CN108492179A - Time-frequency spectrum generation method and equipment - Google Patents
Time-frequency spectrum generation method and equipment Download PDFInfo
- Publication number
- CN108492179A CN108492179A CN201810146215.2A CN201810146215A CN108492179A CN 108492179 A CN108492179 A CN 108492179A CN 201810146215 A CN201810146215 A CN 201810146215A CN 108492179 A CN108492179 A CN 108492179A
- Authority
- CN
- China
- Prior art keywords
- time
- frequency
- frequency spectrum
- medium
- frequency information
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The object of the present invention is to provide a kind of time-frequency spectrum generation methods and equipment applied to securities data quantization transaction, the present invention passes through recursion low-pass filter, after isolating low-frequency information, in time-frequency spectrum, frequency is confined in medium-high frequency limited range, to solve the problems, such as the low frequency of data, can obtain generating the medium-high frequency information needed for time-frequency spectrum.Time-frequency spectrum is established using recursion wavelet algorithm, solves the instant sex chromosome mosaicism of securities market data.Difference processing is carried out additionally by being composed to the dynamic time-frequency, overcomes the problems, such as that the time-frequency spectrum precision caused by Heisenberg's uncertain factor is not high.
Description
Technical field
The present invention relates to computer realm more particularly to a kind of time-frequency spectrum generation methods and equipment.
Background technology
Due to the continuous generation of academic theory, the continuous renewal of trading environment, computer technology be constantly progressive and continue
Development, quantization trading strategies show various advantages, and compare some advantages of other trading strategies, to present
Thriving gesture.In the form of quantization transaction is considered as a kind of " can making money by recumbency " by many trend investors, they think only
A set of mathematic system by accurate calculating is worked out, transaction can be quantified, and profit can become have no pressure.So
And quantify transaction other than the advantage shown, there are the problem of the problem faced of knead dough be also obvious.
Quantify trading strategies, exactly use quantification means built-up and carries out the trading strategies of decision.Specific explanations
Get up, this definition includes two layers of meaning.First, during building trading strategies, the means of quantification should account for mainly at
Point.Here quantification means include portraying, entire transaction flow and transacting targeted quantity to the optimal of quantified goal
Change, to methods such as the Quantitative Estimations of Policy Result.But still allow for difinite quality in this part or it is artificial
Subjective ingredient exists, and tactful research and development after all are the processes of a manual operation.Secondly, trading strategies are finished in construction, are used
Come when being traded decision, it is necessary to have specific quantification rule is completely absent the ingredient of subjective judgement.This characteristic
It ensure that entire strategy can carry out backtracking test under the setting quantified completely, be the necessary condition of previous feature.Simultaneously
Meet the restriction in terms of the two, then can be referred to as quantization trading strategies.
Due to quantifying the speciality of trading strategies itself, there are following several big advantages：
(1) metrizability
It is built-up and carry out the trading strategies of decision to be taken as quantification means, in building process and decision mistake
Cheng Zhong, can be by precisive.Contrastingly, although subjectivityization, trading strategies qualitatively are durings discs etc.
The transaction results of quantification can be obtained, but due to lacking the precisive of globality, the quantification result of part is past
Toward fluctuate it is larger, do not have largely and stable portray ability.
(2) verifiability
Due to the non-availability of Future Data, actually deal maker when judging a trading strategies, largely all according to
Rely the test result in historical data in strategy.On the basis of construction of strategy and strategy statement quantification, multiplicating is gone through
History backtracking test can obtain consistent result.If test result is positive, it can at least illustrate quantization transaction plan
Slightly there is profitability in history inspection.
(3) objectivity
Quantification means are occupied an leading position during due to structure trading strategies, and trade decision is even more to have specific number
Quantizing rule, therefore quantify trading strategies and can largely evasion tactics developer get sth into one's head, in the mistake of structure
It obtains in journey and objectively treats always.
(4) independence
The subjective judgement that quantization trading strategies do not need deal maker can also completely instruct entire transaction flow.Although amount
Change trading strategies can not possibly help we completely evade subjective trading strategies there are the problem of, but the frame of quantification and rule
Damage caused by these factors leading to social instability can be then minimized really.
(5) consistency
Quantify trading strategies and ensure trading strategies in the process of implementation, using identical trading rules, including buying in
Point, the determination etc. for selling point, bin size of merchandising.History verification process and practical friendship can be accomplished by quantifying trading strategies simultaneously
The easy consistency for being, because of whether true sale decision or land parcel change trace test, the trading rules of institute's reference be all by
Quantitative Tables reach explication.This consistency on the whole, overwhelming majority subjectivity trading strategies can not all ensure.
(6) portable
Unless using specific quantization factor, it is however generally that quantization trading strategies are easier in a market or money
After being confirmed effectively in production, it is transplanted to and uses in other markets or assets.Data obtains used in quantization trading strategies
Property is stronger, and transfer ability is also stronger.
Conventional quantization strategy will establish model according to statistical principle, since it is exactly one theoretic to be " model "
Thing is to be based on a large amount of hypothesis, simplify, compromise, obtained from historical experience, but it is known that real world will not be so
Idealization, history can not possibly be repeated simply, and model failure is unavoidable.
There are one very great disadvantages for conventional quantization trading strategies, in that quantifies this characteristic itself.Really this
One is characterized as that quantization trading strategies bring a variety of advantages, but due to this feature so that quantization trading strategies are treating nothing
When the factor of method quantization, it can only take and give up treating method more helpless in this way.Therefore, quantization trading strategies lost very
It mostly is possible to bring the information of profit in fact, but also the range that strategy is covered when handling information becomes narrow.Certainly,
With the development of science and technology, the factors that can not be quantized before some initially enter quantization trading strategies research range it
It is interior, such as the investor sentiment etc. that the network information depicts.But even if technological means, which can allow, quantifies trading strategies energy
The range of information enough handled is wider deeper, and for compared with subjective trading strategies, such defect is quantization trading strategies always
It can not thoroughly break away from.This shortcomings that deriving from self-characteristic, it can only improve and can not effect a radical cure.
Simultaneously as quantifying trading strategies in building process using the method for quantification, need a certain number of
Data sample is studied, and corresponding data are all as the time gradually generates, therefore when the construction of quantization trading strategies
Form not change substantially when, the quantification feature extracted from data also only can be as the time gradually changes, tactful institute
The transaction of formation also can only slowly change.When great change occurs for market situation, this slowly varying characteristic can cause
Quantization trading strategies can not adapt to the market of period of transfer, cause larger loss in a short time.In comparison, a part is fixed
The trading strategies of property are traded due to being based primarily upon thinking in logic, can when market situation changes
Carry out Developing Tactics substantially rapidly based on subjective logic.It is also that quantization trading strategies one are more difficult to turn to slowly this feature
Improved defect.
Invention content
It is an object of the present invention to provide a kind of time-frequency spectrum generation method and equipment, can solve traditional security quantization and hand over
The easily inaccurate problem of strategy.
According to an aspect of the invention, there is provided a kind of time-frequency spectrum generation method, this method include：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information divided
From；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates date, ordinate table
Show that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to the energy of corresponding medium-high frequency information
It is worth, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
Further, in the above method, by Hilbert transform, instantaneous frequency is sought to the low-frequency information after the separation
Rate.
Further, in the above method, by Hilbert transform, instantaneous frequency is sought to the low-frequency information after the separation
Rate, including：
Hilbert transform is carried out to low-frequency information：zn=hn*yn
Wherein
ynFor the low-frequency information of the securities market data after low-pass filtering；
Construct securities market data complex signal sn=yn+izn, at any one moment, reading corresponding instantaneous phase is
Instantaneous frequency μ is sought according to instantaneous phasen=θn-θn-1。
Further, in the above method, the recurrence formula of the low-pass filter is as follows：
Wherein,
f0For frequency filtering, xnFor a recursion median, FnFor discrete securities market data, initial value F0It is risen for data
Initial value, ynFor the low-frequency information of the securities market data after low-pass filtering, xnAnd ynInitial value be 0.
Further, in the above method, time-frequency spectrum is obtained using the medium-high frequency information, including：
Using recursion wavelet algorithm, dynamic time-frequency spectrum is obtained according to the medium-high frequency information.
Further, in the above method, using recursion wavelet algorithm, dynamic time-frequency is obtained according to the medium-high frequency information
Spectrum, including：
According to medium-high frequency information, determine that the Complex wavelet that can be used for generating time-frequency spectrum, function expression are
Wherein, coefficient ω0=2 π, δ=π, i are complex symbol, and t is the time；
By the Complex wavelet function of time-frequency spectrumObtain securities market data FnWavelet transformation：
Wherein, WfFor wavelet coefficient, f is the frequency analysis range of time-frequency spectrum, τ=- 5, -4, -3, -2, -1,0,1,2,3,
4,5, s be scale variable；It is morther wavelet discrete series, it is reflectedContent；
According to the transform of sequence and its convolution property, securities market data Fn, morther wavelet sequenceAnd small echo
Coefficient sequence WfThe transform of (s, n) is：
Wherein,
It enablesTo the morther wavelet sequence
Transform is carried out, is obtained：
Wherein,
It willW (z) is substituted into obtain：
Then
Z inverse transformations are carried out to formula (6) and obtain recurrence formula：
According to the W of formula (7)f(s, n) calculates separately W (s, n-1), W (s, n-2), W (s, n-3), W (s, n-4), to the right
Individual event recurrence calculation goes out all wavelet coefficient sequence Wf(s, n)；
According to all wavelet coefficient Wf(s, n) obtains the dynamic time-frequency spectrum.
Further, in the above method, using recursion wavelet algorithm, dynamic time-frequency spectrum is obtained according to the medium-high frequency information
Later, further include：
The dynamic time-frequency is composed and carries out difference processing, is composed with obtaining high-precision dynamic time-frequency.
Further, in the above method, the dynamic time-frequency is composed and carries out difference processing, when obtaining high-precision dynamic
Frequency spectrum, including：
It is 1-50 that difference step size, which is respectively adopted, to dynamic time-frequency spectrum, 50 difference summation process is carried out, to obtain height
The dynamic time-frequency of precision is composed.
Further, in the above method, obtaining time-frequency spectrum using the medium-high frequency information includes：
Using the medium-high frequency information, the positive frequency of time-frequency spectrum is obtained；
Using low-frequency information remaining near the medium-high frequency information, the negative frequency of time-frequency spectrum is obtained.
According to another aspect of the present invention, a kind of time-frequency spectrum generation equipment is additionally provided, which includes：
Acquisition device, for obtaining securities market data；
Separator, by recursion low-pass filter, by the low-frequency information and medium-high frequency letter in the securities market data
Breath is detached；
Time-frequency spectral apparatus, for obtaining time-frequency spectrum using the medium-high frequency information, the abscissa of the time-frequency spectrum indicates day
Phase, ordinate indicate that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to corresponding medium-high frequency
The energy value of information, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
According to another aspect of the present invention, a kind of equipment based on calculating is additionally provided, wherein including：
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the place when executed
Manage device：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information divided
From；
Time-frequency spectrum is obtained using the medium-high frequency information, the abscissa of the time-frequency spectrum indicates that date, ordinate indicate institute
The frequency of medium-high frequency information is stated, each location point in the time-frequency spectrum corresponds to the energy value of corresponding medium-high frequency information, institute
It states the equal adjacent spots of energy value in time-frequency spectrum and is linked to be closed curve.
According to another aspect of the present invention, a kind of computer readable storage medium is additionally provided, computer is stored thereon with
Executable instruction, wherein the computer executable instructions make processor when being executed by processor：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information divided
From；
Time-frequency spectrum is obtained using the medium-high frequency information, the abscissa of the time-frequency spectrum indicates that date, ordinate indicate institute
The frequency of medium-high frequency information is stated, each location point in the time-frequency spectrum corresponds to the energy value of corresponding medium-high frequency information, institute
It states the equal adjacent spots of energy value in time-frequency spectrum and is linked to be closed curve.
Compared with prior art, the present invention is by recursion low-pass filter, after isolating low-frequency information, in time-frequency spectrum,
Frequency is confined in medium-high frequency limited range, solves the problems, such as the low frequency of data, can be obtained in generating needed for time-frequency spectrum
High-frequency information.Time-frequency spectrum is established using recursion wavelet algorithm, solves the instant sex chromosome mosaicism of securities market data.Additionally by described
Dynamic time-frequency spectrum carries out difference processing, overcomes the problems, such as that the time-frequency spectrum precision caused by Heisenberg's uncertain factor is not high.
Description of the drawings
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, of the invention other
Feature, objects and advantages will become more apparent upon：
Fig. 1 shows the securities market data Dynamic High-accuracy time-frequency spectrum of one embodiment of the invention；
Fig. 2 shows the side of one embodiment of the invention plain (stock code 600516) Dynamic High-accuracy time-frequency spectrum dominant frequency of big charcoal to pick up
Take figure；
Fig. 3 shows 38 stock targets according to an embodiment of the invention pond real-time deal record figure；
Fig. 4 shows 38 stock target pond automated transaction fund curves and its exponential curve of one embodiment of the invention
Figure.
Same or analogous reference numeral represents same or analogous component in attached drawing.
Specific implementation mode
Present invention is further described in detail below in conjunction with the accompanying drawings.
In a typical configuration of this application, terminal, the equipment of service network and trusted party include one or more
Processor (CPU), input/output interface, network interface and memory.
Memory may include computer-readable medium in volatile memory, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology realizes information storage.Information can be computer-readable instruction, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase transition internal memory (PRAM), static RAM (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), fast flash memory bank or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storages, magnetic tape cassette, magnetic tape disk storage or other magnetic storage apparatus or
Any other non-transmission medium can be used for storage and can be accessed by a computing device information.As defined in this article, computer
Readable medium does not include non-temporary computer readable media (transitory media), such as data-signal and carrier wave of modulation.
Conventional quantization trading strategies carry out model foundation on single time dimension, and there are the theoretical systems that history cannot recur
Count error so that the parameter in model sample can not be extrapolated to outside sample.It, need to be from securities market data in order to avoid this kind of error
Essence set out it studied.It is well known that either Value investment or short-term opportunistic practice, it is necessary to be low spot
Buying climax is sold, and is otherwise exactly the transaction to fail.Securities market data is a kind of time-varying non-stationary signal for essence,
With wave characteristic, following market height is determined by frequency parameter, if having made the master changed over time clear
Frequency information has also just grasped the height of market.And dominant frequency information is unrelated with history, there is no the vacations that history can recur
If therefore being the breakthrough merchandised to conventional quantization and overturning.In the processing for non-stationary signal, time frequency analysis is a weight
It is exactly non-stationary signal to want field, main study subject.By establishing Dynamic High-accuracy time-frequency spectrum, may be implemented to Number
According to dynamic extrapolation analysis.
The task of time frequency analysis is how the frequency spectrum of description signal changes in time, studies and understands time varying spectrum in number
The correspondence between physics is learned, suitable time-frequency distributions is constructed and carries out appropriate processing, reaches unlike signal processing
Purpose.Therefore, finding suitable, function admirable time-frequency distributions becomes an important research of Non-stationary Signal Analysis and processing
Content.Currently, the research common method of non-stationary signal has short time discrete Fourier transform, Eugene Wigner-Weir distribution, Koln class etc.,
Different analysis methods have the characteristics that different.The spectrogram that short time discrete Fourier transform is set up be it is most simple, most intuitive one
Kind time-frequency distributions, but when analyzing non-stationary signal, time frequency resolution is unable to adaptively changing.For one Weir of Eugene Wigner point
Cloth and Koln class can accurately estimate that the time-frequencies such as signal transient frequency, instant bandwidth are joined although having good time-frequency characteristic
Number, but due to there are cross-interference terms, influencing their practical ranges.Wavelet analysis is that have in Fourier analysis development history
There is the progress of milestone significance.In recent years, with the development of wavelet theory and application, the mathematical theory and method of wavelet analysis are got over
More to cause the extensive attention of people.Wavelet analysis is it in time domain and frequency domain while having better than the place of Fourier transform
Good localization property.Due to it to radio-frequency component using gradually fine time-domain samples step-length, so as to focus on pair
The arbitrary details of elephant, so it is described as school microscop by people.
Whether can accurately determine that signal can be divided into determining signal and random signal according to the value of signal at any time.
For random signal, the statistic of signal plays an important role.Most common statistic has first order statistic, such as average
It is worth (mean value)；Second-order statistic, such as correlation function, power spectral density, in addition there are the High Order Moments such as three ranks, quadravalence, high-order are tired
The high-order statistics such as accumulated amount and higher-order spectrum.
Time frequency analysis is the most basic content of Non-stationary Signal Analysis, basic task be exactly establish one with time t and
Frequencies omega is the two-dimentional joint distribution function P (t. ω) of variable, abbreviation time-frequency distributions.Using the distribution function P t. ω), can be with
The frequency of a certain determination and the Energy distribution of time range are asked, the density in the frequency of a certain specific time is calculated, calculates this point
Each rank square of entirety and part of cloth, such as average condition frequency and its local expansion, are discussed non-flat using this time-frequency distributions
Frequency dependent characteristic when steady signal, such as instantaneous frequency, instant bandwidth, group delay.
For a long time, in terms of various signals and data processing, especially in spectrum analysis, Fourier transform is most base
This one of tool.By the development in a century, it is the most powerful that classical Fourier transform has become field of signal processing
Analysis method and tool, this is mainly determined by its orthogonality and distinct physical significance and quickly succinct computational methods
Fixed.
From the definition of Fourier transform
F (ω) is the transform of f (t) (stock curve) in formula, and f (t) is the picture original function of F (ω), and t indicates that the time becomes
Amount, ω indicate frequency variable.
As can be seen that general function (or signal) is expressed as having the harmonic function of different frequency by Fourier transform
Linear superposition, to convert the research of original function (in time domain or spatial domain) to the weight coefficient research to this superposition,
That is, the research to Fourier transform (in frequency domain).In short, for any irregular curve, such as stock
Curve can be decomposed into a series of sine wave regular curve of various amplitude different frequencies by Fourier transform, wherein shaking
This maximum sine curve is main cyclic curve.For securities data, if primary period curve obtains, go
The height of feelings is also known that.
Although Fourier transform has an enormous advantage, there are congenital defects in terms of handling non-stationary signal for it.It
It is at the time of a segment signal can only be obtained generally comprising which frequency content, but occurred to each ingredient and ignorant.Namely use Fu
Vertical leaf transformation can only obtain frequency spectrums of the signal f (t) in (∞ | ∞) time range, and be difficult to understand signal in certain time range
Interior property, i.e., it is infinite to the resolution ratio of frequency, and is then zero to the resolution of time, cannot be simultaneously to time and frequency
With preferable resolution ratio, this is because Fourier transform is quadratured to variable t, the time-varying letter in non-stationary signal is eliminated
Number, it is only suitable in deterministic stationary signal, and is then difficult to fully portray to time-varying non-stationary signal, while Fourier transform
Resolution ratio to time domain is also constant on any section, thus is not enough to that signal is described or determined in arbitrarily small range
f(t)。
However, it is directed to securities data, characteristic of the exactly signal that we are concerned about in subrange.Such as extrapolate in firm offer
In, the position of signals of interest dominant frequency mutation, and it is desirable that know frequency content of the signal corresponding to the mutation moment.Obviously, at this
Kind in the case of, Fourier transform can not meet demand, Fourier transform using the trigonometric function of endless be used as basic function, accumulate
It is allocated as with the smooth mutagenic components of non-stationary process, to reflect the feature on regional area, therefore cannot be used for part point
Analysis.
For the jump signal and non-stationary signal of similar securities data etc., new analysis method need to be sought, make it can
The advantages of keeping Fourier transform, and can overcome the disadvantages that its deficiency.Wavelet transformation develops rapidly in this background
A kind of new theory.
" small echo " is exactly as its name suggests small waveform.So-called small, referring to it has Decay Rate, and is referred to as " wave ", then
It is its fluctuation, i.e., its amplitude is in the oscillation form of alternate positive and negative.The basic thought of wavelet transformation is from the flexible of function
With translation.It is to meet condition by one
Function Ψ (t), the family of functions generated by translating and stretching
Wherein, Ψ (t) is mother wavelet, and a is contraction-expansion factor (also referred to as scale factor), and b is shift factor.
It is empty that wavelet transformation constitutes a series of different rectangular projection of resolution ratio by the flexible and translation of wavelet basis function
Between and its corresponding base, then gone to indicate or approach a certain signal or function with this group of base.There are two variable, scales for wavelet transformation
Factor a and shift factor b, flexible, the translation of shift factor b control wavelet functions of scale factor a control wavelet functions.Scale
Corresponding to frequency (inverse ratio), translational movement corresponds to the time.
Different from the trigonometric function base of the endless in Fourier transform, the base of wavelet transformation is time-limited to decay
Wavelet basis can not only obtain frequency in this way, can also navigate to the time.That is, being not only known that signal with wavelet transformation
Including which frequency content, and it is known that these frequency contents existing specific location in the time domain.In quotation number
According in processing, position of the signal in each time domain is exactly that we are required comprising which frequency content.
Theoretically, time-frequency spectrum can be established by wavelet transformation.However in actual treatment, due to securities data
Time-varying is non-stationary, and time frequency analysis is there are prodigious technological difficulties, and technological difficulties are main existing for securities data time frequency analysis
Show as the following aspects：
(1) the low frequency problem of securities market data
Securities market data is fluctuated around a certain center line, so its frequency range span is very big, is not had
Fixed number, the thing on a not no boundary, is difficult to directly calculate time-frequency spectrum to obtain dominant frequency information；Securities market data packet simultaneously
The frequency contained is very low, can not accurately seek at all.
(2) the instant sex chromosome mosaicism of securities market data
The time-frequency window minimum of wavelet transformation is also to have length, and we are it is desirable that instantaneous frequency, certainly absolute meaning
The instantaneous frequency of justice is not present in fact, singly sees a signal value of a moment point, cannot get its frequency certainly, we
Only use the frequency of a very short segment signal as the frequency at the moment；Stock is the thing of instantaneity, once it is determined that thing
Cannot just change afterwards, it would be desirable to be the indeclinable time-frequency spectrum of history
(3) Heisenberg's uncertainty principle problem
Heisenberg uncertainty principle emphasizes that the time width of signal and bandwidth are impossible while arbitrary narrow, it is non-stationary signal
Very important theorem in analysis.Since the bridge between the time-domain representation of signal and frequency domain representation has been erected in Fourier transform,
Therefore, the time domain specification of signal and frequency domain characteristic are not independent from each other, but connect each other.Have when the duration of signal
In limited time, the frequency bandwidth of signal is then unlimited, conversely, when the infinite duration of signal, the frequency bandwidth of signal is then to have
Limit.That is, signal can not possibly have limited duration and limited frequency bandwidth simultaneously.
Wavelet transformation is still without departing from the constraint of Heisenberg's uncertainty principle.It, cannot be in time and frequency under certain scale
There is very high precision simultaneously in rate, since the precision of time-frequency spectrum is inadequate, cannot be satisfied the requirement of stock automatic operation.
The present invention provides a kind of time-frequency spectrum generation method, including：
Step S1 obtains securities market data；
Step S2, by recursion low-pass filter, by the low-frequency information and medium-high frequency information in the securities market data
It is detached；
Step S3 obtains time-frequency spectrum using the medium-high frequency information, wherein and the abscissa of the time-frequency spectrum indicates the date,
Ordinate indicates that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to corresponding medium-high frequency information
Energy value, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
Here, present embodiments providing above-mentioned securities market data low frequency solution to the problem, non-stationary is believed
For number, since its frequency changes over time, it is critically important to seek instantaneous frequency.And the frequency model of securities market data
Enclose that span is very big, including frequency it is very low, it is difficult to directly calculate time-frequency spectrum to obtain dominant frequency information.If low-frequency information regardless of
It separates out and, then the bad determination of time frequency analysis range, and time-frequency spectrum precision is relatively low.Therefore it by recursion low-pass filter, isolates
After low-frequency information, in time-frequency spectrum, frequency is confined in medium-high frequency limited range, solves the problems, such as the low frequency of data, can be with
It obtains generating the medium-high frequency information needed for time-frequency spectrum.
The dominant frequency range of movement in stock and share, is indicated with the primary period, in 0-50 weeks (week) range, is less than the frequency of this range
Rate information is defined as low-frequency information, and the frequency range of wavelet analysis is 0-50 weeks (week), during this frequency range information is defined as
High-frequency information.
Dominant frequency is sought by time-frequency spectrum, the inverse of dominant frequency is the period, and just there is height in the period, which is market
Dealing point.If time-frequency spectrum is accurate enough, dealing is profit, so the precision of time-frequency spectrum wants very high.
According to closed curve in the form of closed curve and position, the band width of the closed curve and the time-frequency spectrum
Around energy value form and position, can determine information when walking of stock or select stocks.Here it is so-called walk when, refer to
Determine the corresponding position of dynamic time-frequency spectrum upper time and frequency, i.e., the dominant frequency value sometime put.Dominant frequency position is determined, i.e.,
It can determine the height of securities market data, instruct the generation effectively merchandised.In dynamic time-frequency spectrum, the height of energy can be passed through
It is low to judge the position of dominant frequency.
The more empty transaction results formed according to low-frequency information in the present embodiment, total fund curve must be it is steady upward,
Both low frequency is solved the problems, such as, as well as the objectivity of algorithm, without artifical influence factor, for the income for pursuing steady in a long-term
It lays the foundation.
In one embodiment of time-frequency spectrum generation method of the present invention, step S2 is by recursion low-pass filter, by the security
After low-frequency information and medium-high frequency information in market data are detached, further include：
Step S21 seeks instantaneous frequency by Hilbert transform to the low-frequency information after the separation.
Here, in securities data processing, by recursion low-pass filter, by securities market data low frequency and middle height
Frequency is removed, and the single low-frequency information of narrower bandwidth, frequency content is isolated, these low-frequency informations are suitable for becoming by Hilbert
It brings and seeks instantaneous frequency, the trading instruction of low frequency signal can be formed according to the instantaneous frequency of acquisition.It is objective due to algorithm
Property, without artifical influence factor, the income to pursue steady in a long-term lays the foundation.
Relatively low by the curve frequencies obtained after low-pass filter, better regularity, frequency content is relatively single, bandwidth ratio
It is relatively narrow, seek instantaneous frequency suitable for passing through Hilbert transform；And medium-high frequency part, frequency content is abundant will to pass through time-frequency spectrum
Dominant frequency is sought in analysis.
Low frequency signal is extremely important for stock exchange, it is the basis of quantization transaction, and herein, low frequency signal is basis
What Hilbert transform was sought, human factor, and the trading instruction formed according to low frequency signal is not present, in the long term,
It is to stablize and get a profit, this is most important for the stability of system.This also differentiates it from traditional quantization transaction, conventional amounts
The parameter for changing transaction is fitted according to artificial adjustment, and there are unstability.
In one embodiment of time-frequency spectrum generation method of the present invention, step S21, by Hilbert transform, to the separation
Low-frequency information afterwards seeks instantaneous frequency, including：
Step S211 carries out Hilbert transform to low-frequency information：zn=hn*yn
Wherein,
ynFor the low-frequency information of the securities market data after low-pass filtering；
Step S212, construction securities market data complex signal sn=yn+izn, at any one moment, read instantaneous phase
Position is
Step S213 seeks instantaneous frequency μ according to instantaneous phasen=θn-θn-1。
Here, if without low-frequency information, the wave characteristic of long period information cannot be held, the inverse of instantaneous frequency is
For long period, instantaneous frequency has been sought, that is, has grasped macrocyclic fluctuation pattern.
Pass through Hilbert transform so that definition to the instantaneous parameters of sophisticated signal and be calculated as possibility, Neng Goushi
The now extraction of instantaneous signal truly.Hilbert transform approximate can only be applied to narrow band signal, that is, be only applicable to band
The relatively narrow signal of width.Relatively low by the curve frequencies obtained after low-pass filter, better regularity, frequency content is relatively single, band
Width is relatively narrow, and instantaneous frequency is sought suitable for passing through Hilbert transform.
In one embodiment of time-frequency spectrum generation method of the present invention, the recurrence formula of the low-pass filter is as follows：xn=q ×
Fn-q×Fn-1+q×xn-1；
Wherein,
f0For frequency filtering, xnFor a recursion median, FnFor discrete securities market data, initial value F0It is risen for data
Initial value, ynFor the low-frequency information of the securities market data after low-pass filtering, xnAnd ynInitial value be 0.
Here, for for time-frequency spectrum, the primary period is higher than the information of 50 weeks (week), is not suitable for wavelet analysis；And it is right
For quantization transaction, these low-frequency informations are useful, and needs are sought.Hilbert transform is only applicable to low frequency letter
Breath, and it is not suitable for medium-high frequency information, so being directed to low-frequency information and medium-high frequency information, Hilbert transform can be respectively adopted
It is sought with wavelet analysis, to improve the precision of quantization transaction.
In one embodiment of time-frequency spectrum generation method of the present invention, step S3 obtains time-frequency spectrum using the medium-high frequency information,
Including：
Step S31 obtains dynamic time-frequency spectrum using recursion wavelet algorithm according to the medium-high frequency information.
Here, separate the securities data after low frequency is changed into stationary signal by non-stationary signal, at this point, the present embodiment
Change traditional wavelet method, time-frequency spectrum is established using recursion wavelet algorithm, solves the instant sex chromosome mosaicism of securities market data.
Small echo is small again, also has length (minimum zone is -5 to+5, totally 10 points), therefore seek time-frequency spectrum with Traditional Wavelet
When, the history value of spectrum can change over time and change.And stock is the thing of instantaneity, transaction once it is determined that, just can not afterwards
To change, it is desirable that the indeclinable time-frequency spectrum of history.Using recursion small echo, this can be solved the problems, such as.
In one embodiment of time-frequency spectrum generation method of the present invention, step S31, using recursion wavelet algorithm, in described
High-frequency information obtains dynamic time-frequency spectrum, including：
Step S311 determines that the Complex wavelet that can be used for generating time-frequency spectrum, function expression are according to medium-high frequency informationWherein, coefficient ω0=2 π, δ=π, i are complex symbol, and t is the time；
Step S312, by the Complex wavelet function of time-frequency spectrumObtain securities market data FnWavelet transformation：
Wherein, WfFor wavelet coefficient, f is the frequency analysis range of time-frequency spectrum, τ=- 5, -4, -3, -2, -1,0,1,2,3,
4,5, s be scale variable；It is morther wavelet discrete series, it reflects FnInContent；
Step S313, according to the transform of sequence and its convolution property, securities market data Fn, morther wavelet sequence
And wavelet coefficient sequence WfThe transform of (s, n) is：
Wherein,
Step S314 is enabledTo the morther wavelet sequence
Transform is carried out, is obtained：
Wherein,
Step S315, willW (z) is substituted into obtain：
Then
Step S316 carries out Z inverse transformations to formula (6) and obtains recurrence formula：
Step S317, according to the W of formula (7)f(s, n) calculate separately W (s, n-1), W (s, n-2), W (s, n-3), W (s,
N-4), individual event recurrence calculation goes out all wavelet coefficient sequence W to the rightf(s, n)；
Here, according to existing historical quotes data, be calculated 4 initial value W (s, n-1) of wavelet transformation, W (s, n-2),
W (s, n-3), W (s, n-4) only individual event recursion can need to calculate all wavelet coefficient sequences to the right later；
Step S318, according to all wavelet coefficient Wf(s, n) obtains the dynamic time-frequency spectrum.
