CN108733921A - Coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation - Google Patents
Coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation Download PDFInfo
- Publication number
- CN108733921A CN108733921A CN201810481085.8A CN201810481085A CN108733921A CN 108733921 A CN108733921 A CN 108733921A CN 201810481085 A CN201810481085 A CN 201810481085A CN 108733921 A CN108733921 A CN 108733921A
- Authority
- CN
- China
- Prior art keywords
- granulation
- neural network
- hot point
- fuzzy information
- fluctuation range
- 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
- 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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation that the invention discloses a kind of includes the number of window when being determined to be granulated according to Fuzzy Information Granulation quality, the effective information of initial data is extracted with Fuzzy Information Granulation technology；Prediction model based on wavelet neural network is built based on effective information, and the structural parameters of wavelet neural network are optimized by harmonic search algorithm；Coiling hot point of transformer temperature fluctuation range is predicted with the model after parameter adjustment using structure.The predictablity rate of the present invention is higher, has certain directive significance to the operation and maintenance of transformer.
Description
Technical field
The coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation that the present invention relates to a kind of.
Background technology
Power transformer is highly important power transmission and transforming equipment, is all the object of emphasis monitoring and protection all the time.Funeral
The main reason for due insulating capacity is most of transformer life termination is lost, and coiling hot point of transformer temperature is to influence
One of an important factor for its insulating capacity.Hot spot temperature of winding is defined as the temperature that transformer winding most thermal region reaches, by
Confirmation is to influence the key factor of transformer load ability and probable life.When coiling hot point of transformer temperature is more than permissible value
It can lead to transformer insulated damage.Therefore the hot(test)-spot temperature of in-depth study transformer, finds coiling hot point of transformer in advance
The exception of temperature, to instructing the operation and maintenance of transformer to have a very important significance.
Domestic and international experts and scholars expand a large amount of research with regard to the prediction of coiling hot point of transformer temperature.Have in existing method
The transformer temperature of top prediction technique proposed based on T-S fuzzy models, model preferably complete temperature of top with
Track, and being compared with Recurrent Neural Network, IEEE directive/guides, but temperature of top is not real hot(test)-spot temperature, therefore model
With certain limitation.Some documents consider environment temperature, top-oil temperature, bottom oil temperature, upper dead angle temperature, lower dead angle temperature
5 influence factors are spent, the prediction model of coiling hot point of transformer temperature is established based on generalized regression nerve networks.Some documents
Load change and top-oil temperature are considered, coiling hot point of transformer temperature real-time estimation mould is established based on Kalman filtering algorithm
Type.Some documents construct transformer temperature of top prediction model based on least square method supporting vector machine.Some documents will load
Electric current, environment temperature, top-oil temperature, bottom oil temperature, upper dead angle temperature, lower dead angle temperature are taken into account, and genetic optimization is constructed
The hot(test)-spot temperature prediction model of support vector machines.
The studies above preferably realizes the tracking for transformer winding temperature, however these models predict change in advance
The ability of depressor hot spot temperature of winding exception is relatively poor.
Invention content
The present invention is to solve the above-mentioned problems, it is proposed that a kind of coiling hot point of transformer temperature based on Fuzzy Information Granulation
Fluctuation range prediction technique, it is relatively incomplete for the prediction of coiling hot point of transformer temperature sequence and fluctuation range forecasting research
Present situation, based on Fuzzy Information Granulation and harmonic search algorithm (Harmony Search, HS) Optimization of Wavelet neural network
The coiling hot point of transformer temperature fluctuation range combination forecasting of (Wavelet neural network, WNN), energy of the present invention
The ability of enough preferable precognition coiling hot point of transformer temperature anomaly.
To achieve the goals above, the present invention adopts the following technical scheme that：
A kind of coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation, including:
The number that window when being granulated is determined according to Fuzzy Information Granulation quality, original number is extracted with Fuzzy Information Granulation technology
According to effective information；
Prediction model based on wavelet neural network is built based on effective information, and by harmonic search algorithm to wavelet neural network
Structural parameters optimize；
Coiling hot point of transformer temperature fluctuation range is predicted with the model after parameter adjustment using structure.
In more detail, including:
Update historical data, to historical data sample data carry out Fuzzy Information Granulation, obtain average value, lower bound and on
Boundary calculates Fuzzy Information Granulation quality factor, chooses optimal granulation result；
Historical data after screening granulation, the wavelet neural network of a variety of different structures is tentatively established with training set sample
Prediction model judges input layer number and hidden layer number of nodes by the model, and timing updates；
The structural parameters that input layer, hidden layer and output layer determine that needs optimize are determined, optimization object is carried out
Coding defines fitness value function, is screened to the structural parameters of wavelet neural network by harmony search；
Average value, lower bound and the upper bound are predicted respectively using the wavelet neural network after screening, obtain transformer around
Group hot(test)-spot temperature fluctuation range；
Model prediction performance is assessed using three mean square error, mean absolute error and related coefficient indexs.
