阅读说明：本技术 一种gis设备的局部放电类型分类方法 (A kind of shelf depreciation genre classification methods of GIS device ) 是由 彭兆裕 岳丹 洪志湖 代维菊 文刚 颜冰 马御棠 李佑明 张云 于 2019-10-14 设计创作，主要内容包括：本申请公开了一种GIS设备的局部放电类型分类方法,利用uPMT传感器阵列和电压传感器采集局部放电的光辐射强度-工频电压相位-光辐射波长三维数据；利用特征提取法提取光辐射强度-工频电压相位-光辐射波长三维数据,得到特征波段下的光辐射强度-工频电压相位信息；光辐射强度-工频电压相位信息分别通过多个特征运算,得到多个特征运算对应的特征量；多个特征量分别输入分类器中,得到局部放电的放电类型。本申请中,通过特征提取法提取采集到的数据量庞大的三维数据,作为放电信息特征参量,减少了数据分析量,提高数据分析效率；经过多个特征运算得到对应的特征量,并将多个特征量输入分类器中,得到放电类型,提高分类准确性。(This application discloses a kind of shelf depreciation genre classification methods of GIS device, utilize light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data of uPMT sensor array and voltage sensor acquisition shelf depreciation；Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data is extracted using feature extraction, obtains light radiation intensity-power-frequency voltage phase information under characteristic wave bands；Light radiation intensity-power-frequency voltage phase information by multiple characteristic operations, obtains the corresponding characteristic quantity of multiple characteristic operations respectively；Multiple characteristic quantities input in classifier respectively, obtain the electric discharge type of shelf depreciation.In the application, the huge three-dimensional data of collected data volume is extracted by feature extraction, as discharge information characteristic parameter, reduces data amount of analysis, improves data analysis efficiency；Corresponding characteristic quantity is obtained by multiple characteristic operations, and multiple characteristic quantities are inputted in classifier, electric discharge type is obtained, improves classification accuracy.)

1. a kind of shelf depreciation genre classification methods of GIS device, which is characterized in that the described method includes:

Utilize light radiation intensity-power-frequency voltage phase-light spoke of uPMT sensor array and voltage sensor acquisition shelf depreciation The long three-dimensional data of ejected wave；

Light radiation intensity-power-frequency voltage phase-light radiation wavelength the three-dimensional data is extracted using feature extraction, obtains feature Light radiation intensity-power-frequency voltage phase information under wave band；

By multiple characteristic operations, it is corresponding to obtain multiple characteristic operations respectively for the light radiation intensity-power-frequency voltage phase information Characteristic quantity；

Multiple characteristic quantities input in classifier respectively, obtain the electric discharge type of shelf depreciation.

2. shelf depreciation genre classification methods according to claim 1, which is characterized in that described to utilize uPMT sensor array Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data of column and voltage sensor acquisition shelf depreciation, comprising:

The light radiation intensity of ultraviolet, visible light and infrared three wave bands using uPMT sensor array acquisition shelf depreciation；

Using the power-frequency voltage phase of voltage sensor acquisition shelf depreciation；

The light radiation intensity and power-frequency voltage phase of ultraviolet, visible light and infrared three wave bands are collected, it is strong to obtain light radiation Degree-power-frequency voltage phase-light radiation wavelength three-dimensional data.

3. shelf depreciation genre classification methods according to claim 1, which is characterized in that described to be mentioned using feature extraction Light radiation intensity-the power-frequency voltage phase-light radiation wavelength three-dimensional data is taken, the light radiation intensity-under characteristic wave bands is obtained Power-frequency voltage phase information, comprising:

Extract the light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data, and by ultraviolet, visible light and red Outer three broadbands are divided into narrow-band；

Correlation between each narrow-band and original wave band, the correlation between each narrow-band are analyzed, extracting has discharge signal The multiple groups narrow-band of feature obtains light radiation intensity-power-frequency voltage phase information under characteristic wave bands as characteristic wave bands.

4. shelf depreciation genre classification methods according to claim 1, which is characterized in that the light spoke under the characteristic wave bands Penetrating intensity-power-frequency voltage phase information includes light radiation intensity-power-frequency voltage phase information, visible light wave range under ultraviolet band Light radiation intensity-power-frequency voltage phase information under lower light radiation intensity-power-frequency voltage phase information and infrared band.

