CN109447309A - A kind of method for digging that waving data and system - Google Patents
A kind of method for digging that waving data and system Download PDFInfo
- Publication number
- CN109447309A CN109447309A CN201811029046.0A CN201811029046A CN109447309A CN 109447309 A CN109447309 A CN 109447309A CN 201811029046 A CN201811029046 A CN 201811029046A CN 109447309 A CN109447309 A CN 109447309A
- Authority
- CN
- China
- Prior art keywords
- impact factor
- factor
- transmission line
- class
- impact
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000005540 biological transmission Effects 0.000 claims abstract description 63
- 230000005611 electricity Effects 0.000 claims abstract description 33
- 238000012216 screening Methods 0.000 claims abstract description 18
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims description 30
- 238000012706 support-vector machine Methods 0.000 claims description 18
- 238000001556 precipitation Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 235000013399 edible fruits Nutrition 0.000 claims description 7
- 230000001755 vocal effect Effects 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 4
- 239000012141 concentrate Substances 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 239000004020 conductor Substances 0.000 description 3
- 238000010219 correlation analysis Methods 0.000 description 3
- 238000007418 data mining Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 208000035126 Facies Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of method for digging that waving data and system, comprising: the potential impact factor pair based on transmission line galloping, which waves information record sheet and carries out screening, obtains first kind impact factor；The supporting vector machine model constructed based on the corresponding related coefficient of the first kind impact factor and in advance, each impact factor in the first kind impact factor is ranked up, selects the impact factor for meeting preset condition as the second class impact factor from the first kind impact factor after sequence；The corresponding transmission line of electricity related data of the second class impact factor described in information record sheet will be waved as wave analysis and wave data.The present invention can extract and wave the closely related impact factor of incidence, improve the accuracy rate for waving prediction and warning.
Description
Technical field
It prevents and reduces natural disasters field the present invention relates to transmission line of electricity, and in particular to a kind of method for digging for waving data and system.
Background technique
Transmission line galloping accident is to threaten one of the important disaster form of power network safety operation, transmission line galloping
It easily leads to line tripping, wire strand breakage, metal bolt to loosen, when accident is waved on a wide range of route of power transmission road of generation, will lead to
Electric grid large area power cut seriously affects bulk power grid safe and stable operation.In recent years, with the fast development of artificial intelligence technology,
Its power grid prevent and reduce natural disasters field application increasingly deeply, people by artificial intelligence technology introduce transmission line galloping prediction and warning
In work, related data largely is waved using accumulation, prediction and warning is carried out to the possibility occurrence for the accident of waving, so that
Route O&M department can prejudge the generation of disaster in advance, formulate counter-measure, and the damage that accident generates power grid is waved in reduction.
However, the accuracy rate for waving prediction is also not fully up to expectations, reason is that waving the utilization of data and excavation level need to be improved,
It waves that related data is various informative, how to be extracted from the data of magnanimity complexity and study most valuable information for waving,
It is to improve basis and the key of waving intelligent early-warning precision and accuracy rate, therefore, it is necessary to deep to data mining technology progress is waved
Enter research.
Summary of the invention
In order to solve the problems, such as waving in the presence of the prior art, predictablity rate is low, and the present invention provides one kind and waves number
According to method for digging and system.
Present invention provide the technical scheme that a kind of method for digging for waving data, comprising:
Potential impact factor pair based on transmission line galloping waves information record sheet and carries out screening acquisition first kind influence
The factor；
The supporting vector machine model constructed based on the corresponding related coefficient of the first kind impact factor and in advance, to described
Each impact factor in first kind impact factor is ranked up, and selection meets default item from the first kind impact factor after sequence
The impact factor of part is as the second class impact factor；
The corresponding transmission line of electricity related data of the second class impact factor described in information record sheet will be waved as waving point
Data are waved in analysis.
Preferably, the support vector machines constructed based on the corresponding related coefficient of the first kind impact factor and in advance
Model is ranked up each impact factor in the first kind impact factor, comprising:
From the power transmission line waved and the first impact factor and first impact factor is selected to be related in information record sheet
Road related data generates analytical data；
Based on the corresponding mean value of impact factor each in the analytical data calculating first kind impact factor and mean square deviation；
The related coefficient of each impact factor is calculated based on the corresponding mean value of each impact factor and mean square deviation；
The absolute value of related coefficient based on each impact factor to each impact factor in the first kind impact factor into
Row sequence obtains the first ranking results；
Based on analytical data and supporting vector machine model, each impact factor in the first kind impact factor is carried out
Sequence obtains the second ranking results.