In one embodiment of time-frequency spectrum generation method of the present invention, step S31, using recursion wavelet algorithm, in described
After high-frequency information obtains dynamic time-frequency spectrum, further include：
S32 composes the dynamic time-frequency and carries out difference processing, is composed with obtaining high-precision dynamic time-frequency.
Here, the present embodiment dynamic data is difficult to seek the higher time-frequency spectrum of precision, the present embodiment algorithmically, using not
Same difference processing improves time-frequency spectral sensitivity and accurately obtains dominant frequency information in conjunction with bandwidth information, overcomes because of Heisenberg as possible
The not high problem of time-frequency spectrum precision caused by uncertain factor obtains dominant frequency for automation in next step and lays the foundation.
In one embodiment of time-frequency spectrum generation method of the present invention, S32 composes the dynamic time-frequency and carries out difference processing, with
High-precision dynamic time-frequency spectrum is obtained, including：
It is 1-50 that difference step size, which is respectively adopted, to dynamic time-frequency spectrum, 50 difference summation process is carried out, to obtain height
The dynamic time-frequency of precision is composed.
Here, the present embodiment dynamic data is difficult to seek the higher time-frequency spectrum of precision, the present embodiment algorithmically, using not
Same difference step size (difference step size 1-50) seeks energy gradient spectrum by 50 difference, improves time-frequency spectrum spirit as possible
Sensitivity accurately obtains dominant frequency information in conjunction with bandwidth information, overcomes the time-frequency spectrum precision caused by Heisenberg's uncertain factor
Not high problem obtains dominant frequency for automation in next step and lays the foundation.
In one embodiment of time-frequency spectrum generation method of the present invention, step S3 obtains time-frequency spectrum using the medium-high frequency information
Including：
Using the medium-high frequency information, the positive frequency that basic frequency in time-frequency spectrum was 0 to 50 weeks is obtained；
Using low-frequency information remaining near the medium-high frequency information, it is 0 Dao minus 50 weeks to obtain basic frequency in time-frequency spectrum
The negative frequency in (week, week).
Here, the time-frequency spectrum includes positive frequency and negative frequency, the analyst coverage in the negative frequency corresponding primary period is 0
To minus 50 weeks (week).
As shown in Figure 1, negative frequency concept is introduced in the time-frequency spectrum, the analyst coverage in primary period corresponding with negative frequency
Minus 50 weeks (week) is arrived for 0, the primary period analyst coverage of medium-high frequency information is 0 to 50 all (week).Negative frequency is act as：
(1) more accurately the time is sold in determination；(2) although having done low frequency separation, some still are located at the low frequency near medium-high frequency information
In the medium-high frequency information of information residual after isolation, these low-frequency informations for being higher than 50 weeks (week) can be rolled into negative frequency range
It is interior so that Time-frequency Spectrum Analysis signal is more comprehensive, more accurately holds market trend.
Specifically, when Fig. 1 is the Dynamic High-accuracy of Chinese international trade (stock code 600007) market data that processing obtains
Frequency spectrum.Horizontal axis represents the time in figure, and the longitudinal axis represents frequency.Since low-frequency information has been stripped, therefore during time frequency analysis only reflects
High-frequency information, frequency analysis range is also within the limited range of medium-high frequency.In time-frequency spectrum, energy ladder can be reflected with color
Degree.For example, from blue to orange, energy is gradually incremented by.In theory, the maximum position of energy is the dominant frequency position of data
It sets, wherein what the form of trap class represented is the information of band width.Pushed away as the date prolongs backward, time-frequency spectrum therewith dynamic backward
Extend, and the time-frequency spectrum of generating portion remains stationary as namely history will not be changed.
Due to Dynamic High-accuracy time-frequency spectrum of the present invention foundation there is no History repeats itself it is assumed that therefore application high-precision is dynamic
The external recursion of dynamic of securities data may be implemented in state time-frequency spectrum, is truly realized results of measuring inside and outside sample and links up unanimously, breaks through
With the pattern for overturning conventional quantization transaction.It, can by identifying the specific modality of bandwidth variation in time-frequency spectrum in dynamic time-frequency spectrum
To obtain accurate primary period information, real automated transaction is realized.
By taking the 38 stock targets randomly selected as an example, its group is built up into a target pond, in the time-frequency spectrum of each target
Full-automatic dynamic dominant frequency pickup is carried out, one of target Dynamic High-accuracy time-frequency spectrum is illustrated in figure 2.It is each perpendicular in spectrum in figure
Line is the dominant frequency initial position of automatic Picking in extrapolation process, and with the switching of dominant frequency, pickup point automatically switches, under time-frequency spectrum
Wherein single line is the energy curve obtained according to Dynamic High-accuracy time-frequency spectrum.It is according to the dynamic dominant frequency of pickup as a result, raw in real time
At transaction record, as shown in figure 3, realizing the automated transaction of securities data.Fund curve after automated transaction is complete, it is steady on
It pushes away, as shown in figure 4, horizontal axis is automated transaction curve monthly returns, the longitudinal axis is accumulated earnings ratio, really realizes quantization transaction mesh
Mark.
According to another aspect of the present invention, a kind of time-frequency spectrum generation equipment is additionally provided, which includes：
Acquisition device, for obtaining securities market data；
Separator, by recursion low-pass filter, by the low-frequency information and medium-high frequency letter in the securities market data
Breath is detached；
Time-frequency spectral apparatus, for obtaining time-frequency spectrum using the medium-high frequency information, wherein the abscissa table of the time-frequency spectrum
Show that date, ordinate indicate the frequency of the medium-high frequency information, during each location point in the time-frequency spectrum corresponds to accordingly
The energy value of high-frequency information, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
According to another aspect of the present invention, a kind of equipment based on calculating is additionally provided, wherein including：
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the place when executed
Manage device：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information divided
From；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates date, ordinate table
Show that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to the energy of corresponding medium-high frequency information
It is worth, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
According to another aspect of the present invention, a kind of computer readable storage medium is additionally provided, computer is stored thereon with
Executable instruction, wherein the computer executable instructions make processor when being executed by processor：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information divided
From；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates date, ordinate table
Show that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to the energy of corresponding medium-high frequency information
It is worth, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
For details, reference can be made to each realities of method for the detailed content of each embodiment of equipment and computer readable storage medium of the present invention
The corresponding part of example is applied, details are not described herein.