5. shelf depreciation genre classification methods according to claim 1, which is characterized in that the characteristic operation includes deflection Degree, standout, degree of asymmetry, slope, binaryzation coding and normalization.

6. shelf depreciation genre classification methods according to claim 1, which is characterized in that the electric discharge class of the shelf depreciation Type includes point discharge, suspended discharge, creeping discharge, the electric discharge of free metal particle.

Technical field

This application involves switched power equipment state diagnostic field more particularly to a kind of shelf depreciation types of GIS device Classification method.

Background technique

GIS device has the significant advantages such as operational reliability is high, the time between overhauls(TBO) is long, equipment volume is small, safety and stability fortune Row has great significance to the reliability of electric system.However the gas-insulateds electricity such as gas-insulated switch, gas power transfer conduit Power equipment inevitably results from conductor tip, scuffing, floating potential etc. in production, transport, assembly and operational process Insulation defect, such insulation defect can generate the serious safeties things such as shelf depreciation even insulation breakdown under forceful electric power field action Therefore.Therefore GIS device failure initial stage find in time and diagnose fault type to ensure equipment safe and reliable operation have it is aobvious The meaning of work.

Partial Discharge Detection is considered as detecting most effective detection means at electrical equipment fault initial stage, and common part is put Electric detection method is superfrequency method.There are when partial discharge phenomenon inside GIS device, generated extra-high frequency electromagnetic waves can It is distally propagated along the tube body of GIS device, the identification for shelf depreciation type, it is accurate to extract characteristic parameter and improve classification Device nicety of grading is always the research emphasis of shelf depreciation electric discharge type identification work, is detected at this stage in partial-discharge ultrahigh-frequency Middle characteristic parameter is concentrated mainly on extraction image grayscale fractal characteristic, statistical nature, small echo in shelf depreciation statistical distribution map Feature, single pulse waveforms feature etc., the ununified standard of the selection of characteristic parameter, are voluntarily extracted by researcher, basic herein It is upper to realize electric discharge type pattern-recognition using classifiers such as neural network, support vector machine fuzzy clusterings.

Due to electromagnetic wave signal by inside GIS device there are the decaying that will cause electromagnetic wave energy when discontinuity point and Reflection, the parameters such as the waveform of signal and amplitude change when electromagnetic wave signal travels to extra-high video sensor in GIS device, To increase the complexity of detection local discharge signal, while the partial discharge monitoring system of large electric equipment is in various In the site environment of interference, superfrequency electromagnetic wave detection device is easy to be interfered, and follow-up mode recognition accuracy also will receive It influences.

Summary of the invention

This application provides a kind of shelf depreciation genre classification methods of GIS device, to solve to be based on superfrequency detection side The low technical problem of the classification accuracy of method.

In order to solve the above-mentioned technical problem, the embodiment of the present application discloses following technical scheme:

A kind of shelf depreciation genre classification methods of GIS device, method include: to be passed using uPMT sensor array and voltage Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data of sensor acquisition shelf depreciation；It is mentioned using feature extraction Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data is taken, light radiation intensity-power frequency under characteristic wave bands is obtained Voltage phase information；Light radiation intensity-power-frequency voltage phase information by multiple characteristic operations, obtains multiple characteristic operations respectively Corresponding characteristic quantity；Multiple characteristic quantities input in classifier respectively, obtain the electric discharge type of shelf depreciation.

Optionally, light radiation intensity-power frequency electric of uPMT sensor array and voltage sensor acquisition shelf depreciation is utilized Press phase-light radiation wavelength three-dimensional data, comprising: shelf depreciation is ultraviolet using the acquisition of uPMT sensor array, visible light and The light radiation intensity of infrared three wave bands；Using the power-frequency voltage phase of voltage sensor acquisition shelf depreciation；It will be ultraviolet, visible The light radiation intensity and power-frequency voltage phase of light and infrared three wave bands collect, and obtain light radiation intensity-power-frequency voltage phase- Light radiation wavelength three-dimensional data.

Optionally, light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data is extracted using feature extraction, Obtain light radiation intensity-power-frequency voltage phase information under characteristic wave bands, comprising: extract light radiation intensity-power-frequency voltage phase- Light radiation wavelength three-dimensional data, and ultraviolet, visible light and infrared three broadbands are divided into narrow-band；Analyze each narrow-band The correlation between correlation, each narrow-band between original wave band extracts the multiple groups narrow-band with discharge signal feature As characteristic wave bands, light radiation intensity-power-frequency voltage phase information under characteristic wave bands is obtained.

Optionally, the light radiation intensity under characteristic wave bands-power-frequency voltage phase information includes the light radiation under ultraviolet band Light spoke under light radiation intensity-power-frequency voltage phase information and infrared band under intensity-power-frequency voltage phase information, visible light wave range Penetrate intensity-power-frequency voltage phase information.

Optionally, characteristic operation includes degree of skewness, standout, degree of asymmetry, slope, binaryzation coding and normalization.

Optionally, the electric discharge type of shelf depreciation includes point discharge, suspended discharge, creeping discharge, free metal particle Electric discharge.

The utility model has the advantages that this application provides a kind of shelf depreciation genre classification methods of GIS device, firstly, utilizing uPMT Light radiation intensity-power-frequency voltage phase-three dimension of light radiation wavelength of sensor array and voltage sensor acquisition shelf depreciation According to；Secondly, extracting light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data using feature extraction, feature is obtained Light radiation intensity-power-frequency voltage phase information under wave band；Again, light radiation intensity-power-frequency voltage phase information passes through respectively Multiple characteristic operations obtain the corresponding characteristic quantity of multiple characteristic operations；Finally, obtaining office in multiple characteristic quantity input classifiers The electric discharge type of portion's electric discharge.In the application, the huge three-dimensional data of collected data volume is extracted by feature extraction, as Discharge information characteristic parameter reduces data amount of analysis, improves data analysis efficiency；It is obtained by multiple characteristic operations corresponding Characteristic quantity, and multiple characteristic quantities are inputted in classifier, electric discharge type is obtained, classification accuracy is improved.

Detailed description of the invention

In order to illustrate more clearly of the technical solution of the application, letter will be made to attached drawing needed in the embodiment below Singly introduce, it should be apparent that, for those of ordinary skills, without any creative labor, It is also possible to obtain other drawings based on these drawings.

Fig. 1 is a kind of flow chart of the shelf depreciation genre classification methods for GIS device that application provides.

Specific embodiment

It is a kind of flow chart of the shelf depreciation genre classification methods of GIS device provided by the present application referring to Fig. 1, it is known that, This application provides a kind of shelf depreciation genre classification methods of GIS device, method includes:

S01: light radiation intensity-power-frequency voltage phase of uPMT sensor array and voltage sensor acquisition shelf depreciation is utilized Position-light radiation wavelength three-dimensional data.

S011: the light spoke of ultraviolet, visible light and infrared three wave bands using uPMT sensor array acquisition shelf depreciation Penetrate intensity.

The light radiation letter of ultraviolet, visible light and infrared three wave bands using uPMT sensor array acquisition shelf depreciation Number, and electric signal is converted by optical radiation signal.

S012: using the power-frequency voltage phase of voltage sensor acquisition shelf depreciation.

S013: the light radiation intensity and power-frequency voltage phase of ultraviolet, visible light and infrared three wave bands are collected, obtained Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data.

Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data filters noise signal by pretreatment It removes, and useful signal is amplified.Light radiation intensity-power frequency is collected by uPMT sensor array and voltage sensor Voltage-phase-light radiation wavelength three-dimensional data, can reduce electromagnetic interference capability, and by pretreatment after, increase acquisition letter Number validity.

S02: light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data is extracted using feature extraction, is obtained Light radiation intensity-power-frequency voltage phase information under characteristic wave bands.

Light radiation intensity-power-frequency voltage phase information under characteristic wave bands includes light radiation intensity-work under ultraviolet band Light radiation intensity-under light radiation intensity-power-frequency voltage phase information and infrared band under frequency voltage phase information, visible light wave range Power-frequency voltage phase information.

S021: extract light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data, and by ultraviolet, visible light with And infrared three broadbands are divided into narrow-band；

S022: analyzing correlation between each narrow-band and original wave band, the correlation between each narrow-band, and extraction has The multiple groups narrow-band of discharge signal feature obtains light radiation intensity-power-frequency voltage phase under characteristic wave bands as characteristic wave bands Information.

The extraction principle of characteristic wave bands be seek that correlation between each narrow-band and original wave band is big and each narrow-band it Between correlation is small, n wave band with discharge signal feature is as characteristic wave bands.Operation can be reduced by feature extracting method Data eliminate redundant data.

S03: light radiation intensity-power-frequency voltage phase information by multiple characteristic operations, obtains multiple characteristic operations respectively Corresponding characteristic quantity.

Characteristic operation includes degree of skewness, standout, degree of asymmetry, slope, binaryzation coding and normalization.By between wave band The operation of related data enhances otherness and feature between data, increases characteristic parameter quantity and dimension for classification Degree.

S04: multiple characteristic quantities input in classifier respectively, obtain the electric discharge type of shelf depreciation.

The electric discharge type of shelf depreciation includes point discharge, suspended discharge, creeping discharge, the electric discharge of free metal particle.

In view of different types of sorting algorithm is for the having differences property of classifying quality of different electric discharge types, therefore this Shen Middle a variety of sorting algorithms based on different classifications criterion please be selected as assembled classifier, and by all kinds of algorithm classification results The mode being weighted obtains final classification as a result, weighting scheme includes but is not limited to assignment coefficient weighting method, calculation method Are as follows:

In formula, f_{Sharp, outstanding, edge, gold}It is that classification results are put for point discharge, suspended discharge, creeping discharge, free metal particle respectively The probability of electricity, f_{Wave spectrum angle 1,2,3,4}Respectively wave spectrum angle matching judgment is the probability that 4 classes are discharged, f_{BP1、2、3、4}Respectively BP neural network It is judged as the probability of 4 classes electric discharge, f_{Support vector machines 1,2,3,4}Respectively support vector machines matching judgment is the probability that 4 classes are discharged, f_{Random forest 1,2,3,4}Respectively random forest is judged as the probability that 4 classes are discharged, f_{Parallelepiped 1,2,3,4}Respectively parallelepiped is judged as 4 The probability of class electric discharge, f_{Fuzzy clustering 1,2,3,4}Respectively fuzzy clustering is judged as the probability that 4 classes are discharged.A, b, c, d are respectively various sides The weight coefficient of method, it is model which, which needs above formula, and training obtains after bringing a large amount of actual measured results into.Finally compare Compared with f_Point,f_Outstanding,f_Along,f_GoldValue, one of maximum probability is classification results.The sorting algorithm that different classifications device uses has differences Property, classifying quality also have differences with data type and quantity etc., can be effective by Combination of Multiple Classifiers and joint classification Improve nicety of grading.

This application provides a kind of shelf depreciation genre classification methods of GIS device, firstly, utilizing uPMT sensor array With light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data of voltage sensor acquisition shelf depreciation；Secondly, sharp Light radiation intensity-power-frequency voltage phase-light radiation wavelength three-dimensional data is extracted with feature extraction, obtains the light under characteristic wave bands Radiation intensity-power-frequency voltage phase information；Again, light radiation intensity-power-frequency voltage phase information is transported by multiple features respectively It calculates, obtains the corresponding characteristic quantity of multiple characteristic operations；Finally, obtaining putting for shelf depreciation in multiple characteristic quantity input classifiers Electric type.In the application, the huge three-dimensional data of collected data volume is extracted by feature extraction, as discharge information spy Parameter is levied, data amount of analysis is reduced, improves data analysis efficiency；Corresponding characteristic quantity is obtained by multiple characteristic operations, and Multiple characteristic quantities are inputted in classifier, electric discharge type is obtained, improve classification accuracy.

Since embodiment of above is that reference combination is illustrated on other modes, have between different embodiments There is identical part, identical, similar part may refer to each other between each embodiment in this specification.Herein no longer in detail It illustrates.

It should be noted that in the present specification, the relational terms of such as " first " and " second " or the like are used merely to It distinguishes one entity or operation from another entity or operation, and not necessarily requires or imply these entities or operation Between there are any this actual relationship or sequences.Moreover, the terms "include", "comprise" or its any other variant are intended to Cover non-exclusive inclusion, so that the circuit structure, article or the equipment that include a series of elements not only include those Element, but also including other elements that are not explicitly listed, or further include for this circuit structure, article or equipment Intrinsic element.In the absence of more restrictions, the element for thering is sentence "including a ..." to limit, it is not excluded that There is also other identical elements in circuit structure, article or equipment including the element.

Those skilled in the art will readily occur to its of the application after considering specification and practicing the disclosure applied here His embodiment.This application is intended to cover any variations, uses, or adaptations of the application, these modifications, purposes or Person's adaptive change follows the general principle of the application and including the undocumented common knowledge in the art of the application Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the application are wanted by right The content asked is pointed out.

Above-described the application embodiment does not constitute the restriction to the application protection scope.