Preferably, select the impact factor for meeting preset condition as in the first kind impact factor from after sequence
Two class impact factors, comprising:
Compare in the first ranking results in each impact factor ranking results and the second ranking results of first kind impact factor
Each impact factor ranking results of first kind impact factor；
Multiple impact factors when being sequentially threshold value in first ranking results and the second ranking results are identical, then select
Identical impact factor is selected as the second class impact factor.
Preferably, described to be based on analytical data and supporting vector machine model, to each in the first kind impact factor
Impact factor is ranked up to obtain the second sequence as a result, including:
Primitive character set is set by whole impact factors in first kind impact factor, while being arranged one is empty set
Feature ordering collection；
Using first kind impact factor and the corresponding transmission line of electricity related data of the first kind impact factor as training sample
This；
The first formula Training Support Vector Machines classifier based on the training sample and setting obtains the weight of hyperplane
Vector；
Second formula of weight vector and setting based on the hyperplane calculates ranking criteria score；
The corresponding impact factor of the smallest ranking criteria score is deleted from the primitive character set, and institute will be based on
It states the result feature that the smallest ranking criteria score obtains and is added to the feature ordering concentration；
The corresponding impact factor of the smallest ranking criteria score and transmission line of electricity dependency number are deleted from current training sample
According to obtaining updated training sample；
The corresponding impact factor of the smallest ranking criteria score, Zhi Daosuo are continued to search based on training sample after the update
When stating primitive character collection and being combined into empty set, current signature sequence is collected into corresponding impact factor as the second ranking results.
Preferably, the weight vector of the hyperplane, is calculated as follows:
In formula: ω: the weight vector of hyperplane；αi: the Lagrange multiplier of i-th of impact factor；xi: i-th influence because
The value of son；yi: the class label of i-th of impact factor；N: training sample number.
Preferably, the ranking criteria score, is calculated as follows:
In formula: ck: the ranking criteria score of k-th of impact factor；The weight of the hyperplane of k-th of impact factor to
Amount；The value of k is the number of impact factor in primitive character set.
Preferably, the result feature, is calculated as follows:
P=arg min ck
In formula: p: the result feature being made of the value of k；ck: the ranking criteria score of k-th of impact factor.
Preferably, the related coefficient of each impact factor, is calculated as follows:
In formula: the related coefficient of R (k): k-th impact factor；μk(+): k-th of impact factor corresponds to the equal of (+1) class
Value；μk(-): the mean value of corresponding (- 1) class of k-th of impact factor；σk(+): the mean square deviation of corresponding (+1) class of k-th of impact factor:
σk(-): the mean square deviation of corresponding (- 1) class of k-th of impact factor.
Preferably, the potential impact factor pair based on transmission line galloping waves information record sheet and carries out screening acquisition
First kind impact factor, comprising:
It is related that the potential impact factor based on transmission line galloping obtains the transmission line of electricity that the potential impact factor is related to
Data；
Information is waved in the transmission line of electricity related data building being related to based on the potential impact factor and the potential impact factor
Record sheet；
Impact factor composition first kind impact factor to be analyzed is filtered out from the information record sheet of waving；
Wherein, the potential impact factor of the transmission line galloping, comprising: wind speed, wind direction, line alignment and wind direction press from both sides
Angle, temperature, humidity, precipitation form, wire icing thickness, landform, height above sea level, voltage class, shaft tower feeder number, twisted wire section
Product, arrangement of conductor and conducting wire division number.
Preferably, the potential impact factor pair based on transmission line galloping waves information record sheet and carries out screening acquisition
First kind impact factor, further includes:
Numerical value shape will be converted by the conversion principle of setting by the impact factor of verbal description in first kind impact factor
Formula.
Based on the same inventive concept, the present invention also provides a kind of digging systems for waving data, comprising:
Screening module, for the potential impact factor pair based on transmission line galloping wave information record sheet carry out screening obtain
Obtain first kind impact factor；
Sorting module, the supporting vector for constructing based on the corresponding related coefficient of the first kind impact factor and in advance
Machine model is ranked up each impact factor in the first kind impact factor, from the first kind impact factor after sequence
Select the impact factor for meeting preset condition as the second class impact factor；
Analysis module, for the corresponding transmission line of electricity dependency number of the second class impact factor described in information record sheet will to be waved
Data are waved according to as wave analysis.
Preferably, the sorting module, comprising:
Generate submodule, for from it is described wave selected in information record sheet the first impact factor and it is described first influence because
The transmission line of electricity related data that son is related to generates analytical data；
First computational submodule, for calculating each impact factor pair in the first kind impact factor based on analytical data
The mean value and mean square deviation answered；
Second computational submodule, for calculating the phase of each impact factor based on the corresponding mean value of each impact factor and mean square deviation
Relationship number；
First sequence is born fruit module, and the absolute value for the related coefficient based on each impact factor is to the first kind shadow
Each impact factor rung in the factor is ranked up to obtain the first ranking results；
Second sequence is born fruit module, for being based on analytical data and supporting vector machine model, to the first kind shadow
Each impact factor rung in the factor is ranked up to obtain the second ranking results.
Compared with prior art, the invention has the benefit that
Technical solution provided by the invention, the potential impact factor pair based on transmission line galloping wave information record sheet into
Row screening obtains first kind impact factor；The support constructed based on the corresponding related coefficient of the first kind impact factor and in advance
Vector machine model is ranked up each impact factor in the first kind impact factor, from after sequence the first kind influence because
Select the impact factor for meeting preset condition as the second class impact factor in son；The second class described in information record sheet will be waved
The corresponding transmission line of electricity related data of impact factor waves data as wave analysis, can extract and to wave incidence close
Relevant impact factor is cut, the accuracy rate for waving prediction and warning is improved.
Technical solution provided by the invention can be adapted under all different zones and DIFFERENT METEOROLOGICAL CONDITIONS, analysis transmission of electricity
The impact factor of line oscillation.
Technical solution provided by the invention, can effectively excavate the influence highly relevant with transmission line galloping because
Son makes major contribution for transmission line galloping prediction modeling, has striven for preciousness to dispose transmission line of electricity anti-dance measure in advance
Time.
Detailed description of the invention
Fig. 1 is a kind of method for digging flow chart for waving data provided by the invention；
Fig. 2 waves information record sheet schematic diagram to be provided in an embodiment of the present invention；
Fig. 3 is line alignment coordinate transition diagram in the embodiment of the present invention；
Fig. 4 is the column schematic diagram of the first ranking results in the embodiment of the present invention.
Specific embodiment
For a better understanding of the present invention, the contents of the present invention are done further with example with reference to the accompanying drawings of the specification
Explanation.
Embodiment 1:
S1, the potential impact factor pair based on transmission line galloping wave information record sheet and carry out screening acquisition first kind shadow
Ring the factor；
S2, the supporting vector machine model constructed based on the corresponding related coefficient of the first kind impact factor and in advance are right
Each impact factor in the first kind impact factor is ranked up, and selection meets pre- from the first kind impact factor after sequence
If the impact factor of condition is as the second class impact factor；
S3, the corresponding transmission line of electricity related data of the second class impact factor described in information record sheet will be waved as waving
Data are waved in analysis.
For the text data used in the present embodiment is " waving information record sheet ", the excavation side that provides through the invention
Method obtains the data of waving that transmission line of electricity wave analysis, and part of data are as shown in Figure 2.
S1, the potential impact factor pair based on transmission line galloping wave information record sheet and carry out screening acquisition first kind shadow
The factor is rung, is specifically included:
1.1 wave data description
1.1.1 initial data and its formatting
Transmission line galloping related data is numerous and jumbled, and form is polynary, including the multiple types such as image, text, video, the present invention
It is studied just for text data.
The potential impact factor based on transmission line galloping obtains the transmission line of electricity related data that the potential impact factor is related to；
Information record is waved in the transmission line of electricity related data building being related to based on the potential impact factor and the potential impact factor
Table.
It marks the record for being to be denoted as positive sample in Fig. 2, is denoted as negative sample labeled as the non-record waved, it is clear that positive and negative
Sample distribution is unbalanced, and ratio is about 1:7.
Due to the present invention is directed to analyze the major influence factors of transmission line galloping, it only chooses and waves genetic facies
The data of pass are analyzed.The potential impact factor of transmission line galloping can be classified as three categories, first is that meteorologic factor, second is that ground
Reason factor, third is that the structure feature of transmission line of electricity itself.
1) meteorologic factor: wind speed, wind direction, line alignment and wind direction angle, temperature, humidity, precipitation form, wire icing are thick
Degree.
2) geographic factor: landform, height above sea level.
3) structure feature of transmission line of electricity itself: voltage class, shaft tower feeder number, twisted wire sectional area, arrangement of conductor,
Conducting wire division number.
According to waving genesis mechanism and wave the theoretical analysis result of rule, the following 12 impact factors composition of screening the
A kind of impact factor is analyzed: voltage class, height above sea level, line alignment, wind speed, wind direction, wind direction and route angle, temperature
Degree, humidity, precipitation form, wire icing thickness, outer twisted wire area and conducting wire division number.
From waving the transmission line of electricity dependency number for selecting the first impact factor and the first impact factor to be related in information record sheet
According to generation analytical data.
Numerical value shape will be converted by the conversion principle of setting by the impact factor of verbal description in first kind impact factor
Formula.
As shown in Figure 2, line alignment, wind direction, 3 impact factors of precipitation form value be verbal description, remaining 9 influence
The factor is numeric type information, in order to be identified by Data Mining Tools, needs to be converted to verbal description numeric type letter
Breath.Wherein line alignment and wind direction are converted according to coordinate shown in Fig. 3.Specific format conversion principle such as table 1.1 is to table
1.3 shown.
1.1 line alignment format conversion principle of table
The conversion of 1.2 wind direction format of table
| Wind direction | Corresponding numerical value |
| Northern (by north) | 90 |
| Northeast | 45 |
| Northwest | 135 |
The conversion of 1.3 precipitation formal mode of table
| Precipitation form | Corresponding numerical value |
| Rime | 1 |
| Glaze | 2 |
| Snow slush | 3 |
S2, the supporting vector machine model constructed based on the corresponding related coefficient of the first kind impact factor and in advance are right
Each impact factor in the first kind impact factor is ranked up, and selection meets pre- from the first kind impact factor after sequence
If the impact factor of condition is specifically included as the second class impact factor:
1.2 degrees of correlation waving impact factor and whether waving
In order to excavate and highly relevant main impact factor occurs for the accident of waving, need to analyze each impact factor
Sample and the non-importance for waving sample are waved for differentiation.Correlation analysis and recursive feature is respectively adopted in the present embodiment
Null method come analyze each impact factor for wave occur whether importance.
1.2.1 correlation analysis
Correlation analysis is to be analyzed using Principle of Statistics data, using related coefficient come between gauge variable
Degree of correlation, in the present embodiment i.e. calculate the corresponding related coefficient of each impact factor, according to related coefficient absolute value into
Row sequence, to judge each impact factor to the correlation size for waving accident rate.
Whether wave and belong to two class problems, is i.e. two-value (binary) dispersive target variable, corresponds to two values+1 or -1,
The calculation formula of its related coefficient is as follows:
The wherein related coefficient of R (k): k-th impact factor, R>0 indicate related to (+1) class, and R<0 is indicated and (- 1) class
The absolute value of correlation, R is bigger, and correlation is stronger；μk(+): the mean value of corresponding (+1) class of k-th of impact factor；μk(-): k-th
The mean value of corresponding (- 1) class of impact factor；σk(+): the mean square deviation of corresponding (+1) class of k-th of impact factor: σk(-): k-th of influence
Factor pair answers the mean square deviation of (- 1) class.
Calculated result is as shown in figure 4, from the point of view of ranking results, for current data record, voltage class, outer twisted wire face
Importance is higher whether product, conducting wire division number, line alignment, wind speed etc. wave generation for differentiation, and wind direction, wind direction and route
The importance that angle etc. waves identification is taken second place.
First ranking results are as follows: voltage class, outer twisted wire area, conducting wire division number, line alignment, wind speed, precipitation form,
Height above sea level, wire icing thickness, temperature, wind direction and route angle and wind direction.
1.2.2 recursive feature null method
The main thought that recursive feature eliminates (Recursive Feature Elimination, RFE) is using a machine
Device learning model (such as support vector machines or regression model) repetition training is multiple, every time after training, eliminates worst spy
It levies (can be selected according to coefficient), this process is then repeated in remaining feature, until all features all traverse.This
The order that feature is eliminated in the process is exactly the sequence of feature.The feature eliminated at first is least important, and the feature finally eliminated is most
To be important, this method is a kind of greedy algorithm for finding optimal feature subset.
It is special using the recurrence based on support vector machines (Support Vector Machine, SVM) model in the present embodiment
It levies null method (SVM-RFE), support vector machines (SVM) principle summary, in detail reference can be made to document " data mining introduction " (people's postal
Electric publishing house) 5.2.2 section content, comprising:
Known training sample setWherein, xi∈RD, yi∈ {+1, -1 } is xiClass label, N be training
Number of samples, D are the intrinsic dimensionality of sample.
SVM seeks optimal classification surface ω x+b=0, wherein ω is the weight vector of hyperplane, and b is threshold value, so that two
Class interval between class sample maximizes, to obtain weight vector and threshold value, optimization problem below demand solution:
Wherein, C > 0 is punishment parameter, ζiIt is slack variable.
Parameter C plays the role of control and divides sample to mistake carrying out punishment degree, realize ratio in error sample with
The optimization problem of SVM can be changed into antithesis below by introducing Lagrange multiplier by the compromise between algorithm complexity
Planning problem:
Wherein, αiIt is Lagrange multiplier.
Relationship between weight vector and the solution of antithesis optimization (formula 11) are as follows:
The discriminant function of SVM are as follows:
Wherein, sgn () is sign function.
SVM-RFE feature selecting algorithm is that all features of needs are initialized as a characteristic set, then every time repeatedly
In generation, rejects a smallest feature of ranking criteria score, and until obtaining final feature set, therefore, SVM-RFE is one and is based on
Sequence backward selection (Sequential Backward Selection, SBS) algorithm of the largest interval principle of SVM.
In SVM-RFE, ith feature ranking criteria score definition are as follows:
Remove the smallest feature of ranking criteria score in each iteration, is then training SVM with remaining feature, carrying out
Next iteration.
The second sequence is obtained as a result, including: using SVM-RFE feature selecting algorithm in the present embodiment
Input: training sample
Output: feature ordering collection R；
1) it initializes, primitive character set S={ 1,2 ..., D }, feature ordering collection
2) circulation following procedure until
1. obtaining the training sample with candidate feature set, the training sample in the present embodiment is analytical data；
2. obtaining ω with formula (4) training SVM classifier；
3. calculating ranking criteria score with formula (6)
4. finding out the smallest feature of ranking criteria scoreWork as ckWhen being minimized, the value of k；
5. updating feature set R={ p } ∪ R；
This feature S=S/p is removed in S.
Finally obtain the impact factor that k in feature ordering set R, R corresponds to primitive character set.
Impact factor is ranked up using REF method, the results are shown in Table 4 for the second sequence:
4 RFE method impact factor ranking results of table
| Sequence | Impact factor |
| 1 | Wind speed |
| 2 | Voltage class |
| 3 | Line alignment |
| 4 | Outer twisted wire area |
| 5 | Conducting wire division number |
| 6 | Humidity |
| 7 | Height above sea level |
| 8 | Wind direction and route angle |
| 9 | Wire icing thickness |
| 10 | Precipitation form |
| 11 | Wind direction |
| 12 | Temperature |
Ranking results show whether wind speed, voltage class, outer twisted wire area, line alignment and conducting wire division number are for sending out
The raw influence waved is more important, consistent with the result that correlation coefficient process obtains.
From the first ranking results and the second ranking results select wind speed, voltage class, outer twisted wire area, line alignment and
Conducting wire division number forms the second class impact factor；
Data are waved using the corresponding transmission line of electricity related data of the second class impact factor as wave analysis.
Embodiment 2:
Based on same inventive concept, the present invention also provides a kind of digging systems for waving data, comprising:
Screening module, for the potential impact factor pair based on transmission line galloping wave information record sheet carry out screening obtain
Obtain first kind impact factor；
Sorting module, the supporting vector for constructing based on the corresponding related coefficient of the first kind impact factor and in advance
Machine model is ranked up each impact factor in the first kind impact factor, from the first kind impact factor after sequence
Select the impact factor for meeting preset condition as the second class impact factor；
Analysis module, for the corresponding transmission line of electricity dependency number of the second class impact factor described in information record sheet will to be waved
Data are waved according to as wave analysis.
In embodiment, the sorting module, comprising:
Generate submodule, for from it is described wave selected in information record sheet the first impact factor and it is described first influence because
The transmission line of electricity related data that son is related to generates analytical data；
First computational submodule, for calculating each impact factor pair in the first kind impact factor based on analytical data
The mean value and mean square deviation answered；
Second computational submodule, for calculating the phase of each impact factor based on the corresponding mean value of each impact factor and mean square deviation
Relationship number；
First sequence is born fruit module, and the absolute value for the related coefficient based on each impact factor is to the first kind shadow
Each impact factor rung in the factor is ranked up to obtain the first ranking results；
Second sequence is born fruit module, for being based on analytical data and supporting vector machine model, to the first kind shadow
Each impact factor rung in the factor is ranked up to obtain the second ranking results.
In embodiment, the sorting module, further includes:
Comparative sub-module, for comparing each impact factor ranking results of first kind impact factor in the first ranking results,
With each impact factor ranking results of first kind impact factor in the second ranking results；
Submodule is recombinated, for multiple shadows when being sequentially threshold value in first ranking results and the second ranking results
It is identical to ring the factor, then selects identical impact factor as the second class impact factor.
In embodiment, second sequence is born fruit module, comprising:
Initial cell, for setting primitive character set for whole impact factors in first kind impact factor, simultaneously
The feature ordering collection that one is empty set is set；
Training sample unit is used for first kind impact factor and the corresponding transmission line of electricity phase of the first kind impact factor
Data are closed as training sample；
Weight vector unit is calculated, for the first formula Training Support Vector Machines point based on the training sample and setting
Class device obtains the weight vector of hyperplane；
Ranking criteria score unit is calculated, the second formula for weight vector and setting based on the hyperplane calculates
Ranking criteria score；
Processing unit, for deleting the corresponding impact factor of the smallest ranking criteria score from the primitive character set
It removes, and the result feature obtained based on the smallest ranking criteria score is added to the feature ordering and is concentrated；
Updating unit, for deleting the corresponding impact factor of the smallest ranking criteria score and defeated from current training sample
Electric line related data obtains updated training sample；
Cycling element, for continuing to search the corresponding shadow of the smallest ranking criteria score based on training sample after the update
The factor is rung, when the primitive character collection is combined into empty set, current signature sequence is collected into corresponding impact factor as second row
Sequence result.
In embodiment, the screening module, comprising:
Acquiring unit obtains what the potential impact factor was related to for the potential impact factor based on transmission line galloping
Transmission line of electricity related data；
Construction unit, the transmission line of electricity related data for being related to based on the potential impact factor and the potential impact factor
Information record sheet is waved in building；
Recomposition unit, for filtering out impact factor composition first kind influence to be analyzed from the information record sheet of waving
The factor；
Wherein, the potential impact factor of the transmission line galloping, comprising: wind speed, wind direction, line alignment and wind direction press from both sides
Angle, temperature, humidity, precipitation form, wire icing thickness, landform, height above sea level, voltage class, shaft tower feeder number, twisted wire section
Product, arrangement of conductor and conducting wire division number.
Obviously, described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on the present invention
In embodiment, all other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
The above is only the embodiment of the present invention, are not intended to restrict the invention, all in the spirit and principles in the present invention
Within, any modification, equivalent substitution, improvement and etc. done, be all contained in apply pending scope of the presently claimed invention it
It is interior.
Claims (12)
1. a kind of method for digging for waving data characterized by comprising
Potential impact factor pair based on transmission line galloping waves information record sheet and carries out screening acquisition first kind impact factor；
The supporting vector machine model constructed based on the corresponding related coefficient of the first kind impact factor and in advance, to described first
Each impact factor in class impact factor is ranked up, and selection meets preset condition from the first kind impact factor after sequence
Impact factor is as the second class impact factor；
The corresponding transmission line of electricity related data of the second class impact factor described in information record sheet will be waved as waving analysis
Wave data.
2. method for digging as described in claim 1, which is characterized in that described to be based on the corresponding phase of the first kind impact factor
Relationship number and the supporting vector machine model constructed in advance are ranked up each impact factor in the first kind impact factor,
Include:
From the transmission line of electricity phase waved and the first impact factor and first impact factor is selected to be related in information record sheet
Data are closed, analytical data is generated；
Based on the corresponding mean value of impact factor each in the analytical data calculating first kind impact factor and mean square deviation；
The related coefficient of each impact factor is calculated based on the corresponding mean value of each impact factor and mean square deviation；
The absolute value of related coefficient based on each impact factor arranges each impact factor in the first kind impact factor
Sequence obtains the first ranking results；
Based on analytical data and supporting vector machine model, each impact factor in the first kind impact factor is ranked up
Obtain the second ranking results.
3. method for digging as claimed in claim 2, which is characterized in that selected in the first kind impact factor from after sequence
Meet the impact factor of preset condition as the second class impact factor, comprising:
Compare in the first ranking results first in each impact factor ranking results and the second ranking results of first kind impact factor
Each impact factor ranking results of class impact factor；
Multiple impact factors when being sequentially threshold value in first ranking results and the second ranking results are identical, then select phase
Same impact factor is as the second class impact factor.
4. method for digging as claimed in claim 3, which is characterized in that described to be based on analytical data and support vector machines mould
Type is ranked up to obtain the second sequence as a result, including: to each impact factor in the first kind impact factor
Primitive character set is set by whole impact factors in first kind impact factor, while the spy that one is empty set is set
Sign sequence collection；
Using first kind impact factor and the corresponding transmission line of electricity related data of the first kind impact factor as training sample；
The first formula Training Support Vector Machines classifier based on the training sample and setting obtains the weight vector of hyperplane；
Second formula of weight vector and setting based on the hyperplane calculates ranking criteria score；
The corresponding impact factor of the smallest ranking criteria score is deleted from the primitive character set, and will be based on described in most
The result feature that small ranking criteria score obtains is added to the feature ordering and concentrates；
The corresponding impact factor of the smallest ranking criteria score and transmission line of electricity related data are deleted from current training sample, are obtained
To updated training sample；
The corresponding impact factor of the smallest ranking criteria score is continued to search based on training sample after the update, until the original
When beginning characteristic set is empty set, current signature sequence is collected into corresponding impact factor as the second ranking results.
5. method for digging as claimed in claim 4, which is characterized in that the weight vector of the hyperplane is calculated as follows:
In formula: ω: the weight vector of hyperplane；αi: the Lagrange multiplier of i-th of impact factor；xi: i-th impact factor
Value；yi: the class label of i-th of impact factor；N: training sample number.
6. method for digging as claimed in claim 4, which is characterized in that the ranking criteria score is calculated as follows:
In formula: ck: the ranking criteria score of k-th of impact factor；The weight vector of the hyperplane of k-th of impact factor；k
Value be primitive character set in impact factor number.
7. method for digging as claimed in claim 4, which is characterized in that the result feature is calculated as follows:
P=argminck
In formula: p: the result feature being made of the value of k；ck: the ranking criteria score of k-th of impact factor.
8. method for digging as claimed in claim 3, which is characterized in that the related coefficient of each impact factor is counted as the following formula
It calculates:
In formula: the related coefficient of R (k): k-th impact factor；μk(+): the mean value of corresponding (+1) class of k-th of impact factor；μk
(-): the mean value of corresponding (- 1) class of k-th of impact factor；σk(+): the mean square deviation of corresponding (+1) class of k-th of impact factor: σk(-):
The mean square deviation of corresponding (- 1) class of k-th of impact factor.
9. method for digging as described in claim 1, which is characterized in that the potential impact factor based on transmission line galloping
Screening acquisition first kind impact factor is carried out to information record sheet is waved, comprising:
The potential impact factor based on transmission line galloping obtains the transmission line of electricity related data that the potential impact factor is related to；
Information record is waved in the transmission line of electricity related data building being related to based on the potential impact factor and the potential impact factor
Table；
Impact factor composition first kind impact factor to be analyzed is filtered out from the information record sheet of waving；
Wherein, the potential impact factor of the transmission line galloping, comprising: wind speed, wind direction, line alignment and wind direction angle, temperature
Degree, precipitation form, wire icing thickness, landform, height above sea level, voltage class, shaft tower feeder number, twisted wire sectional area, is led humidity
Line arrangement mode and conducting wire division number.
10. method for digging as claimed in claim 9, which is characterized in that it is described based on the potential impact of transmission line galloping because
Son carries out screening acquisition first kind impact factor to information record sheet is waved, further includes:
Numeric form will be converted by the conversion principle of setting by the impact factor of verbal description in first kind impact factor.
11. a kind of digging system for waving data characterized by comprising
Screening module, waves information record sheet for the potential impact factor pair based on transmission line galloping and carries out screening and obtain the
A kind of impact factor；
Sorting module, the support vector machines mould for constructing based on the corresponding related coefficient of the first kind impact factor and in advance
Type is ranked up each impact factor in the first kind impact factor, selects from the first kind impact factor after sequence
Meet the impact factor of preset condition as the second class impact factor；
Analysis module is made for that will wave the corresponding transmission line of electricity related data of the second class impact factor described in information record sheet
Data are waved for wave analysis.
12. digging system as claimed in claim 11, which is characterized in that the sorting module, comprising:
Submodule is generated, for selecting the first impact factor and first impact factor to relate in information record sheet from described wave
And transmission line of electricity related data, generate analytical data；
First computational submodule, it is corresponding for calculating each impact factor in the first kind impact factor based on analytical data
Mean value and mean square deviation；
Second computational submodule, for calculating the phase relation of each impact factor based on the corresponding mean value of each impact factor and mean square deviation
Number；
First sequence is born fruit module, for the related coefficient based on each impact factor absolute value on the first kind influence because
Each impact factor in son is ranked up to obtain the first ranking results；
Second sequence is born fruit module, for being based on analytical data and supporting vector machine model, on the first kind influence because
Each impact factor in son is ranked up to obtain the second ranking results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811029046.0A CN109447309A (en) | 2018-09-05 | 2018-09-05 | A kind of method for digging that waving data and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811029046.0A CN109447309A (en) | 2018-09-05 | 2018-09-05 | A kind of method for digging that waving data and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109447309A true CN109447309A (en) | 2019-03-08 |
Family
ID=65530249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811029046.0A Pending CN109447309A (en) | 2018-09-05 | 2018-09-05 | A kind of method for digging that waving data and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109447309A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929808A (en) * | 2019-12-11 | 2020-03-27 | 国网湖南省电力有限公司 | Multi-element intelligent correction method and system for waving temperature |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203839A (en) * | 2016-07-13 | 2016-12-07 | 国网湖南省电力公司 | Transmission line galloping affects key factor discrimination method and system |
| CN106503751A (en) * | 2016-11-10 | 2017-03-15 | 国网河南省电力公司电力科学研究院 | A kind of power transmission line Louis dance potential prediction method based on SVM classifier |
| CN107451621A (en) * | 2017-08-17 | 2017-12-08 | 国网湖南省电力公司 | Power network waves key influence factor clustering method and system |
-
2018
- 2018-09-05 CN CN201811029046.0A patent/CN109447309A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203839A (en) * | 2016-07-13 | 2016-12-07 | 国网湖南省电力公司 | Transmission line galloping affects key factor discrimination method and system |
| CN106503751A (en) * | 2016-11-10 | 2017-03-15 | 国网河南省电力公司电力科学研究院 | A kind of power transmission line Louis dance potential prediction method based on SVM classifier |
| CN107451621A (en) * | 2017-08-17 | 2017-12-08 | 国网湖南省电力公司 | Power network waves key influence factor clustering method and system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110929808A (en) * | 2019-12-11 | 2020-03-27 | 国网湖南省电力有限公司 | Multi-element intelligent correction method and system for waving temperature |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Moradi Kordmahalleh et al. | A sparse recurrent neural network for trajectory prediction of atlantic hurricanes | |
| CN109492830B (en) | Mobile pollution source emission concentration prediction method based on time-space deep learning | |
| Siwek et al. | Data mining methods for prediction of air pollution | |
| CN110110080A (en) | Textual classification model training method, device, computer equipment and storage medium | |
| CN108985380B (en) | Point switch fault identification method based on cluster integration | |
| CN108898177A (en) | A kind of power transmission line corridor earth surface classification method, system, medium and equipment | |
| CN107341497A (en) | The unbalanced weighting data streams Ensemble classifier Forecasting Methodology of sampling is risen with reference to selectivity | |
| CN110232499A (en) | A kind of power distribution network information physical side method for prewarning risk and system | |
| CN104881689A (en) | Method and system for multi-label active learning classification | |
| CN112285807A (en) | Meteorological information prediction method and device | |
| CN112507861A (en) | Pedestrian detection method based on multilayer convolution feature fusion | |
| CN106644035B (en) | Vibration source identification method and system based on time-frequency transformation characteristics | |
| CN116050666A (en) | Photovoltaic power generation power prediction method for irradiation characteristic clustering | |
| CN114021425B (en) | Power system operation data modeling and feature selection method and device, electronic equipment and storage medium | |
| CN112380945B (en) | Method and system for identifying potential hazards of power transmission channel environment based on satellite remote sensing image | |
| CN109447309A (en) | A kind of method for digging that waving data and system | |
| Aillaud et al. | Day-ahead forecasting of regional photovoltaic production using deep learning | |
| CN113536944A (en) | Distribution line inspection data identification and analysis method based on image identification | |
| Ahmadlou et al. | Comparing ANN and CART to model multiple land use changes: A case study of Sari and Ghaem-Shahr cities in Iran | |
| Noor et al. | Prediction map of rainfall classification using random forest and inverse distance weighted (IDW) | |
| Nagahamulla et al. | Selecting most suitable members for neural network ensemble rainfall forecasting model | |
| CN103942403B (en) | A kind of method and apparatus screened to magnanimity variable | |
| CN106652443B (en) | Method for predicting short-time traffic volume of expressway with similar longitudinal and transverse dimensions | |
| Liu et al. | Swarm intelligence for classification of remote sensing data | |
| CN111680572B (en) | Dynamic judgment method and system for power grid operation scene |
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 |