of data sequence, and then obtains generating data sequence
Predicted value, then by the inverse generating process of data, so as to predict original things future developing trend.
Because the prediction curve of gray model is a smooth monotonous curve, larger to fluctuation data fitting compared with
Difference.Original data sequence is subtracted after Grey Model value, and the randomness of height is presented.This remaining random signal can be with
Portrayed with the theory of random process.Markoff process is a kind of typical random process, is the state for studying a system
And its theory of transfer, it determines shape by the research of the transition probability between the probability and state to different conditions
The variation tendency of state.
These conventional Forecasting Methodologies of gray model, Markov model are often for single-point clock signal.To two
The prediction of dimension or three-D sequential field of turbulent flow needs pointwise to carry out.Prediction field of turbulent flow often violates actual physics caused by so
Rule, obtains being unsatisfactory for the result of hydrodynamics fundamental equation.
The content of the invention
In order to solve the above technical problems, the embodiments of the invention provide a kind of Forecasting Methodology of turbulence signal and device, energy
Enough make the prediction of turbulence signal more accurate.
The Forecasting Methodology of turbulence signal provided in an embodiment of the present invention, including：
Turbulence signal is predicted using gray model, the first data result is obtained；
Markov model is built, first data result is modified using the Markov model, obtained
Second data result；
Second data result is modified using physical constraint condition, the 3rd data result, the described 3rd is obtained
Data result meets physics law corresponding with the physical constraint condition.
In the embodiment of the present invention, the utilization gray model is predicted to turbulence signal, obtains the first data result, bag
Include：
The first data sequence corresponding to the turbulence signal is pre-processed so that first data sequence meets institute
State the corresponding application conditions of gray model；
Accumulation process is carried out to first data sequence, the second data sequence, the rule of the second data sequence is generated
Sexual satisfaction preparatory condition；
Second data sequence is predicted using gray model, the first data result is obtained.
In the embodiment of the present invention, the structure Markov model is counted using the Markov model to described first
It is modified according to result, obtains the second data result, including：
The first data sequence and first data result in gray model, determine quantity of state, and according to institute
State quantity of state and determine Markov sequence；
According to the Markov sequence, Markov Transition Probabilities matrix is calculated；
Using the Markov Transition Probabilities matrix, the first data result that the Grey Model is obtained is corrected,
Obtain the second data result.
In the embodiment of the present invention, the utilization physical constraint condition is modified to second data result, including：
If the turbulence signal is one-dimensional physical quantity, based on the target area model belonging to second data result
Enclose, second data result is modified.
In the embodiment of the present invention, the utilization physical constraint condition is modified to second data result, including：
If the turbulence signal is two-dimensional physical amount or three dimensional physical amount, using corresponding with the physical quantity
Constraint equation is modified to second data result.
The prediction meanss of turbulence signal provided in an embodiment of the present invention, including：
Grey Model module, for being predicted using gray model to turbulence signal, obtains the first data result；
Prediction of Markov module, for building Markov model, using the Markov model to described first
Data result is modified, and obtains the second data result；
Physics correcting module, for being modified using physical constraint condition to second data result, obtains the 3rd
Data result, the 3rd data result meets physics law corresponding with the physical constraint condition.
In the embodiment of the present invention, the Grey Model module includes：
Pretreatment unit, for being pre-processed to corresponding first data sequence of the turbulence signal so that described
One data sequence meets the corresponding application conditions of the gray model；
Accumulation process unit, for carrying out accumulation process to first data sequence, generates the second data sequence, described
The rule sexual satisfaction preparatory condition of second data sequence；
Predicting unit, for being predicted using gray model to second data sequence, obtains the first data result.
In the embodiment of the present invention, the Prediction of Markov module includes：
Structural unit, for the first data sequence in gray model and first data result, determines shape
State amount, and Markov sequence is determined according to the quantity of state；
Computing unit, for according to the Markov sequence, calculating Markov Transition Probabilities matrix；
Amending unit, for utilizing the Markov Transition Probabilities matrix, corrects what the Grey Model was obtained
First data result, obtains the second data result.
In the embodiment of the present invention, the physics correcting module, specifically for：If the turbulence signal is one-dimensional physics
Amount, then based on the target area scope belonging to second data result, be modified to second data result.
In the embodiment of the present invention, the physics correcting module, specifically for：If the turbulence signal is two-dimensional physical amount
Or three dimensional physical amount, then second data result is modified using the constraint equation corresponding with the physical quantity.
In the technical scheme of the embodiment of the present invention, turbulence signal is predicted using gray model, the first data are obtained
As a result；Markov model is built, first data result is modified using the Markov model, second is obtained
Data result；Second data result is modified using physical constraint condition, the 3rd data result, the described 3rd is obtained
Data result meets physics law corresponding with the physical constraint condition.Using the technical scheme of the embodiment of the present invention, 1) knot
The precision of prediction can be effectively improved by having closed the advantage of gray level model and Markov model；2) caused by introducing physical constraint
Predict obtained physical field closer to real situation；3) it is relatively simple easy to realize, has saved calculating cost.
Brief description of the drawings
Fig. 1 is the schematic flow sheet one of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention；
Fig. 2 is the schematic flow sheet two of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention；
Fig. 3 is the schematic flow sheet three of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention；
Fig. 4 is the schematic flow sheet four of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention；
The one-dimensional turbulent velocity signal graph that Fig. 5 measures for the LDV of the embodiment of the present invention；
Fig. 6 (a) is the equal length interval division schematic diagram of the embodiment of the present invention；
Fig. 6 (b) divides schematic diagram for the intervals of equal probability of the embodiment of the present invention；
Fig. 7 is the flow field cloud atlas of the prediction of the embodiment of the present invention；
Fig. 8 is the structure composition schematic diagram of the prediction meanss of the turbulence signal of the embodiment of the present invention.
Embodiment
In order to more fully hereinafter understand the features of the present invention and technology contents, below in conjunction with the accompanying drawings to the reality of the present invention
Now it is described in detail, appended accompanying drawing purposes of discussion only for reference, not for limiting the present invention.
Reality is often violated using gray model and Markov model (abbreviation grey-Markov model) prediction field of turbulent flow
The physics law on border, obtains being unsatisfactory for the result of hydrodynamics fundamental equation.Therefore, pre- using grey-Markov module
Survey after field of turbulent flow, it is necessary to be modified with physical constraint to predicting the outcome, to improve precision of prediction.
Based on this, the embodiment of the present invention is using a kind of improved grey-Markov model based on physical constraint to rapids
Stream signal is predicted, i.e., carry out short-term forecast to turbulence signal first with gray model, then pass through Markov characteristic
Optimization Prediction result, the physical characteristic correction of result is predicted eventually through physical constraint, so as to realize that turbulent flow is instantaneously believed
Number short-term forecast.
It will be appreciated by those skilled in the art that the corresponding physical quantity of turbulence signal in following examples of the present invention can with but
It is not limited to following physical quantity：Speed, pressure, temperature, kinetic energy, momentum, vorticity and stress etc..
Fig. 1 is the schematic flow sheet one of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention, as shown in figure 1, the rapids
The Forecasting Methodology of stream signal comprises the following steps：
Step 101：Turbulence signal is predicted using gray model, the first data result is obtained.
Specifically, the first data sequence corresponding to the turbulence signal is pre-processed so that the first data sequence
Row meet the corresponding application conditions of the gray model；
Accumulation process is carried out to first data sequence, the second data sequence, the rule of the second data sequence is generated
Sexual satisfaction preparatory condition；
Second data sequence is predicted using gray model, the first data result is obtained.
Step 102：Markov model is built, first data result is repaiied using the Markov model
Just, the second data result is obtained.
Specifically, in gray model the first data sequence and first data result, determine quantity of state, and
Markov sequence is determined according to the quantity of state；
According to the Markov sequence, Markov Transition Probabilities matrix is calculated；
Using the Markov Transition Probabilities matrix, the first data result that the Grey Model is obtained is corrected,
Obtain the second data result.
Step 103：Second data result is modified using physical constraint condition, the 3rd data result is obtained,
3rd data result meets physics law corresponding with the physical constraint condition.
In one embodiment, if the turbulence signal is one-dimensional physical quantity, based on the second data result institute
The target area scope of category, is modified to second data result.
In another embodiment, if the turbulence signal be two-dimensional physical amount or three dimensional physical amount, using with
The corresponding constraint equation of the physical quantity is modified to second data result.
In the embodiment of the present invention, markoff process passes through the transfer between the probability and state to different conditions
The research of probability determines the variation tendency of state, and predicting the outcome for gray model is modified using Markov model,
The prediction that the gray model data sequence larger to stochastic volatility can be made up is not enough, improves precision of prediction.By introducing thing
Reason constraint causes the physical field that prediction is obtained closer to real situation.
Fig. 2 is the schematic flow sheet two of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention, as shown in Fig. 2 the rapids
The Forecasting Methodology of stream signal comprises the following steps：
Step 201：Turbulence signal is predicted using gray model.
Here, gray model is the forecast model for single-point simple component data sequence, and this step is specifically included：
Step 2011：Data sequence is pre-processed.
If the time series data of some physical quantity (if vector only takes a certain component) of raw velocity certain point (corresponds to
First data sequence) be：
x(0)=(x(0)(1), x(0)(2) ..., x(0)(n)) (1)
Wherein, n >=1.Inspection processing is done to the data first, examines whether it is applied to grey forecasting model.Specific side
Formula is the level ratio for calculating the ordered series of numbers：
If λ (k) all falls within intervalIt is interior, then it is assumed that data x(0)Suitable for grey forecasting model.
If being unsatisfactory for the judgement, need to do conversion process to initial data：
y(0)(k)=x(0)(k)+c, k=1,2 ..., n (3)
Wherein, c is constant.For turbulent boundary layer rate signal, exhibition in addition to main flow to the velocity component with normal direction, its
Numerical value, which has just, negative, thus need do with up-conversion process, using ensure data for just and meet level than judge.
Step 2012：Original data series are done with accumulation process, generation there are stronger regular data sequence (the second data of correspondence
Sequence).The usual way of data generation has Accumulating generation, inverse accumulated generating and weighted accumulation generation.For example, to Accumulating generation, it is first
First make accumulation calculating：
Obtain new ordered series of numbers：
x(1)=(x(1)(1), x(1)(2) ..., x(1)(n))。 (5)
Then to ordered series of numbers x(1)The adjacent value generation of weighting is done again：
z(1)(k)=α x(1)(k)+(1-α)x(1)(k-1), α ∈ [0,1] (6)
Wherein, α is weight coefficient.As α=0.5, z(1)(k) it is referred to as waiting the adjacent value generation number of power.α=0.5 is made herein.
Step 2013：It is predicted using gray model.
Initially set up gray model GM (1,1)：
x(0)(k)+az(1)(k)=b (7)
Bringing given data into has：
Introduce matrix notation：
Then gray model is represented by
Y=Bu (10)
Use least square method obtain the estimate of coefficient matrix for：
Bring albefaction model into：
It must solve：
Then predicted value is obtained：
It is correspondingly made available the predicted value of original data series：
Step 202：Markov is modeled and predicted.
Markov modeling and prediction are carried out on the basis of Grey Model result, is specifically included：
Step 2021：Quantity of state is determined, Markov sequence is constructed.
, it is necessary to select suitable quantity of state first before Markov sequence is defined.Obtained according to gray model original
Data sequence x(0)(t) predicted valueIt is defined as follows concept：
The two residual error：
The two relative residual error：
The two ratio：
The selection of quantity of state can be defined by residual error, relative residual error or the two ratio, and it is research shape that this, which sentences residual error δ,
State is described further.Codomain to residual error is split, and carrys out equidistant partition by fixed width h or is divided by equal-probability distribution
Into small interval [ai(t), bi(t)] (i=1,2 ..., Num).For the former bi(t)-ai(t)=h (i=1,2 ..., Num),
And for the latter P (δ (t) ∈ [ai(t), bi(t)])=1/Num (i=1,2 ..., Num) (P represents probability here).Assuming that its
In, Num is interval number, i.e. status number.I-th of state E of physical quantityiIt is defined as：
Ei=δ (t) ∈ [ai(t), bi(t)] (i=1,2 ..., Num) (19)
If physical quantity to be predicted is vector, such as the prediction to three-component (vector) velocity field, the definition of state can
To combine all components.U velocity component residual error δ u (t) codomain is divided into L area first, is respectively [ai, bi] (i=1,
2 ..., L)；To v velocity components, its residual error δ v (t) is divided into M area, is respectively [ci, di] (i=1,2 ..., M)；To w to
Velocity component, N number of state is divided into by its residual error δ w (t), is respectively [ei, fi] (i=1,2 ..., N).
State E (i, j, k) definition is that (δ u, δ v, δ w) is met simultaneously：
The state of Unify legislation scalar physical quantity and vector physical quantity for convenience, here by state E (i, j, k) (i, j, k
=1,2) sequentially formed a line according to certain.Such as state E (i, j, k) comes i-th+(j-1) L+ (j-1) (k- of the sequence
1) M position.Remember that the state arranged remembers E respectively1, E2..., ENum(Num=L × M × N).
Step 2022：Calculate Markov Transition Probabilities matrix.
According to state demarcation above, some states are designated as E respectively1, E2..., ENum, state EiTurn by a time step
Move on to EjProbability be referred to as transition probability, be designated as：
Wherein, mijFor state EiE is transferred to by a time stepjNumber of times, MiFor state EiThe total degree of appearance.pij
Matrix form is：
Wherein,
Step 2023：Modified grey model predicted value.
If current state is Ei, it is considered to the i-th row in P, if the maximum max (P of the rowij)=Pik, then it is assumed that lower a period of time
Etching system is most possible from state EiSteering state Ek。
For scalar status, EkCorresponding state interval [ak, bk], then the Markov correction value of gray prediction value is：
For vector state, such as velocity (δ u, δ v, δ w), it is assumed that EkCorresponding state interval [ak1, bk1], [ck2,
dk2], [ek3, fk3], then the Markov correction value of gray prediction value is：
Step 203：The amendment of physical constraint is carried out to prediction physical field using physical constraint.
Can be using the method for fixed codomain scope to above-mentioned step for the physical quantity such as one-dimensional temperature signal of single-point
Suddenly the result predictedMake further amendment.Specifically, the x of data with existing is counted(0)(k), (k=1,2 ..., being averaged n)
ValueAnd variances sigma (x), then make following amendment：
The method that prediction for two dimension or three-dimensional physical quantity can be constrained using equation is modified, such as to pressure
Power gradient fields are constrained using irrotationality, and velocity field uses the constraint of continuity equation.
By taking 3D velocity field as an example, the step of introducing physical constraint amendment.It is all pre- to test the speed after pointwise completes prediction
The three-dimensional predetermined speed of degree compositionHydromechanical continuity equation should be met in view of velocity field：
Need to be modified the velocity field of prediction.The embodiment of the present invention passes through solution using the principle of minimum amendment
The minimum of optimization problem is worth to erection rate below：
Here s.t. is followed by after constraints, min being minimum target function.| | | | 2 norms of representative function, definition
For
To the step for specific implementation be related to the discrete of above-mentioned steps.Assuming that real data distribution nx × ny ×
On the mesh point of nz specification.Ijk represents the node numbering of three coordinate directions.Discrete velocity is indexed with upper table (i, j, k).This
The above-mentioned optimization problem of sample can be write as：
HereIt is the difference coefficient for calculating derivative.According to Second-Order Central Difference form, side
Forward or a backward difference scheme is used in boundary, these difference coefficients can constitute following matrix form.
Solution to above mentioned problem, can be solved by conventional conjugate gradient method or other Mathematics Optimization Methods.
The turbulent-velocity field predicted using the embodiment of the present invention, disclosure satisfy that the continuous side of one of hydromechanical fundamental equation
Journey can more reflect the velocity field of true flow field structure there is provided one, and the research to flow mechanism for behind is provided preferably
Material.
Prediction for other physical quantitys such as barometric gradient is similar with said process, simply replaces corresponding constraint equation
The curl of the equation that should be met for barometric gradient, i.e. barometric gradient should be zero.
The scheme of the embodiment of the present invention is described in further detail with reference to concrete application scene.
Embodiment one
X is that rate signal is flowed in the turbulent boundary layer that LDV is obtained, as shown in Figure 5.X has 1000 samples
This point.The numerical value at x the 1001st moment is predicted with the method for the embodiment of the present invention below.Fig. 3 is the embodiment of the present invention
Turbulence signal Forecasting Methodology schematic flow sheet three, as shown in figure 3, comprising the following steps：
Step 301：Grey Model.
Step 3011：Original data signal is examined, level ratio is calculated
It was found that λ (k) fails to fall in intervalIt is interior, then it is assumed that data x(0)It is not suitable for grey forecasting model, in
It is that conversion process is done to initial data：
x′(0)=x(0)+100000 (33)
Examine again and find x′(0)Suitable for Grey Model, then the prediction to x can be by predicting x ' realizations.
Step 3012：To x ' Accumulating generations：
Obtain new ordered series of numbers：
x′(1)=(x′(1)(1), x′(1)(2) ..., x′(1)(n)) (35)
Then to ordered series of numbers x(1)The adjacent value generation of weighting is done again：
z(1)(k)=0.5x′(1)(k)+0.5x′(1)(k-1) (36)
Step 3013：It is predicted using gray model.Note：
Calculate：
Obtain predicted value：
So as to obtain the predicted value of original data series：
Then x predicted value is
Step 302：Markov is modeled.
Step 3021：Construct Markov sequence.Selection obtains original data sequence x according to gray model(0)(t) pre-
Measured valueDefinition status amount is the two residual error：
δ (t) probability density distribution is counted, and waits siding-to-siding block length or δ (t) codomain is equiprobably divided into 4
Individual interval [ai(t), bi(t)], i=1,2,3,4.In this example, δ (t) probability density function and correspondingly two kinds of intervals stroke
Divide shown in mode such as Fig. 6 (a), 6 (b).
Following step is independent of specific division methods.For the sake of simplicity, in following calculating, acquiescence is used etc.
The result that probability is divided.
I-th of state EiIt is defined as：
Ei=δ (t) ∈ [ai(t), bi(t)], i=1,2,3,4 (42)
Step 3022：Calculate Markov Transition Probabilities matrix.
According to state demarcation above, some states are designated as E respectively1, E2, E3, E4, state EiE is transferred to by 1 stepj's
Probability is designated as：
Wherein, mijFor state EiE is transferred to by 1 stepjNumber of times, MiFor state EiThe number of times of appearance.Here, its matrix
Form is：
Step 3023：Modified grey model predicted value.
1000th the moment δ=117.6, belong to the 4th state E4.Then the 4th row in P is considered, if the maximum of the row
Value max (P44)=0.3, then it is assumed that subsequent time (t=1001) system is most possibly from state E4Steering state E4。
State E4Corresponding state interval [93.4,324.0], then the Markov correction value of gray prediction value be：
Step 303：Correct predicted value.
Above-mentioned predict the outcome belongs to the three times variance scope of initial data, i.e., within the scope of [26.0,847.1], therefore according to
According to formula 28, do not made an amendment to predicting the outcome.
Embodiment two
Three-dimensional flow field measurement has obtained the velocity field (u, v, w) of 200 time serieses.3D velocity field is distributed in 175 ×
On 105 × 21 mesh node.Below the 201st velocity field is predicted with the method for the present invention.Due to gray model and Ma Erke
Husband's prediction is that pointwise is carried out, therefore when describing following step 401-402, only with the speed on three-dimensional grid center
Exemplified by degree.Fig. 4 is the schematic flow sheet four of the Forecasting Methodology of the turbulence signal of the embodiment of the present invention, as shown in figure 4, including following
Step：
Step 401：Grey Model.Prediction of the gray model to three components of velocity field is independently carried out.Therefore here
Only illustrated by taking a velocity component u as an example.
Step 4011：Examine original data signal.Raw velocity signal is translated first：
u′(0)=u(0)+100000 (46)
Then level is made than range check to the rate signal after translation, detailed process is shown in step 3011.Through examining, after translation
Rate signal be applied to gray model.
Step 4012：To u ', Accumulating generation is done respectively：
Obtain new ordered series of numbers：
u′(1)=(u′(1)(1), u′(1)(2) ..., u′(1)(n)) (48)
Then to ordered series of numbers u(1)The adjacent value generation of weighting is done again：
z(1)(k)=0.5u′(1)(k)+0.5u′(1)(k-1) (49)
Step 4013：It is predicted using gray model.Note：
Calculate：
Obtain predicted value：
So as to obtain the predicted value of original data series：
Then u predicted value is
Step 402：Markov is modeled.
Step 4021：Construct Markov sequence.Selection obtains original data sequence u according to gray model(0)(t), v(0)
And w (t)(0)(t) predicted valueWithDefinition status amount is the two residual error.
Statistics δ u (t), δ v (t) and δ w (t) probability density distribution, and equiprobably by δ u (t), δ v (t), δ w respectively
(t) codomain is divided into 2 interval [ai(t), bi(t)], [cj(t), dj] and [e (t)k(t), fk(t)], i, j, k=1,2, and P
(δu(t)∈[ai(t), bi(t)])=P (δ v (t) ∈ [ai(t), bi(t)])=P (δ w (t) ∈ [ek(t), fk(t)])=1/2.
Table 1 below summarizes three velocity components totally six interval scopes.
Table 1 (three velocity component state intervals are divided)
| δu | δv | δw | |
| Interval 1 | [- 107.06, -1.56] | [- 55.42, -0.98] | [- 37.17, -1.49] |
| Interval 2 | [- 1.56,78.98] | [- 0.98,62.43] | [- 1.49,38.54] |
State E (i, j, k) is defined as：
Therefore the sum of state is 2 × 2 × 2=8.
Step 4022：Calculate Markov Transition Probabilities matrix.
According to state demarcation above, some states are designated as E respectively1, E2..., E8, state EiE is transferred to by 1 stepj's
Probability is referred to as transition probability, is designated as：
Wherein, mijFor state EiE is transferred to by 1 stepjNumber of times, MiFor state EiThe number of times of appearance.Herein, shift
The concrete numerical value of probability matrix is as shown in the table.
Table 2 (transitional provavility density matrix numerical value)
| P (%) | E (1,1,1) | E (2,1,1) | E (1,2,1) | E (2,2,1) | E (1,1,2) | E (2,1,2) | E (1,2,2) | E (2,2,2) |
| E (1,1,1) | 20.00 | 15.00 | 15.00 | 5.00 | 0.00 | 10.00 | 30.00 | 5.00 |
| E (2,1,1) | 3.33 | 16.67 | 6.67 | 30.00 | 10.00 | 3.33 | 10.00 | 20.00 |
| E (1,2,1) | 10.00 | 15.00 | 10.00 | 20.00 | 15.00 | 5.00 | 20.00 | 5.00 |
| E (2,2,1) | 6.25 | 18.75 | 6.25 | 12.50 | 12.50 | 12.50 | 15.63 | 15.63 |
| E (1,1,2) | 6.45 | 12.90 | 12.90 | 16.13 | 25.81 | 6.45 | 12.90 | 6.45 |
| E (2,1,2) | 11.11 | 11.11 | 0.00 | 16.67 | 16.67 | 16.67 | 11.11 | 16.67 |
| E (1,2,2) | 9.38 | 9.38 | 15.63 | 18.75 | 18.75 | 6.25 | 18.75 | 3.13 |
| E (2,2,2) | 14.29 | 23.81 | 4.76 | 4.76 | 19.05 | 19.05 | 4.76 | 9.52 |
Step 4023：Modified grey model predicted value.
200th moment δ, u=-2.03, δ v=-14.81, δ w=16.94 belonged to the 5th state E (1,1,2).Then examine
The 5th row in P is considered, if the maximum max (P of the row55)=25.81%, then it is assumed that subsequent time (t=201) system most has can
Can be from state E (1,1,2) steering state E (1,1,2).State E4Corresponding state interval be (δ u, δ v, δ w) ∈ [- 107.06 ,-
1.56] × [- 55.42, -0.98] × [- 1.49,38.54], then the Markov correction value of gray prediction value be：
Step 403：Physics amendment.
After pointwise completes to the three-component prediction of velocity field, all pre- results that test the speed are constituted to the prediction of Three-dimendimal fusion
Velocity field.Then, by solving the optimization problem as shown in formula 26, the physics amendment to predetermined speed is completed.After amendment
Speed to flow to distribution of the component in middle section as shown in Figure 7.
Fig. 8 is the structure composition schematic diagram of the prediction meanss of the turbulence signal of the embodiment of the present invention, as shown in figure 8, described
Device includes：
Grey Model module 81, for being predicted using gray model to turbulence signal, obtains the first data knot
Really；
Prediction of Markov module 82, for building Markov model, using the Markov model to described
One data result is modified, and obtains the second data result；
Physics correcting module 83, for being modified using physical constraint condition to second data result, obtains
Three data results, the 3rd data result meets physics law corresponding with the physical constraint condition.
In the embodiment of the present invention, the Grey Model module 81 includes：
Pretreatment unit 811, for being pre-processed to corresponding first data sequence of the turbulence signal so that described
First data sequence meets the corresponding application conditions of the gray model；
Accumulation process unit 812, for carrying out accumulation process to first data sequence, generates the second data sequence,
The rule sexual satisfaction preparatory condition of the second data sequence；
Predicting unit 813, for being predicted using gray model to second data sequence, obtains the first data knot
Really.
In the embodiment of the present invention, the Prediction of Markov module 82 includes：
Structural unit 821, for the first data sequence in gray model and first data result, it is determined that
Quantity of state, and Markov sequence is determined according to the quantity of state；
Computing unit 822, for according to the Markov sequence, calculating Markov Transition Probabilities matrix；
Amending unit 823, for utilizing the Markov Transition Probabilities matrix, corrects the Grey Model and obtains
The first data result, obtain the second data result.
In the embodiment of the present invention, the physics correcting module 83, specifically for：If the turbulence signal is one-dimensional physics
Amount, then based on the target area scope belonging to second data result, be modified to second data result.
In the embodiment of the present invention, the physics correcting module 83, specifically for：If the turbulence signal is two-dimensional physical
Amount or three dimensional physical amount, then repaiied using the constraint equation corresponding with the physical quantity to second data result
Just.
It will be appreciated by those skilled in the art that each unit in the prediction meanss of turbulence signal shown in Fig. 8 realizes work(
The associated description of the Forecasting Methodology of foregoing turbulence signal can be can refer to and understood.
In the embodiment of the present invention, each module can pass through central processing unit in the prediction meanss of the turbulence signal
(Central Processing Unit, CPU), digital signal processor (Digital Signal Processor, DSP) or
Programmable logic array (Field-Programmable Gate Array, FPGA) is realized.
, can be in any combination in the case where not conflicting between technical scheme described in the embodiment of the present invention.
In several embodiments provided by the present invention, it should be understood that disclosed method and smart machine, Ke Yitong
Other modes are crossed to realize.Apparatus embodiments described above are only schematical, for example, the division of the unit, only
Only a kind of division of logic function, can have other dividing mode, such as when actually realizing：Multiple units or component can be tied
Close, or be desirably integrated into another system, or some features can be ignored, or do not perform.In addition, shown or discussed each group
Into part coupling each other or direct-coupling or communication connection can be by some interfaces, equipment or unit it is indirect
Coupling is communicated to connect, and can be electrical, machinery or other forms.
The above-mentioned unit illustrated as separating component can be or may not be it is physically separate, it is aobvious as unit
The part shown can be or may not be physical location, you can positioned at a place, can also be distributed to multiple network lists
In member；Part or all of unit therein can be selected to realize the purpose of this embodiment scheme according to the actual needs.
In addition, each functional unit in various embodiments of the present invention can be fully integrated into a second processing unit,
Can also be each unit individually as a unit, can also two or more units it is integrated in a unit；
Above-mentioned integrated unit can both be realized in the form of hardware, it would however also be possible to employ hardware adds the form of SFU software functional unit real
It is existing.
The foregoing is only a specific embodiment of the invention, but protection scope of the present invention is not limited thereto, any
Those familiar with the art the invention discloses technical scope in, change or replacement can be readily occurred in, should all be contained
Cover within protection scope of the present invention.
Claims (10)
1. a kind of Forecasting Methodology of turbulence signal, it is characterised in that methods described includes：
Turbulence signal is predicted using gray model, the first data result is obtained；
Markov model is built, first data result is modified using the Markov model, second is obtained
Data result；
Second data result is modified using physical constraint condition, the 3rd data result, the 3rd data is obtained
As a result physics law corresponding with the physical constraint condition is met.
2. according to the method described in claim 1, it is characterised in that the utilization gray model is predicted to turbulence signal,
The first data result is obtained, including：
The first data sequence corresponding to the turbulence signal is pre-processed so that first data sequence meets the ash
The corresponding application conditions of color model；
Accumulation process is carried out to first data sequence, the second data sequence is generated, the regularity of the second data sequence expires
Sufficient preparatory condition；
Second data sequence is predicted using gray model, the first data result is obtained.
3. method according to claim 2, it is characterised in that the structure Markov model, utilizes the Ma Erke
Husband's model is modified to first data result, obtains the second data result, including：
The first data sequence and first data result in gray model, determine quantity of state, and according to the shape
State amount determines Markov sequence；
According to the Markov sequence, Markov Transition Probabilities matrix is calculated；
Using the Markov Transition Probabilities matrix, the first data result that the Grey Model is obtained is corrected, is obtained
Second data result.
4. according to the method described in claim 1, it is characterised in that the utilization physical constraint condition is to the second data knot
Fruit is modified, including：
If the turbulence signal is one-dimensional physical quantity, right based on the target area scope belonging to second data result
Second data result is modified.
5. according to the method described in claim 1, it is characterised in that the utilization physical constraint condition is to the second data knot
Fruit is modified, including：
If the turbulence signal is two-dimensional physical amount or three dimensional physical amount, using the constraint corresponding with the physical quantity
Equation is modified to second data result.
6. a kind of prediction meanss of turbulence signal, it is characterised in that described device includes：
Grey Model module, for being predicted using gray model to turbulence signal, obtains the first data result；
Prediction of Markov module, for building Markov model, using the Markov model to first data
As a result it is modified, obtains the second data result；
Physics correcting module, for being modified using physical constraint condition to second data result, obtains the 3rd data
As a result, the 3rd data result meets physics law corresponding with the physical constraint condition.
7. device according to claim 6, it is characterised in that the Grey Model module includes：
Pretreatment unit, for being pre-processed to corresponding first data sequence of the turbulence signal so that first number
The corresponding application conditions of the gray model are met according to sequence；
Accumulation process unit, for carrying out accumulation process to first data sequence, generates the second data sequence, described second
The rule sexual satisfaction preparatory condition of data sequence；
Predicting unit, for being predicted using gray model to second data sequence, obtains the first data result.
8. device according to claim 7, it is characterised in that the Prediction of Markov module includes：
Structural unit, for the first data sequence in gray model and first data result, determines quantity of state,
And Markov sequence is determined according to the quantity of state；
Computing unit, for according to the Markov sequence, calculating Markov Transition Probabilities matrix；
Amending unit, for utilizing the Markov Transition Probabilities matrix, corrects the Grey Model is obtained first
Data result, obtains the second data result.
9. device according to claim 6, it is characterised in that the physics correcting module, specifically for：If the rapids
Stream signal is one-dimensional physical quantity, then based on the target area scope belonging to second data result, to the second data knot
Fruit is modified.
10. device according to claim 6, it is characterised in that the physics correcting module, specifically for：If described
Turbulence signal is two-dimensional physical amount or three dimensional physical amount, then using the constraint equation corresponding with the physical quantity to described the
Two data results are modified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710134437.8A CN107038279A (en) | 2017-03-08 | 2017-03-08 | The Forecasting Methodology and device of a kind of turbulence signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710134437.8A CN107038279A (en) | 2017-03-08 | 2017-03-08 | The Forecasting Methodology and device of a kind of turbulence signal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107038279A true CN107038279A (en) | 2017-08-11 |
Family
ID=59533201
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710134437.8A Pending CN107038279A (en) | 2017-03-08 | 2017-03-08 | The Forecasting Methodology and device of a kind of turbulence signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107038279A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103106256A (en) * | 2013-01-23 | 2013-05-15 | 合肥工业大学 | Gray model (GM) (1,1) prediction method of orthogonal interpolation based on Markov chain |
| CN103545846A (en) * | 2013-11-11 | 2014-01-29 | 湖南大学 | Microgrid economic operation method based on generalized load prediction |
| CN103996071A (en) * | 2014-02-25 | 2014-08-20 | 沈阳理工大学 | Wind power plant wind speed prediction method based on Markov theory |
| CN105825040A (en) * | 2015-12-29 | 2016-08-03 | 海南电力技术研究院 | Short-term power load prediction method |
-
2017
- 2017-03-08 CN CN201710134437.8A patent/CN107038279A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103106256A (en) * | 2013-01-23 | 2013-05-15 | 合肥工业大学 | Gray model (GM) (1,1) prediction method of orthogonal interpolation based on Markov chain |
| CN103545846A (en) * | 2013-11-11 | 2014-01-29 | 湖南大学 | Microgrid economic operation method based on generalized load prediction |
| CN103996071A (en) * | 2014-02-25 | 2014-08-20 | 沈阳理工大学 | Wind power plant wind speed prediction method based on Markov theory |
| CN105825040A (en) * | 2015-12-29 | 2016-08-03 | 海南电力技术研究院 | Short-term power load prediction method |
Non-Patent Citations (2)
| Title |
|---|
| 杜占龙 等: "基于多渐消因子强跟踪UKF和约束AR模型的故障估计与预测", 《控制与决策》 * |
| 端木京顺 等: "《航空事故预测预警预防理论方法》", 30 June 2013, 北京：国防工业出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Aly | An intelligent hybrid model of neuro Wavelet, time series and Recurrent Kalman Filter for wind speed forecasting | |
| Zhang et al. | Efficient calibration techniques for large-scale traffic simulators | |
| Goldstein et al. | Probabilistic formulations for transferring inferences from mathematical models to physical systems | |
| Wandel et al. | Spline-pinn: Approaching pdes without data using fast, physics-informed hermite-spline cnns | |
| Osorio et al. | Simulation-based optimization: achieving computational efficiency through the use of multiple simulators | |
| CN106326585A (en) | Prediction analysis method based on bayesian network reasoning and device thereof | |
| CN104048676B (en) | MEMS (Micro Electro Mechanical System) gyroscope random error compensating method based on improved particle filter | |
| CN108108559A (en) | A kind of structural response acquisition methods and sensitivity acquisition methods based on minor structure | |
| CN106909719A (en) | Integrated arithmetic of linearity regression | |
| CN103366033B (en) | method and system for statistical circuit simulation | |
| Wong et al. | Graph neural network based surrogate model of physics simulations for geometry design | |
| CN107038279A (en) | The Forecasting Methodology and device of a kind of turbulence signal | |
| CN104933103A (en) | Multi-target community discovering method integrating structure clustering and attributive classification | |
| CN106342305B (en) | A kind of testability index requiring towards multitask is determined method | |
| CN108920737A (en) | A kind of particle filter assimilation method of hydrodynamic model, device and calculate equipment | |
| Shahane et al. | Convolutional neural network for flow over single and tandem elliptic cylinders of arbitrary aspect ratio and angle of attack | |
| CN107133104A (en) | A kind of distributed stream data processing method | |
| Shenoy et al. | Estimation of Liquid Level in a Harsh Environment Using Chaotic Observer | |
| JP4911244B2 (en) | Compressed air leak diagnostic device | |
| Schwenker et al. | Echo state networks and neural network ensembles to predict sunspots activity | |
| Muklisin et al. | The Analysis of Airport Passengers Flow by Using Spatial Temporal Graph Neural Networks and Resolving Efficient Dominating Set | |
| Malineni et al. | On the Performance of a Data-Driven Backward Compatible Physics Informed Neural Network (BC-PINN) for Prediction of Flow Past a Cylinder | |
| Chung | Regular sensitivity calculation and gradient-based optimization of chaotic dynamical systems | |
| Liu et al. | Acceleration of steady-state lattice Boltzmann simulations for exterior flows | |
| Heinemann | ML-based visual analysis of droplet behaviour in multiphase flow simulations |
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: 20170811 |
|
| RJ01 | Rejection of invention patent application after publication |