Obviously, those skilled in the art can carry out the application essence of the various modification and variations without departing from the application
God and range.In this way, if these modifications and variations of the application belong to the range of the application claim and its equivalent technologies
Within, then the application is also intended to include these modifications and variations.
It should be noted that the present invention can be carried out in the assembly of software and/or software and hardware, for example, can adopt
With application-specific integrated circuit (ASIC), general purpose computer or any other realized similar to hardware device.In one embodiment
In, software program of the invention can be executed by processor to realize steps described above or function.Similarly, of the invention
Software program (including relevant data structure) can be stored in computer readable recording medium storing program for performing, for example, RAM memory,
Magnetic or optical driver or floppy disc and similar devices.In addition, hardware can be used to realize in some steps or function of the present invention, example
Such as, coordinate to execute the circuit of each step or function as with processor.
In addition, the part of the present invention can be applied to computer program product, such as computer program instructions, when its quilt
When computer executes, by the operation of the computer, it can call or provide according to the method for the present invention and/or technical solution.
And the program instruction of the method for the present invention is called, it is possibly stored in fixed or moveable recording medium, and/or pass through
Broadcast or the data flow in other signal loaded mediums and be transmitted, and/or be stored according to described program instruction operation
In the working storage of computer equipment.Here, including a device according to one embodiment of present invention, which includes using
Memory in storage computer program instructions and processor for executing program instructions, wherein when the computer program refers to
When order is executed by the processor, method and/or skill of the device operation based on aforementioned multiple embodiments according to the present invention are triggered
Art scheme.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power
Profit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent requirements of the claims
Variation includes within the present invention.Any reference signs in the claims should not be construed as limiting the involved claims.This
Outside, it is clear that one word of " comprising " is not excluded for other units or step, and odd number is not excluded for plural number.That is stated in device claim is multiple
Unit or device can also be realized by a unit or device by software or hardware.The first, the second equal words are used for table
Show title, and does not represent any particular order.
Claims (12)
1. a kind of time-frequency spectrum generation method, wherein this method includes：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information detach；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates that date, ordinate indicate institute
The frequency of medium-high frequency information is stated, each location point in the time-frequency spectrum corresponds to the energy value of corresponding medium-high frequency information, institute
It states the equal adjacent spots of energy value in time-frequency spectrum and is linked to be closed curve.
2. according to the method described in claim 1, wherein, by Hilbert transform, asking the low-frequency information after the separation
Take instantaneous frequency.
3. according to the method described in claim 2, wherein, by Hilbert transform, asking the low-frequency information after the separation
Instantaneous frequency is taken, including：
Hilbert transform is carried out to low-frequency information：zn=hn*yn
Wherein,
ynFor the low-frequency information of the securities market data after low-pass filtering；
Construct securities market data complex signal sn=yn+izn, at any one moment, reading instantaneous phase is
Instantaneous frequency μ is sought according to instantaneous phasen=θn-θn-1。
4. according to the method described in claim 1, wherein, the recurrence formula of the low-pass filter is as follows：
xn=q × Fn-q×Fn-1+q×xn-1；
Wherein,
f0For frequency filtering, xnFor a recursion median, FnFor discrete securities market data, initial value F0For data initial value,
ynFor the low-frequency information of the securities market data after low-pass filtering, xnAnd ynInitial value be 0.
5. according to the method described in claim 1, wherein, time-frequency spectrum is obtained using the medium-high frequency information, including：
Using recursion wavelet algorithm, dynamic time-frequency spectrum is obtained according to the medium-high frequency information.
6. according to the method described in claim 5, wherein, using recursion wavelet algorithm, being moved according to the medium-high frequency information
State time-frequency spectrum, including：
According to medium-high frequency information, determine that the Complex wavelet that can be used for generating time-frequency spectrum, function expression are
Wherein, coefficient ω0=2 π, δ=π, i are complex symbol, and t is the time；
By the Complex wavelet function of time-frequency spectrumObtain securities market data FnWavelet transformation：
Wherein, WfFor wavelet coefficient, f is the frequency analysis range of time-frequency spectrum, τ=- 5, -4, -3, -2, -1,0,1,2,3,4,5, s
For scale variable；It is morther wavelet discrete series, it reflects FnInContent；
According to the transform of sequence and its convolution property, securities market data Fn, morther wavelet sequenceAnd wavelet coefficient
Sequence WfThe transform of (s, n) is：
Wherein,
It enablesTo the morther wavelet sequence
Transform is carried out, is obtained：
Wherein,
λ1=-4eA, λ2=6e2A, λ3=-4e3A, λ4=e4A；
It willW (z) is substituted into obtain：
Then
Z inverse transformations are carried out to formula (6) and obtain recurrence formula：
According to formula (7) Wf(s, n) calculates separately W (s, n-1), W (s, n-2), W (s, n-3), W (s, n-4), to the right individual event recursion
Calculate all wavelet coefficient sequence Wf(s, n)；
According to all wavelet coefficient Wf(s, n) obtains the dynamic time-frequency spectrum.
7. according to the method described in claim 5, wherein, using recursion wavelet algorithm, being moved according to the medium-high frequency information
After state time-frequency spectrum, further include：
The dynamic time-frequency is composed and carries out difference processing, is composed with obtaining high-precision dynamic time-frequency.
8. according to the method described in claim 7, wherein, being composed to the dynamic time-frequency and carrying out difference processing, to obtain high-precision
Dynamic time-frequency spectrum, including：
It is 1-50 that difference step size, which is respectively adopted, to dynamic time-frequency spectrum, carries out 50 difference summation process, to obtain high-precision
Dynamic time-frequency spectrum.
9. according to the method described in claim 1, wherein, obtaining time-frequency spectrum using the medium-high frequency information includes：
Using the medium-high frequency information, the positive frequency of the time-frequency spectrum is obtained；
Using low-frequency information remaining near the medium-high frequency information, the negative frequency of the time-frequency spectrum is obtained.
10. a kind of time-frequency spectrum generates equipment, wherein the equipment includes：
Acquisition device, for obtaining securities market data；
Separator, by recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information into
Row separation；
Time-frequency spectral apparatus, for obtaining time-frequency spectrum using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates day
Phase, ordinate indicate that the frequency of the medium-high frequency information, each location point in the time-frequency spectrum correspond to corresponding medium-high frequency
The energy value of information, the equal adjacent spots of energy value are linked to be closed curve in the time-frequency spectrum.
11. a kind of equipment based on calculating, wherein including：
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the processing when executed
Device：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information detach；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates that date, ordinate indicate institute
The frequency of medium-high frequency information is stated, each location point in the time-frequency spectrum corresponds to the energy value of corresponding medium-high frequency information, institute
It states the equal adjacent spots of energy value in time-frequency spectrum and is linked to be closed curve.
12. a kind of computer readable storage medium, is stored thereon with computer executable instructions, wherein the computer is executable
Instruction makes the processor when being executed by processor：
Obtain securities market data；
By recursion low-pass filter, by the securities market data low-frequency information and medium-high frequency information detach；
Time-frequency spectrum is obtained using the medium-high frequency information, wherein the abscissa of the time-frequency spectrum indicates that date, ordinate indicate institute
The frequency of medium-high frequency information is stated, each location point in the time-frequency spectrum corresponds to the energy value of corresponding medium-high frequency information, institute
It states the equal adjacent spots of energy value in time-frequency spectrum and is linked to be closed curve.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810146215.2A CN108492179B (en) | 2018-02-12 | 2018-02-12 | Time-frequency spectrum generation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810146215.2A CN108492179B (en) | 2018-02-12 | 2018-02-12 | Time-frequency spectrum generation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108492179A true CN108492179A (en) | 2018-09-04 |
| CN108492179B CN108492179B (en) | 2020-09-01 |
Family
ID=63340314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810146215.2A Active CN108492179B (en) | 2018-02-12 | 2018-02-12 | Time-frequency spectrum generation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108492179B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112328956A (en) * | 2020-09-22 | 2021-02-05 | 东北大学 | Strong frequency variable signal time-frequency analysis method |
| CN112446329A (en) * | 2020-11-30 | 2021-03-05 | 广州大学 | Method, system, device and storage medium for determining instantaneous frequency of time-varying structure |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104936088A (en) * | 2015-04-21 | 2015-09-23 | 上海大学 | Mixed virtual bass enhancing method |
| US9472498B2 (en) * | 2008-01-09 | 2016-10-18 | Oracle International Corporation | Multiple access over proximity communication |
| US9576583B1 (en) * | 2014-12-01 | 2017-02-21 | Cedar Audio Ltd | Restoring audio signals with mask and latent variables |
| CN106771590A (en) * | 2017-01-12 | 2017-05-31 | 中南大学 | The method and device that effective information is extracted in a kind of active periodic signal |
| CN107515421A (en) * | 2017-08-15 | 2017-12-26 | 中国石油化工股份有限公司江汉油田分公司物探研究院 | Spectral imaging method based on wavelet package transforms |
-
2018
- 2018-02-12 CN CN201810146215.2A patent/CN108492179B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9472498B2 (en) * | 2008-01-09 | 2016-10-18 | Oracle International Corporation | Multiple access over proximity communication |
| US9576583B1 (en) * | 2014-12-01 | 2017-02-21 | Cedar Audio Ltd | Restoring audio signals with mask and latent variables |
| CN104936088A (en) * | 2015-04-21 | 2015-09-23 | 上海大学 | Mixed virtual bass enhancing method |
| CN106771590A (en) * | 2017-01-12 | 2017-05-31 | 中南大学 | The method and device that effective information is extracted in a kind of active periodic signal |
| CN107515421A (en) * | 2017-08-15 | 2017-12-26 | 中国石油化工股份有限公司江汉油田分公司物探研究院 | Spectral imaging method based on wavelet package transforms |
Non-Patent Citations (1)
| Title |
|---|
| 钟强: "《工程科技Ⅱ辑》", 31 March 2017 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112328956A (en) * | 2020-09-22 | 2021-02-05 | 东北大学 | Strong frequency variable signal time-frequency analysis method |
| CN112328956B (en) * | 2020-09-22 | 2024-01-26 | 东北大学 | Strong frequency variable signal time-frequency analysis method |
| CN112446329A (en) * | 2020-11-30 | 2021-03-05 | 广州大学 | Method, system, device and storage medium for determining instantaneous frequency of time-varying structure |
| CN112446329B (en) * | 2020-11-30 | 2023-08-08 | 广州大学 | Time-varying structure instantaneous frequency determining method, system, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108492179B (en) | 2020-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wild et al. | The evolution of post-starburst galaxies from z= 2 to 0.5 | |
| Van de Vrande et al. | How prior corporate venture capital investments shape technological alliances: A real options approach | |
| Margalef-Bentabol et al. | The formation of bulges, discs and two-component galaxies in the CANDELS Survey at z< 3 | |
| Stevens et al. | Mechanisms for stable, robust, and adaptive development of orientation maps in the primary visual cortex | |
| Cheng et al. | Beyond the hubble sequence–exploring galaxy morphology with unsupervised machine learning | |
| Smith-Greenaway et al. | Maternal cumulative prevalence measures of child mortality show heavy burden in sub-Saharan Africa | |
| Kultur et al. | Ensemble of neural networks with associative memory (ENNA) for estimating software development costs | |
| CN108510204A (en) | Methods of marking, device and electric terminal | |
| CN108492179A (en) | Time-frequency spectrum generation method and equipment | |
| CN108416674A (en) | The application process and equipment of time-frequency spectrum | |
| Rockart et al. | Learning in cycles | |
| Clark et al. | Stages of diversification in Africa | |
| Kim | Tackling false positives in business research: A statistical toolbox with applications | |
| Adeola et al. | Tourism and economic wellbeing in Africa | |
| Haga et al. | A neural network approach to measure real activities manipulation | |
| Arpaci‐Ayhan | Foreign aid as a catalyst for improving productive capabilities in recipients | |
| CN108053266A (en) | A kind of patent value predictor method and device | |
| KR20220027606A (en) | Method for predicting default rates for each macro-scenario | |
| Mellers et al. | Predicting the future with humans and AI | |
| Follett et al. | Achieving parsimony in bayesian VARs with the horseshoe prior | |
| van Weert | Predicting Academic Success Using Academic Genealogical Data, a Data Science Approach | |
| Sandbhor et al. | Effect of training sample and network characteristics in neural network-based real property value prediction | |
| Luo et al. | A study on stock ranking and selection strategy based on UTA method under the condition of inconsistence | |
| Baloul et al. | Fuzzy academic advising system for on probation students in colleges of applied sciences—Sultanate of Oman | |
| Ouyang | Time series forecasting: from econometrics to deep learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |