CN103617356A - Self-adaptive on-line monitoring data trend extraction method - Google Patents

Self-adaptive on-line monitoring data trend extraction method Download PDF

Info

Publication number
CN103617356A
CN103617356A CN201310612784.9A CN201310612784A CN103617356A CN 103617356 A CN103617356 A CN 103617356A CN 201310612784 A CN201310612784 A CN 201310612784A CN 103617356 A CN103617356 A CN 103617356A
Authority
CN
China
Prior art keywords
trend
term
sequence
trend term
data
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
Application number
CN201310612784.9A
Other languages
Chinese (zh)
Other versions
CN103617356B (en
Inventor
陈强
林承华
陈金祥
梁曼舒
何金栋
汤振立
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Fujian Electric Power Co Ltd
State Grid Fujian Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Electric Power Research Institute of State Grid Fujian Electric Power Co Ltd
State Grid Fujian Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, Electric Power Research Institute of State Grid Fujian Electric Power Co Ltd, State Grid Fujian Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN201310612784.9A priority Critical patent/CN103617356B/en
Publication of CN103617356A publication Critical patent/CN103617356A/en
Application granted granted Critical
Publication of CN103617356B publication Critical patent/CN103617356B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Remote Monitoring And Control Of Power-Distribution Networks (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention relates to a self-adaptive on-line monitoring data trend extraction method. According to the method, a combined morphological filter method and an empirical mode decomposition method are combined, firstly, a self-adaptive morphological filter structural element is built dynamically, and filtering is performed by the adoption of the morphological filter structural element so as to perform empirical mode decomposition; secondly, a trend term is built according to the characteristics of processed data; finally, trend pre-alarm is performed according to the trend term. The self-adaptive on-line monitoring data trend extraction method is high in operation performance, high in adaptability and good in use effect.

Description

A kind of adaptive online monitoring data trend extracting method
Technical field
The present invention relates to the extract real-time technical field of online monitoring data variation tendency, particularly the adaptive fast online trend extracting method of a kind of power transmission and transforming equipment online monitoring data.
Background technology
In recent years, repair based on condition of component is extensively promoted in State Grid Corporation of China, as the on-line monitoring of the important technical of repair based on condition of component, also obtains develop rapidly.Online monitoring data in real time, abundant, quantity is huge, the in the situation that of equipment routine test cycle stretch-out, is the important channel of grasping equipment state.But on-Line Monitor Device is easily subject to the factors such as ambient temperature and humidity, load, grid switching operation to be disturbed, and causes data frequent fluctuation, and the method that adopts single acquisition data to carry out state alarm is often reported by mistake, has affected the effect of on-line monitoring.
In online monitoring data, contained a large amount of work informations, if analyse in depth, very valuable for engineering application.For this reason, occurred the methods such as mathematical morphology, empirical mode decomposition (Empirical Mode Decomposition is called for short EMD), wavelet analysis, extracted in reflection equipment, the trend of long-term state, and usage trend carries out state early warning.
A kind of new signal analysis method that the people such as the Norden E Huang of empirical mode decomposition Shi You U.S. NASA proposed in 1998.EMD decomposes be linearity that complicated signal decomposition is become to several intrinsic mode functions (Intrinsic Mode Function is called for short IMF) and a remainder with.IMF has reflected the internal feature of time series signal, and remainder has reflected trend.But the screening process operand that EMD decomposes is very large, when data volume increases substantially, the time cost of computing becomes geometric growth, and operational performance sharply declines, and cannot meet the needs in line drawing trend to mass data.In addition, EMD decomposes the phenomenon that has " crossing screening ", and screening IMF component is out often too much, and some components there is no clear and definite physical meaning.
King waits (2009) quietly and directly uses mathematical morphology filter to carry out filtering processing to online data, and filtered data are as trend.Choosing the effect of this method of structural element and window width thereof is most important, and the waveform morphology of on-line monitoring is varied, is difficult to know instantly select the structural element of which kind of shape to be more suitable for.Determining of window width is also difficult to solve, and window width generally can obtain smoother filter effect than maximum impulse noise width great talent in data, but this cannot a priori know.Therefore, the method for mathematical morphology filter cannot meet the needs that online monitoring data trend is extracted adaptively.
Summary of the invention
The object of the present invention is to provide a kind of adaptive online monitoring data trend extracting method, the method operational performance is high, and adaptivity is strong, and result of use is good.
For achieving the above object, technical scheme of the present invention is: a kind of adaptive online monitoring data trend extracting method, the method is the adaptive morphological filter structural element of dynamic construction first, use described morphological filter structural element to carry out filtering, and then carry out empirical mode decomposition, then according to the feature structure trend term of institute's deal with data, finally according to described trend term, carry out trend early warning.
Further, the method comprises the following steps:
(1) standard deviation of dynamic calculation original series, usings described standard deviation as radius structure semicircular structure element, as morphological filter structural element;
(2) adopt described semicircular structure element to carry out filtering processing to original series, remove the noises such as high frequency, pulse, obtain preliminary trend sequence data;
(3) preliminary trend sequence data are decomposed by empirical mode decomposition method, obtain one group of IMF component and a surplus;
(4), according to the feature of institute's deal with data, select the stack of surplus or several IMF components as trend term r (t);
(5) for trend term r (t), calculate the trend development speed of trend term r (t) in corresponding sequence time length, if trend development speed surpasses the threshold values arranging, send trend early warning, trend development speed V is defined as follows formula:
V = ( r ( t n ) - r ( t 0 ) ) · T t n - t 0
Wherein, t 0the initial time of sequence, t nthe closing time of sequence, r (t 0) be t 0the value of moment trend term, r (t n) be t nthe value of moment trend term, T is reference period.
Further, in step (4), the principle that trend term is selected is as follows: for time span, be 1 year and above sequence, the cycle in all IMF components chosen is greater than the component of 90 days and stacks up as trend term r (t); For time span, be 3 months above 1 year following sequence, the cycle in all IMF components chosen is greater than the component of 30 days and stacks up as trend term r (t); For time span, be 1 month above 3 months following sequences, the cycle in all IMF components chosen is greater than the component of 1 day and stacks up as trend term r (t).
Compared to prior art, the invention has the beneficial effects as follows: operational performance is high, do not have the phenomenon of " crossing screening ", the physical meaning of screening IMF component is out clearer and more definite.The method adopts combination form wave filter to carry out filtering to original series, has removed noise contribution, and this has accelerated follow-up EMD screening process, has reduced the IMF component without clear and definite physical significance simultaneously, thereby has greatly improved operational performance.In addition, the trend term of the inventive method can objectively respond Sequence Trend essence, and has very strong adaptivity.The structural element that this method filtering is used is according to original series dynamic calculation, without manually a priori selected, thus needs that can all kinds of sequential filterings of dynamically adapting.