Further, it during the effective information that initial data is extracted with Fuzzy Information Granulation technology, specifically includes：
Coiling hot point of transformer temperature-time sequence resolves into several small subsequences on demand, each subsequence conduct
Window is blurred when one operation；
A triangular form obscure particle is established on each subsequence；
By time series by sorting from small to large, average value, lower bound and the upper bound of each triangular form obscure particle are sought.
Further, during extracting the effective information of initial data with Fuzzy Information Granulation technology, mould is utilized
The number of window when Information Granulating quality factor determine granulation is pasted, the quality factor are specially the data compression of Fuzzy Information Granulation
Than the log values with the amount lost ratio of data after Fuzzy Information Granulation.
Further, it is based in effective information structure prediction model based on wavelet neural network, using Morlet mother wavelet functions,
Using the coiling hot point of transformer temperature history after fuzzy granulation as the input of the input layer of wavelet neural network
Data.
Further, it is searched for using harmonic search algorithm in best wavelet neural network structural parameters, by wavelet neural
The prediction error of network model calls wavelet neural network to be predicted, obtained prediction result conduct as object function
The fitness value of harmony, it would be desirable to input weight, output weight, contraction-expansion factor and the shift factor of the wavelet neural network of optimization
Position vector as each harmony.
Further, Optimization Steps include：
(1) it changes to harmony memory storage capacity, data base probability, tone fine tuning probability, tone fine tuning broadband and maximum
Generation number is initialized；
(2) multiple harmony are generated, initial harmony library is constituted, wavelet neural network is called to be predicted, prediction is calculated and misses
Difference obtains each individual fitness value in original data base；
(3) 0~1 random number is generated at random, if meeting the random number is less than data base probability, according to from note
Recall and randomly selects an individual in library；
(4) individual of selection is finely adjusted；If being unsatisfactory for the random number is less than data base probability, becoming
A new explanation is regenerated in the value range of amount；
(5) harmony data base is updated, the fitness value of generated new explanation is calculated, and carries out the update of data base；
(6) judge whether algorithm terminates, algorithm terminates if meeting maximum iteration, obtains optimal Wavelet Neural Network
Otherwise network structural parameters are back to step (3) and continue to execute.
Further, in the step (3), the process that an individual is randomly selected from data base includes：If with
Machine number is less than data base probability, and new individual is random one in harmony, and otherwise, new individual is in harmony vector
One.
Further, in the step (4), if the random number is less than data base probability, new explanation is：
Rand2 is the random number between 0~1, and k is the number of discrete variable probable value, and PAR is that tone finely tunes probability.
Further, in the step (5), the fitness value of new explanation is calculated, data base is carried out according to fitness value
Update：
xworstFor the worst solution that solution is concentrated, function expression method f () here is object function.
Compared with prior art, beneficial effects of the present invention are：
(1) Information Granulating technology can extract useful information from initial data on demand, reduce answering for target data
Miscellaneous degree, and wavelet neural network has very strong nonlinearity mapping ability.Two methods are combined and effectively realize by the present invention
Prediction for coiling hot point of transformer temperature fluctuation range.
(2) Information Granulating process is optimized in the concept of the Fuzzy Information Granulation quality proposed；It is searched for using harmony
Algorithm screens the structural parameters of wavelet neural network, improves prediction accuracy, and carry out with a variety of prediction models
Comparison demonstrates the superiority of the put forward model of the present invention, demonstrates FIG-HS-WNN model for transformer coiling hotspot temperature
Spend the feasibility of prediction.
(3) predictablity rate of the present invention is higher, has certain directive significance to the operation and maintenance of transformer.The combination
Model can be used not only for the prediction of coiling hot point of transformer temperature, moreover it is possible to provide thinking for the prediction modeling of other field.
Description of the drawings
The accompanying drawings which form a part of this application are used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation do not constitute the improper restriction to the application for explaining the application.
Fig. 1 is the wavelet neural network topology diagram of the present invention；
Fig. 2 is the combination forecasting flow chart of the present invention；
Specific implementation mode：
The invention will be further described with embodiment below in conjunction with the accompanying drawings.
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms that the present invention uses have logical with the application person of an ordinary skill in the technical field
The identical meanings understood.
It should be noted that term used herein above is merely to describe specific implementation mode, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative
It is also intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or combination thereof.
In the present invention, term for example "upper", "lower", "left", "right", "front", "rear", "vertical", "horizontal", " side ",
The orientation or positional relationship of the instructions such as "bottom" is to be based on the orientation or positional relationship shown in the drawings, only to facilitate describing this hair
Bright each component or component structure relationship and the relative of determination, not refer in particular to either component or element in the present invention, cannot understand
For limitation of the present invention.
In the present invention, term such as " affixed ", " connected ", " connection " shall be understood in a broad sense, and indicate may be a fixed connection,
Can also be to be integrally connected or be detachably connected；It can be directly connected, it can also be indirectly connected through an intermediary.For
The related scientific research of this field or technical staff can determine the concrete meaning of above-mentioned term in the present invention as the case may be,
It is not considered as limiting the invention.
For the prediction of coiling hot point of transformer temperature sequence and the relatively incomplete present situation of fluctuation range forecasting research, propose
Based on Fuzzy Information Granulation and harmonic search algorithm (Harmony Search, HS) Optimization of Wavelet neural network (Wavelet
Neural network, WNN) coiling hot point of transformer temperature fluctuation range combination forecasting.
The number of window when being granulated is determined according to Fuzzy Information Granulation quality first, and then is carried with Fuzzy Information Granulation technology
Take the effective information of initial data；
It is then based on effective information structure prediction model based on wavelet neural network, and by harmonic search algorithm to wavelet neural
The structural parameters of network are optimized；
Finally, coiling hot point of transformer temperature fluctuation range is predicted with the model after parameter adjustment using structure.
Experiment shows that the model can be carried out effectively the prediction of coiling hot point of transformer temperature fluctuation range with data results,
There is certain directive significance to the O&M of transformer.
One Fuzzy Information Granulation
1.1 Fuzzy Information Granulation basic principles
Information Granulating proposes by professor Zadeh earliest, this method be one by large amount of complex information by certain regular break
For the process of multiple portions, there is lower computation complexity in each part.While being extracted raw information valid data
Realize the compression to raw information.Information Granulating mainly has based on fuzzy set theory, rough set theory and entropy Space Theory three
Type, the present invention use the model based on fuzzy set theory, study the fuzzy grain of coiling hot point of transformer temperature-time sequence
Change process.Fuzzy information granule is divided into two processes：(1) window when dividing granulation；(2) press certain rule to it is each when window in number
According to being blurred.
It is exactly that coiling hot point of transformer temperature-time sequence is resolved into several small sub- sequences on demand that window, which divides, when granulation
Row, window when each subsequence is as an operation.The blurring of information is the extraction process of an effective information, to it is each when
Data in window establish rational fuzzy set by certain rule.
Window carries out the case where Fuzzy processing when considering single, i.e., a coiling hot point of transformer temperature sequence T=
Window is blurred when [t1, t2 ..., tn] regards one as.The target of blurring is that an obscure particle P is established on T, i.e.,
A determining function A is found to make：
P=A (t), t ∈ T (1)
Obscure particle P mainly has the citation forms such as triangle, ladder type and asymmetric Gaussian, since the present invention carries out transformation
The prediction of device hot spot temperature of winding fluctuation range needs the maxima and minima for knowing window, therefore uses triangle pattern
Particle is pasted, shown in membership function such as formula (2)：
The process of blurring is to determine the value of LOW, R, UP, and the present invention uses W.Pedrycz algorithms：
(1) average R.By time series T by sorting from small to large, might as well set after sorting still as T, then R=median
(T)。
(2) lower bound LOW is sought：
(3) upper bound UP is sought：
(4) obscure particle P is provided：
P=(LOW, R, UP) (5)
The value that can determine parameter LOW, R, UP by the above process, data variation in window when wherein parameter LOW indicates each
Minimum value, when parameter R indicates each in window data variation average level, when parameter UP indicates each in window data variation maximum
Value.
1.2 Fuzzy Information Granulation qualities
In application Fuzzy Information Granulation, the number of window needs artificial given when granulation, inevitably with personal main
Color is seen, redundancy or the loss of information are easily caused.In order to solve this problem, the present invention propose Fuzzy Information Granulation quality because
Number, the number of window when being granulated from the angle-determining of more science.
If hot spot temperature of winding time series T=[t1, t2 ..., tn], LOW=after Fuzzy Information Granulation [a1, a2 ...,
Al], R=[m1, m2 ..., ml], UP=[b1, b2 ..., bl].
The data compression ratio of ambiguity in definition Information Granulating is：
CR=n/ (3l) (6)
In formula (6), n be original series T length, l LOW, R, UP length, the more big then level of data compression of CR more
It is high.
Comentropy can represent the number of information content entrained by one group of data to a certain extent.For this purpose, introducing information
Information content entrained by data before and after granulation is described in entropy.The comentropy of T, LOW, R, UP are solved, if respectively HT, HLOW,
HR, HUP, then the amount lost of data is defined as after Fuzzy Information Granulation：
S is bigger, and loss of data is more serious after showing Fuzzy Information Granulation.
Fuzzy Information Granulation quality factor are defined as：
Quality factor are higher, and Fuzzy Information Granulation result is more excellent, i.e., are found between data compression ratio and amount of data lost
One equalization point, it is desirable to which the result data amount lost after Fuzzy Information Granulation possesses certain compression ratio while less.
Two harmonic search algorithm Optimization of Wavelet neural networks
2.1 wavelet neural network
Wavelet neural network is the product that wavelet analysis is combined with neural network, inherits the time-frequency office of wavelet transformation
Portion's property, at the same have both the self study of neural network, Nonlinear Mapping ability.It is divided by the combination of the two, small echo god
There is loose type wavelet neural network through network and the type wavelet neural network that compacts, loose type wavelet neural network is to locate small echo in advance
Reason is presented with neural network with cascade, and the type that compacts wavelet neural network is then that complete neural network using wavelet transformation hidden
Hide the design of node layer.Present invention research is around type wavelet neural network expansion of compacting.
Due to the nonlinear characteristic of coiling hot point of transformer temperature, using wavelet neural network feature extraction with it is non-linear
Capability of fitting is suitable.The present invention uses the wavelet neural network of single layer output, network topology structure as shown in Figure 1.
In Fig. 1, { x1 ..., xi ..., xn } is the input data of the input layer of wavelet neural network, i.e., fuzzy granulation
Coiling hot point of transformer temperature history afterwards, y are the node prediction output data of the output layer of wavelet neural network, wij
For input layer to hidden layer weight coefficient, wj is hidden layer to output layer weight coefficient.Ψ is the excitation function for hiding node layer,
Here Morlet mother wavelet functions are used.
If i-th of node input is xi (i=1,2 ..., n), then the input value of j-th of hiding node layer is：
Data first carry out flexible and translation transformation after entering hiding node layer, then the input value of mother wavelet function is：
(10) in formula, aj is contraction-expansion factor, and bj is shift factor.Morlet mother wavelets function is as shown in (11) formula, then hidden
It hides node layer and exports result such as (12) formula.
The output node neuron of wavelet neural network uses Sigmoid functions, i.e., f (x)=1/ (1+e-x), then finally
Model prediction result be represented by (13) formula.
The prediction error of Definition Model is：
In formula (14), T is the number of training sample, and YP indicates that the measured data of the P sample, yP indicate the P sample
Prediction data.
Weight wij, wj, contraction-expansion factor aj, the shift factor bj of WNN is affected to prediction error J, it is therefore desirable to choose
Rational model parameter.Traditional wavelet neural network carries out parameter selection by the gradient descent method of additional momentum, makes (14)
Formula minimizes, however this method has that convergence rate is relatively slow, cannot effectively search global optimum.In order to effectively solve
The certainly problem, the present invention expand research to harmonic search algorithm.
2.2 HS optimize WNN structural parameters
Harmonic search algorithm simulates the principle of musical performance, which has stronger ability of searching optimum, and keeps away
Complicated parameter setting problem is exempted from, has all been shown than genetic algorithm, simulated annealing more in many optimization problems
Excellent performance.The present invention is optimized the structural parameters of wavelet neural network using the algorithm, promotes coiling hotspot temperature
Spend the accuracy of prediction.Document exists in the prior art the principle of harmonic search algorithm is described in detail, the present invention
It repeats no more.
Best wavelet neural network structural parameters are searched for using harmonic search algorithm, need to consider emphatically is fitness
The establishment of value function and the encoded question of harmony.The present invention is when being designed by the prediction error of wavelet-neural network model
As object function, that is, wavelet neural network is called to be predicted, using the result that formula (14) obtains as the fitness value of harmony.
To need the input weight wij optimized, output weight wj, contraction-expansion factor aj, shift factor bj as the position of each harmony to
Amount, as shown in formula (15).
X (HMS)=[wij,aj,bj,wj] (15)
In formula, HMS is harmony number, and i is input layer number, and j is hidden layer number of nodes, then harmony dimension D=(i+3)
×j。
Optimization Steps are as follows：
(1) parameter initialization.The parameter being configured is needed to have：Harmony data base HM capacity, data base probability
HMCR, tone finely tune probability P AR, and tone finely tunes broadband bw, maximum iteration Tmax.
(2) initialization harmony library.HMS harmony is generated, initial harmony library is constituted.Wavelet neural network is called to carry out
Prediction calculates prediction error, obtains each individual fitness value in original data base (HM).
(3) new harmony is generated.The random number rand1 of random generation 0~1, if meeting rand1<HMCR, then according to (16)
Formula randomly selects an individual from data base (HM).
The individual of selection is finely adjusted according to (17) formula；If being unsatisfactory for rand1<HMCR, then in the value model of variable
A new explanation is regenerated in enclosing.
In formula (17), rand2 is the random number between 0~1, and k is the number of discrete variable probable value.
(4) harmony data base is updated.The fitness value of the new explanation caused by step 3 is calculated, and is remembered according to the following formula
Recall the update of library (HM).
(5) judge whether algorithm terminates.Algorithm terminates if meeting maximum iteration, obtains optimal Wavelet Neural Network
Network structural parameters.Otherwise, algorithm goes to step 3 and continues to execute.
The prediction modeling of three hot(test)-spot temperature fluctuation ranges
On above-mentioned working foundation, construct based on Fuzzy Information Granulation and harmony chess game optimization wavelet neural network
(FIG-HS-WNN) coiling hot point of transformer temperature fluctuation range combination forecasting.The overall flow of model is as shown in Figure 2.
Modeling procedure is summarized as follows：
(1) historical data is updated, Fuzzy Information Granulation is carried out to sample data according to formula (2)-(5), obtains LOW, R, UP,
Fuzzy Information Granulation quality factor q is calculated according to formula (6)-(8), chooses optimal granulation result.
(2) historical data after screening granulation, the wavelet neural of a variety of different structures is tentatively established with training set sample
Network Prediction Model judges input layer number (number of days for inputting historical data) and hidden layer number of nodes by the model.
It should be noted that the structure for different history data set wavelet neural networks can change, i.e., all needed after updating the data daily
Wavelet neural network is readjusted.
(3) structural parameters that input layer, hidden layer and output layer determine that needs optimize are determined, it is right according to formula (15)
Optimization object is encoded, and formula (14) is defined as fitness value function, by harmony search to the structure of wavelet neural network
Parameter is screened.
(4) LOW, R, UP are predicted respectively using the wavelet neural network after design, obtains coiling hot point of transformer
Temperature fluctuation range.
(5) use three mean square error (MSE), mean absolute error (MAE), related coefficient (r) indexs to model prediction
Performance is assessed.
The foregoing is merely the preferred embodiments of the application, are not intended to limit this application, for the skill of this field
For art personnel, the application can have various modifications and variations.Within the spirit and principles of this application, any made by repair
Change, equivalent replacement, improvement etc., should be included within the protection domain of the application.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.
Claims (10)
1. a kind of coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation, it is characterized in that：Including:
The number that window when being granulated is determined according to Fuzzy Information Granulation quality, initial data is extracted with Fuzzy Information Granulation technology
Effective information；
Prediction model based on wavelet neural network is built based on effective information, and by harmonic search algorithm to the knot of wavelet neural network
Structure parameter optimizes；
Coiling hot point of transformer temperature fluctuation range is predicted with the model after parameter adjustment using structure.
2. a kind of coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation, it is characterized in that：Including:
Historical data is updated, Fuzzy Information Granulation is carried out to historical data sample data, obtains average value, lower bound and the upper bound, is counted
Fuzzy Information Granulation quality factor are calculated, optimal granulation result is chosen；
Historical data after screening granulation, the wavelet neural network that a variety of different structures are tentatively established with training set sample are predicted
Model judges input layer number and hidden layer number of nodes by the model, and timing updates；
The structural parameters that input layer, hidden layer and output layer determine that needs optimize are determined, optimization object is encoded,
Fitness value function is defined, the structural parameters of wavelet neural network are screened by harmony search；
Average value, lower bound and the upper bound are predicted respectively using the wavelet neural network after screening, obtain transformer winding heat
Point temperature fluctuation range；
Model prediction performance is assessed using three mean square error, mean absolute error and related coefficient indexs.
3. a kind of coiling hot point of transformer temperature fluctuation range based on Fuzzy Information Granulation as claimed in claim 1 or 2 is pre-
Survey method, it is characterized in that：During the effective information for extracting initial data with Fuzzy Information Granulation technology, specifically include：
Coiling hot point of transformer temperature-time sequence resolves into several small subsequences on demand, each subsequence is as one
Window is blurred when operation；
A triangular form obscure particle is established on each subsequence；
By time series by sorting from small to large, average value, lower bound and the upper bound of each triangular form obscure particle are sought.
4. a kind of coiling hot point of transformer temperature fluctuation range based on Fuzzy Information Granulation as claimed in claim 1 or 2 is pre-
Survey method, it is characterized in that：During extracting the effective information of initial data with Fuzzy Information Granulation technology, using fuzzy
The number of window when Information Granulating quality factor determine granulation, the quality factor are specially the data compression ratio of Fuzzy Information Granulation
With the log values of the amount lost ratio of data after Fuzzy Information Granulation.
5. a kind of coiling hot point of transformer temperature fluctuation range based on Fuzzy Information Granulation as claimed in claim 1 or 2 is pre-
Survey method, it is characterized in that：It is built in prediction model based on wavelet neural network based on effective information, using Morlet mother wavelet functions,
Using the coiling hot point of transformer temperature history after fuzzy granulation as the input of the input layer of wavelet neural network
Data.
6. a kind of coiling hot point of transformer temperature fluctuation range based on Fuzzy Information Granulation as claimed in claim 1 or 2 is pre-
Survey method, it is characterized in that：It is searched for using harmonic search algorithm in best wavelet neural network structural parameters, by Wavelet Neural Network
The prediction error of network model calls wavelet neural network to be predicted as object function, obtained prediction result be used as and
The fitness value of sound, it would be desirable to which input weight, output weight, contraction-expansion factor and the shift factor of the wavelet neural network of optimization are made
For the position vector of each harmony.
7. a kind of coiling hot point of transformer temperature fluctuation range prediction side based on Fuzzy Information Granulation as claimed in claim 6
Method, it is characterized in that：Optimization Steps include：
(1) to harmony memory storage capacity, data base probability, tone fine tuning probability, tone fine tuning broadband and greatest iteration time
Number is initialized；
(2) multiple harmony are generated, initial harmony library is constituted, wavelet neural network is called to be predicted, prediction error is calculated, obtains
To each individual fitness value in original data base；
(3) 0~1 random number is generated at random, if meeting the random number is less than data base probability, according to from data base
In randomly select an individual；
(4) individual of selection is finely adjusted；If being unsatisfactory for the random number is less than data base probability, in variable
A new explanation is regenerated in value range；
(5) harmony data base is updated, the fitness value of generated new explanation is calculated, and carries out the update of data base；
(6) judge whether algorithm terminates, algorithm terminates if meeting maximum iteration, obtains optimal wavelet neural network knot
Otherwise structure parameter is back to step (3) and continues to execute.
8. a kind of coiling hot point of transformer temperature fluctuation range prediction side based on Fuzzy Information Granulation as claimed in claim 7
Method, it is characterized in that：In the step (3), the process that an individual is randomly selected from data base includes：If random number is less than
Data base probability, new individual are random one in harmony, and otherwise, new individual is one in harmony vector.
9. a kind of coiling hot point of transformer temperature fluctuation range prediction side based on Fuzzy Information Granulation as claimed in claim 7
Method, it is characterized in that：In the step (4), if the random number is less than data base probability, new explanation is：
Rand2 is the random number between 0~1, and k is the number of discrete variable probable value, and PAR is that tone finely tunes probability.
10. a kind of coiling hot point of transformer temperature fluctuation range prediction based on Fuzzy Information Granulation as claimed in claim 7
Method, it is characterized in that：In the step (5), the fitness value of new explanation is calculated, the update of data base is carried out according to fitness value：
xworstWorst solution is concentrated for solution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481085.8A CN108733921B (en) | 2018-05-18 | 2018-05-18 | Transformer winding hot spot temperature fluctuation range prediction method based on fuzzy information granulation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481085.8A CN108733921B (en) | 2018-05-18 | 2018-05-18 | Transformer winding hot spot temperature fluctuation range prediction method based on fuzzy information granulation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108733921A true CN108733921A (en) | 2018-11-02 |
| CN108733921B CN108733921B (en) | 2021-07-16 |
Family
ID=63938458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810481085.8A Active CN108733921B (en) | 2018-05-18 | 2018-05-18 | Transformer winding hot spot temperature fluctuation range prediction method based on fuzzy information granulation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108733921B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109375504A (en) * | 2018-11-06 | 2019-02-22 | 江西北斗变电科技有限公司 | The method of winding insulation bubble formation control based on Fuzzy Neural PID control |
| CN109901030A (en) * | 2019-02-25 | 2019-06-18 | 国网山东省电力公司济南供电公司 | A kind of reactor turn-to-turn insulation state monitoring method, system and application |
| CN110244225A (en) * | 2019-06-27 | 2019-09-17 | 清华大学深圳研究生院 | A method of hot face temperature when prediction power battery electric discharge |
| CN116432406A (en) * | 2023-03-09 | 2023-07-14 | 广东电网有限责任公司佛山供电局 | Method and device for calculating hot spot temperature of working winding of oil immersed transformer |
| CN118091522A (en) * | 2024-04-24 | 2024-05-28 | 呼和浩特市奥祥电力自动化有限公司 | Testing and checking system and method for transformer substation |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6609079B1 (en) * | 1998-05-14 | 2003-08-19 | Va Tech Elin Transformatoren Gmbh | Method and arrangement for ascertaining state variables |
| CN102915355A (en) * | 2012-10-11 | 2013-02-06 | 李英明 | Multiprocessor task scheduling method based on harmony search and simulated annealing |
| CN103150610A (en) * | 2013-02-28 | 2013-06-12 | 哈尔滨工业大学 | Fuzzy information granulation and support vector machine-based heating load prediction method |
| CN105550472A (en) * | 2016-01-20 | 2016-05-04 | 国网上海市电力公司 | Prediction method of transformer winding hot-spot temperature based on neural network |
| CN105590023A (en) * | 2015-12-08 | 2016-05-18 | 三峡大学 | Fuzzy granulation prediction method of performance degradation of rolling bearing on the basis of information entropy |
| US20160293832A1 (en) * | 2015-03-31 | 2016-10-06 | Ngk Insulators, Ltd. | Piezoelectric/electrostrictive material, piezoelectric/electrostrictive body, and resonance driving device |
| CN106073702A (en) * | 2016-05-27 | 2016-11-09 | 燕山大学 | Many time-frequencies yardstick diencephalon myoelectricity coupling analytical method based on small echo transfer entropy |
| CN106527141A (en) * | 2016-12-05 | 2017-03-22 | 清华大学 | Glass furnace air-fuel ratio adjustment method based on variable universe fuzzy rule iterative learning |
| CN107368926A (en) * | 2017-07-28 | 2017-11-21 | 中南大学 | A kind of how natural parameter sensing method for amalgamation processing of intelligent environment carrying robot identification floor |
| CN107506871A (en) * | 2017-09-08 | 2017-12-22 | 广东工业大学 | A kind of method and system of interval prediction |
| CN107784655A (en) * | 2016-12-28 | 2018-03-09 | 中国测绘科学研究院 | A kind of visual attention model SAR naval vessels detection algorithm of adaptive threshold |
-
2018
- 2018-05-18 CN CN201810481085.8A patent/CN108733921B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6609079B1 (en) * | 1998-05-14 | 2003-08-19 | Va Tech Elin Transformatoren Gmbh | Method and arrangement for ascertaining state variables |
| CN102915355A (en) * | 2012-10-11 | 2013-02-06 | 李英明 | Multiprocessor task scheduling method based on harmony search and simulated annealing |
| CN103150610A (en) * | 2013-02-28 | 2013-06-12 | 哈尔滨工业大学 | Fuzzy information granulation and support vector machine-based heating load prediction method |
| US20160293832A1 (en) * | 2015-03-31 | 2016-10-06 | Ngk Insulators, Ltd. | Piezoelectric/electrostrictive material, piezoelectric/electrostrictive body, and resonance driving device |
| CN105590023A (en) * | 2015-12-08 | 2016-05-18 | 三峡大学 | Fuzzy granulation prediction method of performance degradation of rolling bearing on the basis of information entropy |
| CN105550472A (en) * | 2016-01-20 | 2016-05-04 | 国网上海市电力公司 | Prediction method of transformer winding hot-spot temperature based on neural network |
| CN106073702A (en) * | 2016-05-27 | 2016-11-09 | 燕山大学 | Many time-frequencies yardstick diencephalon myoelectricity coupling analytical method based on small echo transfer entropy |
| CN106527141A (en) * | 2016-12-05 | 2017-03-22 | 清华大学 | Glass furnace air-fuel ratio adjustment method based on variable universe fuzzy rule iterative learning |
| CN107784655A (en) * | 2016-12-28 | 2018-03-09 | 中国测绘科学研究院 | A kind of visual attention model SAR naval vessels detection algorithm of adaptive threshold |
| CN107368926A (en) * | 2017-07-28 | 2017-11-21 | 中南大学 | A kind of how natural parameter sensing method for amalgamation processing of intelligent environment carrying robot identification floor |
| CN107506871A (en) * | 2017-09-08 | 2017-12-22 | 广东工业大学 | A kind of method and system of interval prediction |
Non-Patent Citations (4)
| Title |
|---|
| LI ZHANG 等: "A New Prediction Model for Transformer Winding Hotspot Temperature Fluctuation Based on Fuzzy Information Granulation and an Optimized Wavelet Neural Network", 《ENERGIES》 * |
| WESLEY N. ALLEN 等: "Modeling inductance and quality factor of integrated spiral inductors on low-loss substrates up to 5 GHz", 《2017 IEEE INTERNATIONAL CONFERENCE ON MICROWAVES, ANTENNAS, COMMUNICATIONS AND ELECTRONIC SYSTEMS (COMCAS)》 * |
| 李硕: "基于信息熵—灰色关联度法的电力变压器故障诊断研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 王冠 等: "基于改进变分模态分解的有载分接开关机械状态监测", 《湖南大学学报(自然科学版)》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109375504A (en) * | 2018-11-06 | 2019-02-22 | 江西北斗变电科技有限公司 | The method of winding insulation bubble formation control based on Fuzzy Neural PID control |
| CN109901030A (en) * | 2019-02-25 | 2019-06-18 | 国网山东省电力公司济南供电公司 | A kind of reactor turn-to-turn insulation state monitoring method, system and application |
| CN109901030B (en) * | 2019-02-25 | 2021-10-22 | 国网山东省电力公司济南供电公司 | Reactor turn-to-turn insulation state monitoring method, system and application |
| CN110244225A (en) * | 2019-06-27 | 2019-09-17 | 清华大学深圳研究生院 | A method of hot face temperature when prediction power battery electric discharge |
| CN116432406A (en) * | 2023-03-09 | 2023-07-14 | 广东电网有限责任公司佛山供电局 | Method and device for calculating hot spot temperature of working winding of oil immersed transformer |
| CN116432406B (en) * | 2023-03-09 | 2024-02-02 | 广东电网有限责任公司佛山供电局 | Method and device for calculating hot spot temperature of working winding of oil immersed transformer |
| CN118091522A (en) * | 2024-04-24 | 2024-05-28 | 呼和浩特市奥祥电力自动化有限公司 | Testing and checking system and method for transformer substation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108733921B (en) | 2021-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108733921A (en) | Coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation | |
| CN109948029B (en) | Neural network self-adaptive depth Hash image searching method | |
| Lan et al. | Constructive hidden nodes selection of extreme learning machine for regression | |
| CN107665230A (en) | Training method and device for the users' behavior model of Intelligent housing | |
| CN108573303A (en) | It is a kind of that recovery policy is improved based on the complex network local failure for improving intensified learning certainly | |
| CN108932671A (en) | A kind of LSTM wind-powered electricity generation load forecasting method joined using depth Q neural network tune | |
| CN103905246B (en) | Link prediction method based on grouping genetic algorithm | |
| CN108876044B (en) | Online content popularity prediction method based on knowledge-enhanced neural network | |
| CN113784410B (en) | Heterogeneous wireless network vertical switching method based on reinforcement learning TD3 algorithm | |
| CN110851566A (en) | Improved differentiable network structure searching method | |
| CN109413710B (en) | Clustering method and device of wireless sensor network based on genetic algorithm optimization | |
| Yang et al. | Deep reinforcement learning based wireless network optimization: A comparative study | |
| CN115271099A (en) | Self-adaptive personalized federal learning method supporting heterogeneous model | |
| CN109214546A (en) | A kind of Power Short-Term Load Forecasting method based on improved HS-NARX neural network | |
| CN107944617A (en) | A kind of doubtful stealing theme influence factor weight optimization method that logic-based returns | |
| CN112667406A (en) | Task unloading and data caching method in cloud edge fusion heterogeneous network | |
| CN106296434A (en) | A kind of Grain Crop Yield Prediction method based on PSO LSSVM algorithm | |
| TWI452529B (en) | Combined with the system equivalent model of the system and its computer program products | |
| Chen et al. | Outdoor localization for LoRaWans using semi-supervised transfer learning with grid segmentation | |
| CN117520956A (en) | Two-stage automatic feature engineering method based on reinforcement learning and meta learning | |
| Yang et al. | DHQN: a stable approach to remove target network from deep q-learning network | |
| Yu et al. | Proximal Policy Optimization-based Federated Client Selection for Internet of Vehicles | |
| CN113157344B (en) | DRL-based energy consumption perception task unloading method in mobile edge computing environment | |
| CN111274498B (en) | Network characteristic community searching method | |
| Lu et al. | Gaussian process dynamical modeling for adaptive inference over graphs |
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 |