Accompanying drawing explanation
Fig. 1 is the workflow diagram of the inventive method.
Fig. 2 is the embodiment of the present invention Zhong Mou 110kVI of the transformer station section bus mutually resistive current data of lightning arrester A and trend map thereof.
Embodiment
Below in conjunction with drawings and the specific embodiments, the present invention is described in further detail.
The invention provides a kind of adaptive online monitoring data trend extracting method, the method combines combination form wave filter and two kinds of methods of empirical mode decomposition, first the adaptive morphological filter structural element of dynamic construction, use described morphological filter structural element to carry out filtering, and then carry out empirical mode decomposition, then according to the feature structure trend term of institute's deal with data, finally according to described trend term, carry out trend early warning.The method specifically comprises the following steps:
(1) standard deviation of dynamic calculation original series, usings described standard deviation as radius structure semicircular structure element, as morphological filter structural element.
(2) adopt described semicircular structure element to carry out filtering processing to original series, remove the noises such as high frequency, pulse, obtain preliminary trend sequence data.
(3) preliminary trend sequence data are decomposed by empirical mode decomposition method, obtain one group of IMF component and a surplus.
(4), according to the feature of institute's deal with data, select the stack of surplus or several IMF components as trend term r (t).The principle that trend term is selected is as follows: for time span, be 1. 1 year and above sequence, consider that it is the longest influence factor of cycle that online monitoring data is subject to the impact of Seasonal, thereby the cycle of its trend part should be greater than 90 days, the cycle in all IMF components chosen is greater than the component of 90 days and stacks up as trend term r (t); 2. for time span, be 3 months above 1 year following sequence, the cycle in all IMF components chosen is greater than the component of 30 days and stacks up as trend term r (t); 3. for time span, be 1 month above 3 months following sequences, consider the impact that is subject to ambient temperature and humidity and load, the cycle in all IMF components chosen is greater than the component of 1 day and stacks up as trend term r (t).
(5) for trend term r (t), calculate the trend development speed of trend term r (t) in corresponding sequence time length, if trend development speed surpasses the threshold values arranging, send trend early warning, trend development speed V is defined as follows formula:
V = ( r ( t n ) - r ( t 0 ) ) · T t n - t 0
Wherein, t 0the initial time of sequence, t nthe closing time of sequence, r (t 0) be t 0the value of moment trend term, r (t n) be t nthe value of moment trend term, T is reference period, as 30,90,360 days, corresponding was the speed of development in the moon, season, year.
The inventive method operational performance is high, does not have the phenomenon of " crossing screening ", and the physical meaning of screening IMF component is out clearer and more definite.This method adopts combination form wave filter to carry out filtering to original series, has removed noise contribution, and this has accelerated follow-up EMD screening process, has reduced the IMF component without clear and definite physical significance simultaneously, thereby has greatly improved operational performance.
Relatively adopt the operational performance of EMD method and this method below.266 Zinc-Oxide Arrester equipment are randomly drawed in experiment, respectively the data of the total current of online acquisition (144 data every day) 1 month and 2 months are processed, program adopts JAVA language compilation, on the PC of CPU2.93GHZ, internal memory 2G, moves, and statistics is as table 1.
The operational performance comparison of table 1EMD method and this method
Extracting method Time long (moon) Average calculating operation time (ms) Average weight number (individual)
FMD 1 5803 7
This method 1 243 5
EMD 2 44946 8
This method 2 418 5.5
Experiment shows, the trend of 1 month data is extracted, and be 23.8 times of this method the average calculating operation time of EMD method, reached 107 times during 2 months data.Visible, this method has significantly improved operational performance, and particularly, along with the growth of data volume, sharply reduce average calculating operation time especially, can meet online the trend of big data quantity is extracted to requirement in performance.
The trend term of the inventive method can objectively respond Sequence Trend essence, and has very strong adaptivity.The structural element that this method filtering is used is according to original series dynamic calculation, without manually a priori selected, thus needs that can all kinds of sequential filterings of dynamically adapting.Definite quality that has determined that trend is extracted of trend term, this method has proposed to choose which component as the principle of trend term from the angle in IMF component cycle, thereby solved previous methods dependence experience, determines which component is as the problem of trend term.This can be from the explanation of following instantiation: in March, 2009, the trend development speed V of current in resistance property of the 110kVI of ,Mou transformer station section bus lightning arrester A phase reached for the 27.5 μ A/ months (Fig. 2), had surpassed trend early warning range.Continuous Tracking, V value April was the 32.9 μ A/ months, and be the 39 μ A/ months May, and be the 47.3 μ A/ months June, has obvious accelerated development trend.Adopt live testing to survey its current in resistance property, variation tendency trend basic and that MM-EMD extracts matches.Afterwards this lightning arrester is carried out to high-potting, the 7uA by commissioning test in 2003 under DC test 75%U1mA rises to 38uA; Through strip inspection, this lightning arrester insulation bucket makes moist.
Be more than preferred embodiment of the present invention, all changes of doing according to technical solution of the present invention, when the function producing does not exceed the scope of technical solution of the present invention, all belong to protection scope of the present invention.

Claims (3)

1. an adaptive online monitoring data trend extracting method, it is characterized in that, the method is the adaptive morphological filter structural element of dynamic construction first, use described morphological filter structural element to carry out filtering, and then carry out empirical mode decomposition, then according to the feature structure trend term of institute's deal with data, finally according to described trend term, carry out trend early warning.
2. a kind of adaptive online monitoring data trend extracting method according to claim 1, is characterized in that, comprises the following steps:
(1) standard deviation of dynamic calculation original series, usings described standard deviation as radius structure semicircular structure element, as morphological filter structural element;
(2) adopt described semicircular structure element to carry out filtering processing to original series, remove the noises such as high frequency, pulse, obtain preliminary trend sequence data;
(3) preliminary trend sequence data are decomposed by empirical mode decomposition method, obtain one group of IMF component and a surplus;
(4), according to the feature of institute's deal with data, select the stack of surplus or several IMF components as trend term r (t);
(5) for trend term r (t), calculate the trend development speed of trend term r (t) in corresponding sequence time length, if trend development speed surpasses the threshold values arranging, send trend early warning, trend development speed V is defined as follows formula:
V = ( r ( t n ) - r ( t 0 ) ) · T t n - t 0
Wherein, t 0the initial time of sequence, t nthe closing time of sequence, r (t 0) be t 0the value of moment trend term, r (t n) be t nthe value of moment trend term, T is reference period.
3. a kind of adaptive online monitoring data trend extracting method according to claim 2, it is characterized in that, in step (4), the principle that trend term is selected is as follows: for time span, be 1 year and above sequence, the cycle in all IMF components chosen is greater than the component of 90 days and stacks up as trend term r (t); For time span, be 3 months above 1 year following sequence, the cycle in all IMF components chosen is greater than the component of 30 days and stacks up as trend term r (t); For time span, be 1 month above 3 months following sequences, the cycle in all IMF components chosen is greater than the component of 1 day and stacks up as trend term r (t).
CN201310612784.9A 2013-11-27 2013-11-27 A kind of adaptive online monitoring data trend abstraction method Active CN103617356B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310612784.9A CN103617356B (en) 2013-11-27 2013-11-27 A kind of adaptive online monitoring data trend abstraction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310612784.9A CN103617356B (en) 2013-11-27 2013-11-27 A kind of adaptive online monitoring data trend abstraction method

Publications (2)

Publication Number Publication Date
CN103617356A true CN103617356A (en) 2014-03-05
CN103617356B CN103617356B (en) 2016-09-07

Family

ID=50168059

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310612784.9A Active CN103617356B (en) 2013-11-27 2013-11-27 A kind of adaptive online monitoring data trend abstraction method

Country Status (1)

Country Link
CN (1) CN103617356B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105203346A (en) * 2015-10-23 2015-12-30 珠海格力电器股份有限公司 Fault diagnosis method and system for range hood based on EMD (Empirical Mode Decomposition) noise reduction
CN107730494A (en) * 2017-10-24 2018-02-23 河海大学 A kind of anchor pole detection method based on variation mode decomposition
CN110824304A (en) * 2019-10-16 2020-02-21 福建和盛高科技产业有限公司 Method for analyzing insulation degradation trend of zinc oxide arrester
CN111121955A (en) * 2019-12-31 2020-05-08 华侨大学 Vibration signal wavelet analysis trend term removing method adaptive to threshold frequency
CN113324646A (en) * 2021-05-26 2021-08-31 兰州交通大学 Big wind area electrified railway contact net positive feeder line galloping positioning algorithm

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102540021A (en) * 2010-12-20 2012-07-04 上海市电力公司 Method for monitoring GIS (gas insulated switchgear) by aid of high-frequency partial discharge online system

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102540021A (en) * 2010-12-20 2012-07-04 上海市电力公司 Method for monitoring GIS (gas insulated switchgear) by aid of high-frequency partial discharge online system

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
HAN LI,ET AL.: "Fault diagnosis using pattern classification based on one-dimensional adaptive rank-order morphological filter", 《JOURNAL OF PROCESS CONTROL》, 31 December 2012 (2012-12-31), pages 436 - 449, XP 028891576, DOI: doi:10.1016/j.jprocont.2011.12.005 *
汤宝平,等.: "基于形态奇异值分解和经验模态分解的滚动轴承故障特征提取方法", 《机械工程学报》, vol. 46, no. 5, 31 March 2010 (2010-03-31) *
王楠.: "电容型设备绝缘在线监测与故障诊断的研究", 《中国博士学位论文全文数据库 工程科技Ⅱ辑》, no. 01, 15 March 2005 (2005-03-15), pages 042 - 38 *
陈隽,等.: "经验模分解在信号趋势项提取中的应用", 《振动、测试与诊断》, vol. 25, no. 2, 30 June 2005 (2005-06-30) *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105203346A (en) * 2015-10-23 2015-12-30 珠海格力电器股份有限公司 Fault diagnosis method and system for range hood based on EMD (Empirical Mode Decomposition) noise reduction
CN107730494A (en) * 2017-10-24 2018-02-23 河海大学 A kind of anchor pole detection method based on variation mode decomposition
CN110824304A (en) * 2019-10-16 2020-02-21 福建和盛高科技产业有限公司 Method for analyzing insulation degradation trend of zinc oxide arrester
CN111121955A (en) * 2019-12-31 2020-05-08 华侨大学 Vibration signal wavelet analysis trend term removing method adaptive to threshold frequency
CN111121955B (en) * 2019-12-31 2022-02-18 华侨大学 Vibration signal wavelet analysis trend term removing method adaptive to threshold frequency
CN113324646A (en) * 2021-05-26 2021-08-31 兰州交通大学 Big wind area electrified railway contact net positive feeder line galloping positioning algorithm

Also Published As

Publication number Publication date
CN103617356B (en) 2016-09-07

Similar Documents

Publication Publication Date Title
CN103617356A (en) Self-adaptive on-line monitoring data trend extraction method
Franzke Nonlinear trends, long-range dependence, and climate noise properties of surface temperature
CN105629136A (en) Cable insulation state online automatic monitoring and diagnosis system
CN111371188A (en) Real-time early warning system for abnormal power consumption data
CN103197124A (en) Overvoltage identification method based on time-frequency matrix singular value
CN104330676A (en) Transformer substation overvoltage intelligence monitoring system and method
CN105548739B (en) A kind of arrester operating state signal processing method
CN104091056B (en) A kind of laboratory equipment life forecast system and method
CN104545887A (en) Method and device for identifying artifact electrocardiograph waveforms
He et al. Feature extraction of acoustic signals based on complex Morlet wavelet
CN103018537B (en) The Classification of Transient Power Quality Disturbances recognition methods of kurtosis is composed based on CWD
CN103198184A (en) Low-frequency oscillation character noise-like identification method in electric power system
CN103346797A (en) Real-time compression method for gear fault signals
CN109840691A (en) Non-intrusion type subitem electricity estimation method based on deep neural network
CN103308829B (en) A kind of GIS single Partial discharge signal extracts and trigger instants method of adjustment
CN110531234A (en) A kind of identification extracting method of transmission line of electricity discharge pulse
CN105954695A (en) Synchronization-based homogeneous-sensor mutation parameter recognizing method and device
CN112183400A (en) Novel latent fault feature extraction method and system for distribution transformer
CN114184870A (en) Non-invasive load identification method and equipment
Saini et al. Application of Hilbert-Huang Transform in the field of power quality events analysis
CN108594156B (en) Improved current transformer saturation characteristic identification method
CN115144703A (en) High-resistance grounding fault identification method based on zero-sequence differential current and energy moment indexes
Wali et al. High impedance fault detection based on power spectrum technique
Viet et al. A method for monitoring voltage disturbances based on discrete wavelet transform and adaptive linear neural network
CN103258144B (en) Online static load modeling method based on data of fault recorder

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant