CN105512962A - Method for comprehensively evaluating insulation status of gas insulated switchgear (GIS) - Google Patents

Abstract

The invention belongs to the technical field of online monitoring and evaluation of insulation statuses of a gas insulated switchgear (GIS), and particularly relates to a method for comprehensively evaluating the insulation status of the gas insulated switchgear (GIS). The method mainly comprises the steps of GIS insulation status evaluation index system framework construction, GIS insulation evaluation index weight calculation, GIS insulation status partition based on a fuzzy theory, and multi-source status evaluation mathematical model fusion based on an evidence theory. The method involves evaluation indexes, is comprehensive and reliable in evaluation model, has simple and clear evaluation results, and can be popularized and applied to actual engineering fields. In consideration of limited running condition of the GIS, a comprehensive index system reflecting the insulation status of equipment can be constructed by using complementary information under different monitoring conditions, and the fuzzy theory and a data fusing theory are introduced in conjunction with expert experience, so that comprehensive evaluation of the insulation status of the GIS equipment is realized, and a comprehensive and reliable scheme is provided for multi-source information evaluation of the same type of equipment.

Description

A kind of gas insulated combined electrical equipment state of insulation comprehensive estimation method

Technical field

The invention belongs to state of insulation on-line monitoring and the assessment technology field of gas insulated combined electrical equipment, be specifically related to a kind of gas insulated combined electrical equipment state of insulation comprehensive estimation method.

Background technology

Along with the quickening of social process of industrialization, the power equipment put into operation is increasing, and coverage is also increasing, and the power outage caused due to electrical equipment fault is not only made troubles to life, also result in serious economic loss.While carrying out large quantity research to Power System Reliability, the reliability of power equipment also can not be ignored, the aging problem of especially apparatus insulated performance under the Various Complex factor actings in conjunction such as working voltage, environment temperature/humidity.

Gas insulated combined electrical equipment (GasInsulatedSwitchgear, GIS) is also known as SF ₆totally-enclosed insulation in combined electric appliance is key equipment of equal importance with transformer in power transmission and transformation system.It adopts has the SF6 gas of superior isolation and arc extinction performance as insulating medium, is sealed in a metal shell by the elements combinations such as isolating switch, disconnector and mutual inductor.Due to compactedness and the complicacy of GIS interior spatial structure, make GIS internal evaluation and maintenance very difficulty, and fault also can affect the performance of non-faulting element in the propagation of the inner finite space, causes wider fault.GIS device failure cause is various, but accounts for half by the equipment failure caused of insulating.Especially for the GIS device put into operation more than 20 years, the risk occurred due to insulation degradation causing trouble can not be despised especially.

Realize GIS device repair based on condition of component and ensure that namely the primary prerequisite of its safe and reliable operation is accurate evaluation actual insulation state and the early stage Hidden fault of identification.GIS device Condition assessment of insulation refers to the actual motion state current according to equipment, bonding apparatus essential history information, to the comprehensive evaluation that the current residing health status of equipment is implemented.The evaluation index of reaction state of insulation is numerous, comprise the influence factor of not ipsilateral, under same factor, different characteristic parameter differs to assessment result influence degree, and these all make GIS device Condition assessment of insulation not a duck soup, is also one of reason of rarer this respect research at present.

At present, the research of setting up in the Condition assessment of insulation scheme of power transmission and transformation equipment both at home and abroad mainly concentrates on transformer, the equipment such as cable, as " method for evaluating oil paper insulation ageing state based on local discharge characteristic parameter " that the patent No. is ZL200810233096.0, the patent No. is " insulating state on-line monitoring method of cross-linked PE cable " of ZL200610041888.9 etc., and for the status monitoring of gas insulated combined electrical equipment and assessment aspect, more concentrate on the study on monitoring to fault, as " a kind of on-line monitoring locating device of local discharge of gas-insulator switchgear " that the patent No. is ZL201020204616.8, the patent No. is " detecting method and the detection and location device of the inner shelf depreciation of gas insulated combined electric appliance equipment " of ZL201010156542.X etc., data rarely based on the existing off-line to GIS and on-line monitoring provide the scheme assessing GIS dielectric level further.

Due to compactedness and the complicacy of GIS interior spatial structure, make GIS internal evaluation and maintenance very difficulty, and fault also can affect the performance of non-faulting element in the propagation of the inner finite space, causes wider fault.And the average power off time overhauled required for a GIS device is than conventional electric substation head, the economic loss caused is also larger.Therefore, if the early stage sign of latency insulation fault can be identified, and accurate judgement is made to its state of insulation, best maintenance schedule can be arranged before GIS device is about to break down, avoid fault to cause larger economic loss.

Summary of the invention

The object of the invention is for existing SF ₆the various methods that the PD of gas insulated electric apparatus detects its true state of insulation have its limitation, provide kind of a gas insulated combined electrical equipment state of insulation comprehensive estimation method.GIS device state monitoring method is numerous and monitoring index is more, and for there is ambiguity and uncertainty between index, and the dimension of various index is different, and span is also not quite similar situation, and this programme is introduced fuzzy mathematics theory and processed index degree of membership.Meanwhile, the influence degree of different state parameter index for assessment results is also not quite similar, and the present invention program introduces the affect result of evidence theory on Different factor layer and carries out fusion assessment.The present invention sets up and provides technical support based on the GIS device Condition assessment of insulation of multi source status parameter and fault diagnosis model, there is provided scientific basis and basis for finally realizing GIS device repair based on condition of component, carrying out equipment control to electric power enterprise will have higher construction value.

Technical scheme of the present invention is:

A kind of gas insulated combined electrical equipment state of insulation comprehensive estimation method, the method is characterized in that, specifically comprise:

The sub-step that a GIS Condition assessment of insulation System Framework builds: GIS device state of insulation is formed Recurison order hierarchy according to the source attribute difference of three class status informations, is followed successively by the evaluation index layer evaluated layer, factor layer and each factor layer and select from top to bottom; Wherein,

Layer one: evaluate layer and be decided to be H ₁, H ₂, H ₃, H ₄one of four states:

State 1., normal condition H ₁level: every quantity of state of equipment is in normal table;

State 2., attention state H ₂level: patrol and examine some quantity of state of middle discovery and change and have the possible trend close to warning value;

State 3., abnormality H ₃level: a certain or various states amount is only slight beyond allowable value, and reasonable arrangement Strategies of Maintenance does suitable process to abnormality;

State 4., severe conditions H ₄level: monitor signal has strong exception, quantity of state is seriously above standard limit value, should arrange interruption maintenance even more exchange device as early as possible;

Layer two: factor layer then chooses PD electric parameter, PD chemical parameters, GIS device preventive trial parameter are defined as F={f ₁, f ₂, f ₃, namely to the factor classification of returning of considered index;

Layer three: indicator layer chooses the PD electric parameter f of characterization device state of insulation ₁={ e ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, PD chemical parameters and the characteristic parameter f of GIS device preventive trial ₃={ e ₃₁, e ₃₂, e ₃₃the state of insulation indicator layer of state of insulation index;

The sub-step of a GIS insulation evaluation index weight calculation:

Step 1.1,

E ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, e ₂₁, e ₂₂, e ₂₃, e ₂₄, e ₂₅, e ₂₆, e ₃₁, e ₃₂, e ₃₃for indicator layer is defined as ground floor, as the Information base that evaluating system is the most basic, it is selected and sequence plays vital effect to the result of decision, and the significance level to each index of this step qualitative and quantitative carries out weight evaluation;

F={f ₁, f ₂, f ₃factor layer is defined as the second layer, its a certain factor A comprehensively determines by n evaluation index, and m expert can compare any two indices according to experience knowledge, the judgment matrix A that a kth expert obtains ^(k)for: wherein represent and carry out the quantification of expert opinion with modal 1 ~ 9 scaling law by the quantized value that a kth expert obtains when the relative importance of relatively i and j; At Judgement Matricies A ^(k)if process in must meet same position expert and have a when judging between two respectively to the importance ranking of three indexs _i> a _jand a _j> a _k, then necessarily a is had _i> a _k, this ordinal consistency is then necessary observant criterion, otherwise relation between index can be caused confusing; That is: λ _maxfor judgment matrix A ^(k)maximum characteristic root, R.I. refers to Aver-age Random Consistency Index, and C.I. is matrix A ^(k)consistency check index; The A of structure ^(k)for reciprocal matrix, therefore construct antisymmetric matrix B ^(k): if there is Matrix C ^(k), make overall standard variances sigma _ijminimum, then claim C ^(k)for B ^(k)optimum transfer matrix, its element account form is $c_{ij}^{\left(k\right)}=\frac{1}{n}\underset{l=1}{\overset{n}{\Σ}}(b_{il}^{\left(k\right)}-b_{jl}^{\left(k\right)})=\lg \[\left(\sqrt[n]{\underset{l=1}{\overset{n}{\Σ}}{a}_{il}}\right)/\sqrt[n]{\underset{l=1}{\overset{n}{\Σ}}{a}_{jl}}\];$ Thus, resolving the weighted value obtaining a jth evaluation index is: due to C ^(k)naturally namely the consistency principle is met, so do not do consistency check; Be f by M index to factor in last layer factor layer in described ground floor _rmode of Level Simple Sequence weight vectors can be e _1r, e _2r..., e _mr(r=1,2 ..., R), and the second layer is R factor F={ f ₁, f ₂, f ₃total orderweight vector to destination layer, can f be written as ₁, f ₂..., f _r, can obtain bottom indicator layer i-th index thus for the weights of target is namely the weight of this factor is multiplied by by the weights sum of indexs all under certain factor;

Step 1.2, definition relative inferiority degree, namely the true measured value of a certain evaluation index and the relativeness between initial value and demand value, carry out dynamic calibration to weight, and the subjectivity remaining expert opinion also reflects the truth of GIS device state of insulation in real time by the objective assignment method of changeable weight; Definition X _rm0and X _rmabe as the criterion respectively and survey factor layer f _runder a certain evaluation index to dispatch from the factory the attention limit value of initial value and permission, find the attention limit value of dispatch from the factory initial value and the permission of the index of described indicator layer, utilize $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rma}\\ \frac{X_{rm0}-X_{rm}}{X_{rm0}-X_{rma}}& X_{rma}\&le;X_{rm}\&le;X_{rm0}\\ 0& X_{rm}\&GreaterEqual;X_{rm0}\end{array}\right.$ With $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rm0}\\ \frac{X_{rm}-X_{rm0}}{X_{rma}-X_{rm0}}& X_{rm0}\&le;X_{rm}\&le;X_{rma}\\ 0& X_{rm}\&GreaterEqual;X_{rma}\end{array}\right.$ Adjustment, in order to ensure the economy that power equipment runs, generally can allow near relevant code allowable value, have certain floating, represent with coefficient k, then for above-mentioned evaluation index, the limit value that code allows is multiplied by this coefficient and can be approximately X _rmaactual value, obviously for more large more Severe, k > 1, more little more Severe then gets k < 1; To impairment grade normalization, two formulas indicate the more little more Severe of index and the more large more Severe of index respectively;

Step 1.3, along with the change of running status, evaluation index may have the trend of deterioration, so that the degree in Decision Making Effect changes, and the method can combining the dynamic and stalic state weight determines final comprehensive weight $w_{rm,i},w_{rm,i}=\frac{w_{rm,i}^0{(1-x_{rm,i})}^{\mathrm{\&delta;}-1}}{\underset{k=1}{\overset{n}{\&Sigma;}}{w}_{rm,k}^0{(1-x_{rm,k})}^{\mathrm{\&delta;}-1}},$ Wherein, for the static weight under criterion factor layer, x _{rm, i}be then deteriorated weight, δ ∈ [0,1], for becoming weight parameter, mainly causes the intensity of variation of weight according to this index degradation in actual conditions; Obviously, when not having impairment grade to produce, δ=0, maintains original static weight constant; When δ=1, original static weight complete failure is described, now get one close to 0 small numerical approximation comprehensive weight now;

One calculates sub-step based on each index degree of membership of fuzzy theory GIS state of insulation:

For GIS device state of insulation H ₁, H ₂, H ₃, H ₄the fuzzy problem that may exist, with the membership function S (w based on fuzzy theory _m,i)={ μ _{i, H1}, μ _{i, H2}, μ _{i, H3}, μ _{i, H4}judge the H that assessment models result of calculation belongs to described ₁, H ₂, H ₃, H ₄which state, wherein μ _{i, Hm}for index i is under the jurisdiction of state H _mdegree of membership; For a certain evaluation index of described indicator layer, its possible degree adhering to described four kinds of state of insulations separately calculates with four functions:

Function one: described H is made in the distribution of type ridge less than normal shape ₄subordinate function:

${\mathrm{\&mu;}}_4\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_1\\ \frac{1}{2}-\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{ri}\&le;a_2\\ 0& x_{ri}>a_2\end{array}\right.,$

Function two: described H is made in the distribution of osculant ridge shape ₃subordinate function:

${\mathrm{\&mu;}}_3\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_1\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{ri}\&le;a_2\\ 1& a_2<x_{ri}\&le;a_3\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_3<x_{ri}\&le;a_4\\ 0& x_i>a_4\end{array}\right.,$

Function three: the distribution of osculant ridge shape is then respectively as the H described in work ₂subordinate function:

${\mathrm{\&mu;}}_2\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_3\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_3+a_4}{2})& a_3<x_{ri}\&le;a_4\\ 1& a_4<x_{ri}\&le;a_5\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_4+a_3}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right.,$

Function four: the H described in osculant ridge shape distribution work ₁subordinate function:

${\mathrm{\&mu;}}_1\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_5\\ \frac{1}{2}+\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right.;$

In four described subordinate functions, a _i(i=1,2 ..., 6) be the fuzzy separation of four kinds of state grades, according to the known a of definition ₁=0.1, a ₂=0.3, a ₃=0.4, a ₄=0.6, a ₅=0.7, a ₆=0.9, and x _ritwo class adjacent states may be belonged at most simultaneously;

The second layer factor layer f determined by described ground floor indicator layer _rits degree of membership can be expressed as P _r(H): $P_r\left(H\right)=\left(\begin{array}{cccc}{\mathrm{\&mu;}}_1\left(x_{r1}\right)& {\mathrm{\&mu;}}_2\left(x_{r1}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{r1}\right)\\ {\mathrm{\&mu;}}_1\left(x_{r2}\right)& {\mathrm{\&mu;}}_1\left(x_{r2}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{r2}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ {\mathrm{\&mu;}}_1\left(x_{rM}\right)& {\mathrm{\&mu;}}_2\left(x_{rM}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{rM}\right)\end{array}\right),$ In conjunction with described in comprehensive weight formula the comprehensive weight value of each state estimation index calculated, can obtain the original confidence level M obtained after r combined factors fuzzy evaluation in total evaluation system thus _r(H): $M_r\left(H\right)=\underset{m=1}{\overset{M}{\&Sigma;}}{w}_{rm}{P}_r\left(H\right);$

The sub-step of a GIS Condition assessment of insulation model merged based on DS evidential reasoning:

By factor layer { f ₁, f ₂, f ₃be defined as three evidence sources, before the fusion to three evidence sources, introduce degree of confidence factor alpha _r(r=1,2,3) characterize different evidence source to the weight of final decision result, the original confidence level M described in supposing _r(H) be factor layer factor f _roriginal basic probability assignment, can r after weight modification new m function m _r(H) be: m _r(H)=α _rm _r(H), and m _r(θ) for revising the uncertain evidence probability assignments after weight; Degree of confidence factor alpha herein _rby revising original weight value α _kcalculate, namely w _maxfor f ₁, f ₂, f ₃maximal value in weight;

There is r kind in definition evidence source, then take this rule of combination to all m function fusion evidence fusion formulas under n m function of the described indicator layer belonging to each evidence source: $m\left(A\right)=(m_1\&CirclePlus;m_2\&CirclePlus;...\&CirclePlus;m_n)\left(A\right)=\frac{1}{K}\underset{A_1\&cap;A_2\&cap;...A_n=A}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right),$ With $K=\underset{A_1\&cap;A_2\&cap;...A_n\&NotEqual;\mathrm{\&phi;}}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right)$ Carry out fusion to solve, described evidence fusion formula is the strict AND operation to entire evidence source, gets m _r(H) state evaluation belonging to the maximal value in is final GIS state of insulation grade.

Tool of the present invention has the following advantages: 1. the invention provides a kind of gas insulated combined electrical equipment state of insulation comprehensive estimation method, compensate for existing SF ₆gas insulated electric apparatus lacks the appraisal procedure considering the state of insulation of comprehensive factor, is SF ₆fault diagnosis and the maintenance of gas insulated electric apparatus lay the foundation.2. of the present invention in the state index plant process of the equipment of structure, to fully take into account electrically, the comprehensive factor of preventive trial of chemistry and equipment, improve accuracy and the confidence level of the GIS device Condition assessment of insulation method of foundation.3. the present invention builds in the weight design process of GIS device Condition assessment of insulation index, in conjunction with static weight and changeable weight, take into full account the complicated variety of GIS, and the impact of trend of deterioration may be had along with the change evaluation index of running status, guarantee the accuracy of evaluation scheme final decision.4. the present invention builds in the process of the membership function of GIS device four kinds of state of insulations, and the degree of membership that introducing fuzzy theory carries out evaluation index and state grade judges, ensure that the accuracy of state of insulation level evaluation.5. the present invention is in the last evaluation decision process of GIS state of insulation, and by evidence theory fusion multi source status evaluation index, it is the practical again method of economy that the objective distribution realizing original probability obtains the final result of decision accurately.6. the present invention relates to evaluation index, assessment models comprehensively reliable, assessment result is simple and clear, can in Practical Project field popularization application.The present invention considers limit due to GIS self-operating condition, the various methods making PD detect its true state of insulation have its limitation, but the complementary information under different monitoring condition can be utilized to build comprehensive index system of reaction GIS device state of insulation, and in conjunction with expertise, utilize fuzzy theory process different levels, between single level index and state of insulation grade, degree of membership calculates, thus realize the comprehensive assessment of GIS device state of insulation.

Accompanying drawing explanation

Fig. 1 is apparatus insulated State Assessment Index System block diagram of the present invention.

Fig. 2 is that each index of index system of the present invention illustrates schematic diagram.

Fig. 3 is the UHF shelf depreciation diagnostic device DMS data of certain 110kVGIS monitoring of equipment.

Fig. 4 is that the basic parameter of non-phase non-temporal pattern obtains schematic diagram (1. discharge pulse amplitude; 2. the time interval of two continuous discharges; 3. discharging gap).

Fig. 5 is the PD chemical parameters that certain 110kVGIS equipment off-line obtains.

Fig. 6 is that GIS device Condition assessment of insulation model factor layer and indicator layer build schematic diagram.

Fig. 7 is typical case 1 ~ 9 scaling law value criterion used.

Fig. 8 is Aver-age Random Consistency Index value result.

Fig. 9 is Static State Index weight calculation result.

Figure 10 is that the membership function of four kinds of state of insulation grades builds schematic diagram.

Figure 11 is the degree of membership value schematic diagram of evaluation index.

The GIS device Condition assessment of insulation model overall schematic that Figure 12 merges based on fuzzy evidence reasoning.

Figure 13 is the result of calculation schematic diagram in three class evidence sources.

Figure 14 is the result schematic diagram that evidence source fusion reasoning calculates.

Embodiment

Below in conjunction with embodiment, further illustrate the present invention.

First need the real data based on Site Detection, complete the building process of GIS Condition assessment of insulation index system, for described gas insulated combined electrical equipment state of insulation comprehensive estimation method provides appreciation information basis.As shown in Figure 1, described index system implication as shown in Figure 2, carries out SF to certain 110kVGIS equipment to GIS Condition assessment of insulation index system establishment scheme ₆the concrete steps of gas insulated combined electrical equipment Condition assessment of insulation index are as follows:

First, PD electric parameter f ₁={ e ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉state of insulation index: 1. UHFPD data acquisition, PD electric parameter mainly comes from pulse current method and ultra-high-frequency detection method (UltraHighFrequency, UHF) method, gathers PD Monitoring Data by UHF method, i.e. a large amount of time-domain sampling values of the ultra-high frequency signal q of PD.Directly utilize UHFPD signal that UHF shelf depreciation diagnostic device DMS gathers described 110kVGIS equipment #1 interval as shown in Figure 3.With the basic parameter that the change (Δ q/ Δ t) in the UHF signal q amplitude unit interval is PD signal analysis, and then choose three basic parameters of shelf depreciation electric parameter: the time interval (Δ t) of discharge pulse amplitude (p), continuous two discharge pulses and discharging gap (Δ T), as illustration in Fig. 4, extract required basic parameter data.With the reference signal that power frequency sinusoidal ac signal f=Asin (314t) is PD monitor signal data q, Δ T is decided to be 2 power frequency periods and is 0.04s, by monitor signal data-signal q's $(0~\frac{\mathrm{\&Delta;}T}{2},...,i\mathrm{\&Delta;}T~\frac{(2i+1)\mathrm{\&Delta;}T}{2},...,999\mathrm{\&Delta;}T~\frac{1999\mathrm{\&Delta;}T}{2},i=1,2,...)$ Interior obtained data with $(\frac{\mathrm{\&Delta;}T}{2}~\mathrm{\&Delta;}T,...,\frac{(2i-1)\mathrm{\&Delta;}T}{2}~i\mathrm{\&Delta;}T,...,\frac{1999\mathrm{\&Delta;}T}{2}~2000\mathrm{\&Delta;}T,i=1,2,...)$ Interior data are extracted and are arranged as positive half period signal with negative half-cycle signal meanwhile, positive half period signal is got with negative half-cycle signal interior amplitude is greater than 8mV statistics and is decided to be discharge gage once, and the time interval directly subtracts each other with twice adjacent discharging time and obtains, and then the time interval also forms sequence respectively by the front/rear cycle with the data of whole 2000 power frequency periods collected form sequence Δ T ₁..., Δ T _w..., Δ T ₁₀₀₀.2. electric state index extraction, electric state index extraction process is: the first, is extracted pulse amplitude correlation behavior parameter e ₁₁, e ₁₂, e ₁₃.Wherein, e ₁₁: N _mag, for the electric discharge in front/rear cycle the total degree of superposition reflects front/rear periodic discharging repetition rate; e ₁₂: E _mag, wherein for the half wave amplitude average in front/rear cycle reflects strength of discharge, by obtain; e ₁₃: S _mag, amplitude standard deviation, wherein the half wave amplitude standard deviation for the front/rear cycle reflects the change of front/rear periodic discharging pulse amplitude, by $S_{mag}^f=\sqrt{\frac{\underset{k}{\&Sigma;}(p_k^f/E_{mag}^f-1^2)}{N_{mag}^f}}$ With $S_{mag}^b=\sqrt{\frac{\underset{k}{\&Sigma;}{(p_k^b/E_{mag}^b-1)}^2}{N_{mag}^b}}$ Obtain.The second, be extracted the time interval relevant state parameter e ₁₄, e ₁₅, e ₁₆.Wherein, e ₁₄: E _int, in for the maximal value in the average of front/rear time-of-week interval, describe the time interval of double electric discharge, by ${E_{\mathrm{ti}}}^f=\frac{\underset{l}{\mathrm{\&Sigma;}}\mathrm{\&Delta;}{t}_l^f}{{N_{\mathrm{ti}}}^f}$ With $E_{\mathrm{mag}}^b=\frac{\underset{l}{\mathrm{\&Sigma;}}{\mathrm{\&Delta;t}}_l^b}{N_{\mathrm{int}}^b}$ Calculate; e ₁₅: N _int, $N_{\mathrm{int}}=N_{\mathrm{int}}^f+N_{\mathrm{int}}^b$ In for the time interval superposition sum in front/rear cycle, describe discharge pulse sequence time duration; e ₁₆: S _int, in get the time interval standard deviation maximal value in front/rear cycle, describes the change in the time interval, by with obtain; 3rd, be extracted the state parameter e that discharge range is relevant ₁₇, e ₁₈, e ₁₉.Wherein, e ₁₇: Δ T _min, characterize electric discharge interval feature; e ₁₈: P, characterize the polar character of PD; e ₁₉: N _ref, characterize the polar character of PD, power frequency period be on average divided into 20 sub-ranges, be more than or equal in each interval 3 times and above electric discharge as being whether the decision criteria discharged, thus the total discharge range N to the front/rear cycle _refadd up.

Secondly, PD chemical parameters f ₂={ e ₂₁, e ₂₂, e ₂₃, e ₂₄, e ₂₅, e ₂₆state of insulation index.1. gas composition data acquisition: the gas of the 110kVGIS equipment #1 compartment described in collection, carries out with gas chromatograph the SF that gas composition measurement obtains ₆the characteristic gas four kinds of typical CF decomposed ₄, CO ₂, SO ₂f ₂, SOF ₂content.2. chemical state index extraction: chemical state index extraction process is: the first, using all sulfur compound sums as SF ₆sulfur-bearing analyte decomposition total amount state parameter e is extracted in the embodiment that gas molecule decomposes total amount ₂₁: c (SOF ₂)+c (SO ₂f ₂).The second, extract the content ratio state parameter e of gas composition ₂₂, e ₂₃, e ₂₄, wherein state parameter e ₂₂for c (SOF ₂)/c (SO ₂f ₂), e ₂₃for c (CF ₄)/c (CO ₂), e ₂₄for c (SOF ₂+ SO ₂f ₂)/c (CF ₄+ CO ₂).3rd, by the characteristic component content after every 24h sampling be CO with the ratio of time interval Δ t ₂with SO ₂f ₂respective components is at the gas production rate (10 in jth sky ^-6/ day), measure herein and select to carry out 4 days continuously, substitute into measurement data formula ${\mathrm{\&gamma;}}^{\&prime;}\left({\mathrm{CO}}_2\right)=\frac{1}{2}\sqrt{\underset{j=1}{\overset{4}{\&Sigma;}}{\left(\frac{{\mathrm{\&Delta;c}}_{\left({\mathrm{CO}}_2\right)}}{\mathrm{\&Delta;}t}\right)}_j^2}$ With ${\mathrm{\&gamma;}}^{\&prime;}\left({\mathrm{SO}}_2{F}_2\right)=\frac{1}{2}\sqrt{\underset{j=1}{\overset{4}{\&Sigma;}}{\left(\frac{{\mathrm{\&Delta;c}}_{\left({\mathrm{SO}}_2{F}_2\right)}}{\mathrm{\&Delta;}t}\right)}_j^2}$ Calculate γ ' (SO ₂f ₂), γ ' (CO ₂) i.e. state parameter e ₂₅, e ₂₆, obtain data as shown in Figure 5.

Finally, the characteristic parameter f of preventive trial ₃={ e ₃₁, e ₃₂, e ₃₃state of insulation index, based on the preventive trial of described 110kVGIS equipment, the gas of the 110kVGIS equipment #1 compartment described in collection, the moisture e recorded with microwater device ₃₁, test result display arc extinguishing air chamber c (H ₂o)=163ppm.The SF in described 110kVGIS equipment #1 compartment is recorded with air-leakage detector ₆leakage Gas e ₃₂be 0.48 μ L/L.E ₃₃the gas primary circuit insulation against ground resistance measured value of the 110kVGIS equipment #1 compartment described in surveying with Insulation Resistance Tester is about 7500M Ω.

Embodiment 1.

A kind of gas insulated combined electrical equipment state of insulation comprehensive estimation method, " the gas insulated combined electrical equipment state of insulation comprehensive estimation method, " that utilize the applicant to apply for, the concrete steps of GIS being carried out to Condition assessment of insulation are as follows:

(1) GIS device Condition assessment of insulation System Framework builds

Apparatus insulated for described 110kVGIS state is formed Recurison order hierarchy according to the source attribute difference of three class status informations, is followed successively by the evaluation index layer evaluated layer, factor layer and each factor layer and select from top to bottom.Wherein, evaluate layer and be decided to be H according to " power transmission and transformation equipment state overhauling directive/guide " and GIS device operating experience ₁, H ₂, H ₃, H ₄one of four states: 1. normal condition (H ₁level): every quantity of state of equipment is in normal table.2. attention state (H ₂level): patrol and examine some quantity of state of middle discovery and change and have the possible trend close to warning value.3. abnormality (H ₃level): a certain or various states amount is only slight beyond allowable value, and reasonable arrangement Strategies of Maintenance does suitable process to abnormality.4. severe conditions (H ₄level): monitor signal has strong exception, and quantity of state is seriously above standard limit value, should arrange interruption maintenance even more exchange device as early as possible.Factor layer then chooses PD electric parameter, PD chemical parameters, GIS device preventive trial parameter are defined as F={f ₁, f ₂, f ₃, namely factor classification is returned to the institute of considered index.Indicator layer chooses the PD electric parameter f of characterization device state of insulation ₁={ e ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, PD chemical parameters f ₂={ e ₂₁, e ₂₂, e ₂₃, e ₂₄, e ₂₅, e ₂₆and the characteristic parameter f of GIS device preventive trial ₃={ e ₃₁, e ₃₂, e ₃₃the state of insulation index of state of insulation index.

(2) GIS insulation evaluation index weight calculation

By e ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, e ₂₁, e ₂₂, e ₂₃, e ₂₄, e ₂₅, e ₂₆, e ₃₁, e ₃₂, e ₃₃indicator layer is defined as ground floor, and as the Information base that evaluating system is the most basic, it is selected and sequence plays vital effect to the result of decision, and the significance level to each index of this step qualitative and quantitative carries out weight evaluation.F={f ₁, f ₂, f ₃factor layer is defined as the second layer, level is formed as shown in Figure 6, and its a certain factor A comprehensively determines by n evaluation index, and m expert can compare any two indices according to experience knowledge, the judgment matrix A that a kth expert obtains ^(k)for: wherein represent the quantized value that a kth expert obtains when the relative importance of relatively i and j, with modal 1 ~ 9 scaling law as the quantification of expert opinion is carried out in division in Fig. 7.At Judgement Matricies A ^(k)if process in must meet same position expert and have a when judging between two respectively to the importance ranking of three indexs _i> a _jand a _j> a _k, then necessarily a is had _i> a _k, this ordinal consistency is then necessary observant criterion, otherwise relation between index can be caused confusing.That is: λ _maxfor judgment matrix A ^(k)maximum characteristic root, R.I. refer to Aver-age Random Consistency Index directly according to inspection information definition tables of data substitute into calculating as shown in Figure 8, C.I. is matrix A ^(k)consistency check index.Then matrix theory principle is utilized, the A of structure ^(k)for reciprocal matrix, B ^(k): $B^{\left(k\right)}={\left(b_{\mathrm{ij}}^{\left(k\right)}\right)}_{m\&times;n}={1\mathrm{gA}}^{\left(k\right)}.$ And structural matrix C ^(k), utilize ${\mathrm{\&sigma;}}_{\mathrm{ij}}^2=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(c_{\mathrm{ij}}^{\left(k\right)}-b_{\mathrm{ij}}^{\left(k\right)})}^2$ Overall standard variances sigma _ijminimum mathematic condition, the C calculated ^(k)for B ^(k)optimum transfer matrix, use $c_{ij}^{\left(k\right)}=\frac{1}{n}\underset{l=1}{\overset{n}{\&Sigma;}}(b_{il}^{\left(k\right)}-b_{jl}^{\left(k\right)})=\lg \&lsqb;\left(\sqrt[n]{\underset{l=1}{\overset{n}{\&Sigma;}}{a}_{il}}\right)/\sqrt[n]{\underset{l=1}{\overset{n}{\&Sigma;}}{a}_{jl}}\&rsqb;$ Each element is calculated.Thus, the weighted value of a jth evaluation index is used calculate, due to C ^(k)naturally namely the consistency principle is met, so do not do consistency check.Be f by M index to factor in last layer factor layer in described ground floor _rmode of Level Simple Sequence weight vectors can be e _1r, e _2r..., e _mr(r=1,2 ..., R), and the second layer is the total orderweight vector of R factor to destination layer, can be written as f ₁, f ₂..., f _r, can obtain bottom indicator layer i-th index thus for the weights of target is: the result calculated each indicator layer recorded and factor layer one by one as shown in Figure 9.

Find the attention limit value data of dispatch from the factory initial value and the permission of described 110kVGIS equipment, definition X _rm0and X _rmabe as the criterion respectively and survey factor layer f _runder a certain evaluation index to dispatch from the factory the attention limit value of initial value and permission, utilize $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rma}\\ \frac{X_{rm0}-X_{rm}}{X_{rm0}-X_{rma}}& X_{rma}\&le;X_{rm}\&le;X_{rm0}\\ 0& X_{rm}\&GreaterEqual;X_{rm0}\end{array}\right.$ With $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rm0}\\ \frac{X_{rm}-X_{rm0}}{X_{rma}-X_{rm0}}& X_{rm0}\&le;X_{rm}\&le;X_{rma}\\ 0& X_{rm}\&GreaterEqual;X_{rma}\end{array}\right.,$ And near allowable value, have certain float factor k to represent, get the more large more Severe of k < 1 and index according to the more little more Severe of index and get k > 1, easy in order to calculate, float factor directly gets 1 herein, namely not to x _rmcarry out amendment of floating.

Finally, the method for combining the dynamic and stalic state weight determines final comprehensive weight w _{rm, i}, use formula calculate, wherein, for the static weight under criterion factor layer, x _{rm, i}be then deteriorated weight, δ ∈ [0,1], for becoming weight parameter, directly gets δ=0.001 and makes comprehensive weight be similar to static weight.

(3) based on the GIS Condition assessment of insulation model of Fuzzy Evidence Theory

Based on Fuzzy Evidence Theory GIS Condition assessment of insulation model construction thinking as shown in figure 12, membership function S (w is introduced for the apparatus insulated state of described 110kVGIS _m,i)={ μ _{i, H1}, μ _{i, H2}, μ _{i, H3}, μ _{i, H4}judge the H that assessment models result of calculation belongs to described ₁, H ₂, H ₃, H ₄which state, wherein μ _{i, Hm}for index i is under the jurisdiction of state H _mdegree of membership.Membership function as shown in Figure 10, is respectively: described H is made in the distribution of type ridge less than normal shape ₄subordinate function: ${\mathrm{\&mu;}}_4\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_1\\ \frac{1}{2}-\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{ri}\&le;a_2\\ 0& x_{ri}>a_2\end{array}\right.,$ Described H is made in the shape distribution of osculant ridge ₃subordinate function: ${\mathrm{\&mu;}}_3\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_1\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{ri}\&le;a_2\\ 1& a_2<x_{ri}\&le;a_3\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_3<x_{ri}\&le;a_4\\ 0& x_i>a_4\end{array}\right.,$ The shape distribution of osculant ridge is then respectively as the H described in work ₂subordinate function: ${\mathrm{\&mu;}}_2\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_3\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_3+a_4}{2})& a_3<x_{ri}\&le;a_4\\ 1& a_4<x_{ri}\&le;a_5\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_4+a_3}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right.$ H described in doing with the distribution of osculant ridge shape ₁subordinate function: ${\mathrm{\&mu;}}_1\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_5\\ \frac{1}{2}+\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right..$ In four described subordinate functions, a _i(i=1,2 ..., 6) be the fuzzy separation of four kinds of state grades, substitute into a ₁=0.1, a ₂=0.3, a ₃=0.4, a ₄=0.6, a ₅=0.7, a ₆=0.9, each index number of indicator layer is substituted in formula carry out calculating evaluation index degree of membership value as shown in figure 11 one by one.

The second layer factor layer f determined by described ground floor indicator layer _rits degree of membership can be expressed as P _r(H): $P_r\left(H\right)=\left(\begin{array}{cccc}{\mathrm{\&mu;}}_1\left(x_{r1}\right)& {\mathrm{\&mu;}}_2\left(x_{r1}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{r1}\right)\\ {\mathrm{\&mu;}}_1\left(x_{r2}\right)& {\mathrm{\&mu;}}_1\left(x_{r2}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{r2}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ {\mathrm{\&mu;}}_1\left(x_{rM}\right)& {\mathrm{\&mu;}}_2\left(x_{rM}\right)& ...& {\mathrm{\&mu;}}_4\left(x_{rM}\right)\end{array}\right),$ In conjunction with described in comprehensive weight formula the comprehensive weight value of each state estimation index calculated, substitutes into original confidence level M _r(H): in formula, can obtain after calculating obtaining after r combined factors fuzzy evaluation in total evaluation system

$M_r\left(H\right)=\left(\begin{array}{cccc}0.0337& 0.2523& 0.4580& 0.2560\\ 0.0630& 0.3849& 0.4090& 0.1431\\ 0.1061& 0.1537& 0.4337& 0.3065\end{array}\right).$

(4) based on the GIS Condition assessment of insulation model that DS evidential reasoning merges

By factor layer { f ₁, f ₂, f ₃be defined as three evidence sources, before the fusion to three evidence sources, get preferential degree of confidence factor alpha _k=0.9, substitute into formula (w _maxget f ₁, f ₂, f ₃maximal value in weight) degree of confidence factor alpha can be obtained _r=(0.9,0.8350,0.7145), by α _rthe m described in value combination of=(0.9,0.8350,0.7145) _r(H) computing formula m _r(H)=α _rm _r(H) and m _r(θ) for revising the uncertain evidence probability assignments after weight.And then the original probability that can obtain separately corresponding to uncertain judgement distributes m _r(θ)=(0.1,0.1650,0.2855), so by m _r(θ)=(0.1,0.1650,0.2855) with $M_r\left(H\right)=\left(\begin{array}{cccc}0.0337& 0.2523& 0.4580& 0.2560\\ 0.0630& 0.3849& 0.4090& 0.1431\\ 0.1061& 0.1537& 0.4337& 0.3065\end{array}\right)$ Basic probability assignment table such as the Figure 13 forming three class evidence sources shows.

There are 3 kinds of f in evidence source ₁, f ₂, f ₃, then the data under n m function of the described indicator layer belonging to each evidence source in result of calculation and Figure 13, substitute into this rule of combination to all m function fusion evidence fusion formulas: $m\left(A\right)=(m_1\&CirclePlus;m_2\&CirclePlus;...\&CirclePlus;m_n)\left(A\right)=\frac{1}{K}\underset{A_1\&cap;A_2\&cap;...A_n=A}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right),$ With $K=\underset{A_1\&cap;A_2\&cap;...A_n\&NotEqual;\mathrm{\&phi;}}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right)$ Carry out fusion to solve, get m _r(H) state evaluation belonging to the maximal value in is final GIS state of insulation grade, by f ₁, f ₂with f ₁, f ₂, f ₃carry out fusion calculation result as shown in figure 14.According to method same above, respectively respectively Condition assessment of insulation is carried out to #2 ~ #8 with other 7 groups of GIS interval field datas of the 110kVGIS equipment described in collecting, realize 6 groups in 8 groups of samples and correctly assess, 1 group of estimation error, fail to provide the result of decision for 1 group, correct assessment reaches 75%.

Specific embodiment described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Claims (1)

1. a gas insulated combined electrical equipment state of insulation comprehensive estimation method, the method is characterized in that, specifically comprises:

The sub-step that a GIS Condition assessment of insulation System Framework builds: GIS device state of insulation is formed Recurison order hierarchy according to the source attribute difference of three class status informations, is followed successively by the evaluation index layer evaluated layer, factor layer and each factor layer and select from top to bottom; Wherein,

Layer one: evaluate layer and be decided to be H ₁, H ₂, H ₃, H ₄one of four states:

State 1., normal condition H ₁level: every quantity of state of equipment is in normal table;

State 2., attention state H ₂level: patrol and examine some quantity of state of middle discovery and change and have the possible trend close to warning value;

State 3., abnormality H ₃level: a certain or various states amount is only slight beyond allowable value, and reasonable arrangement Strategies of Maintenance does suitable process to abnormality;

State 4., severe conditions H ₄level: monitor signal has strong exception, quantity of state is seriously above standard limit value, should arrange interruption maintenance even more exchange device as early as possible;

Layer two: factor layer then chooses PD electric parameter, PD chemical parameters, GIS device preventive trial parameter are defined as F={f ₁, f ₂, f ₃, namely to the factor classification of returning of considered index;

Layer three: indicator layer chooses the PD electric parameter f of characterization device state of insulation ₁={ e ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, PD chemical parameters and the characteristic parameter f of GIS device preventive trial ₃={ e ₃₁, e ₃₂, e ₃₃the state of insulation indicator layer of state of insulation index;

The sub-step of a GIS insulation evaluation index weight calculation:

Step 1.1,

E ₁₁, e ₁₂, e ₁₃, e ₁₄, e ₁₅, e ₁₆, e ₁₇, e ₁₈, e ₁₉, e ₂₁, e ₂₂, e ₂₃, e ₂₄, e ₂₅, e ₂₆, e ₃₁, e ₃₂, e ₃₃for indicator layer is defined as ground floor, as the Information base that evaluating system is the most basic, it is selected and sequence plays vital effect to the result of decision, and the significance level to each index of this step qualitative and quantitative carries out weight evaluation;

F={f ₁, f ₂, f ₃factor layer is defined as the second layer, its a certain factor A comprehensively determines by n evaluation index, and m expert can compare any two indices according to experience knowledge, the judgment matrix A that a kth expert obtains ^(k)for: wherein represent and carry out the quantification of expert opinion with modal 1 ~ 9 scaling law by the quantized value that a kth expert obtains when the relative importance of relatively i and j; At Judgement Matricies A ^(k)if process in must meet same position expert and have a when judging between two respectively to the importance ranking of three indexs _i> a _jand a _j> a _k, then necessarily a is had _i> a _k, this ordinal consistency is then necessary observant criterion, otherwise relation between index can be caused confusing; That is: λ _maxfor judgment matrix A ^(k)maximum characteristic root, R.I. refers to Aver-age Random Consistency Index, and C.I. is matrix A ^(k)consistency check index; The A of structure ^(k)for reciprocal matrix, therefore construct antisymmetric matrix B ^(k): if there is Matrix C ^(k), make overall standard variances sigma _ijminimum, then claim C ^(k)for B ^(k)optimum transfer matrix, its element account form is $c_{ij}^{\left(k\right)}=\frac{1}{n}\underset{l=1}{\overset{n}{\&Sigma;}}(b_{il}^{\left(k\right)}-b_{jl}^{\left(k\right)})=\lg \&lsqb;\left(\sqrt[n]{\underset{l=1}{\overset{n}{\&Pi;}}{a}_{il}}\right)/\sqrt[n]{\underset{l=1}{\overset{n}{\&Pi;}}{a}_{jl}}\&rsqb;;$ Thus, resolving the weighted value obtaining a jth evaluation index is: $w_j=1/\underset{i=1}{\overset{n}{\&Sigma;}}{10}^{c_{ij}},j=1,2,...,n,$ Due to C ^(k)naturally namely the consistency principle is met, so do not do consistency check; Be f by M index to factor in last layer factor layer in described ground floor _rmode of Level Simple Sequence weight vectors can be e _1r, e _2r..., e _mr(r=1,2 ..., R), and the second layer is R factor F={ f ₁, f ₂, f ₃total orderweight vector to destination layer, can f be written as ₁, f ₂..., f _r, can obtain bottom indicator layer i-th index thus for the weights of target is namely the weight of this factor is multiplied by by the weights sum of indexs all under certain factor;

Step 1.2, definition relative inferiority degree, namely the true measured value of a certain evaluation index and the relativeness between initial value and demand value, carry out dynamic calibration to weight, and the subjectivity remaining expert opinion also reflects the truth of GIS device state of insulation in real time by the objective assignment method of changeable weight; Definition X _rm0and X _rmabe as the criterion respectively and survey factor layer f _runder a certain evaluation index to dispatch from the factory the attention limit value of initial value and permission, find the attention limit value of dispatch from the factory initial value and the permission of the index of described indicator layer, utilize $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rma}\\ \frac{X_{rm0}-X_{rm}}{X_{rm0}-X_{rma}}& X_{rma}\&le;X_{rm}\&le;X_{rm0}\\ 0& X_{rm}\&GreaterEqual;X_{rm0}\end{array}\right.$ With $x_{rm}=\left\{\begin{array}{cc}1& X_{rm}\&le;X_{rm0}\\ \frac{X_{rm}-X_{rm0}}{X_{rma}-X_{rm0}}& X_{rm0}\&le;X_{rm}\&le;X_{rma}\\ 0& X_{rm}\&GreaterEqual;X_{rma}\end{array}\right.$ Adjustment, in order to ensure the economy that power equipment runs, generally can allow near relevant code allowable value, have certain floating, represent with coefficient k, then for above-mentioned evaluation index, the limit value that code allows is multiplied by this coefficient and can be approximately X _rmaactual value, obviously for more large more Severe, k > 1, more little more Severe then gets k < 1; To impairment grade normalization, two formulas indicate the more little more Severe of index and the more large more Severe of index respectively;

Step 1.3, along with the change of running status, evaluation index may have the trend of deterioration, so that the degree in Decision Making Effect changes, and the method can combining the dynamic and stalic state weight determines final comprehensive weight w _{rm, i}, $w_{rm,i}=\frac{w_{rm,i}^0{(1-x_{rm,i})}^{\mathrm{\&delta;}-1}}{\underset{k=1}{\overset{n}{\&Sigma;}}{w}_{rm,k}^0{(1-x_{rm,k})}^{\mathrm{\&delta;}-1}},$ Wherein, for the static weight under criterion factor layer, x _{rm, i}be then deteriorated weight, δ ∈ [0,1], for becoming weight parameter, mainly causes the intensity of variation of weight according to this index degradation in actual conditions; Obviously, when not having impairment grade to produce, δ=0, maintains original static weight constant; When δ=1, original static weight complete failure is described, now get one close to 0 small numerical approximation comprehensive weight now;

One calculates sub-step based on each index degree of membership of fuzzy theory GIS state of insulation:

For GIS device state of insulation H ₁, H ₂, H ₃, H ₄the fuzzy problem that may exist, with the membership function S (w based on fuzzy theory _m,i)={ μ _{i, H1}, μ _{i, H2}, μ _{i, H3}, μ _{i, H4}judge the H that assessment models result of calculation belongs to described ₁, H ₂, H ₃, H ₄which state, wherein μ _{i, Hm}for index i is under the jurisdiction of state H _mdegree of membership; For a certain evaluation index of described indicator layer, its possible degree adhering to described four kinds of state of insulations separately calculates with four functions:

Function one: described H is made in the distribution of type ridge less than normal shape ₄subordinate function:

${\mathrm{\&mu;}}_4\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_1\\ \frac{1}{2}-\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{\mathrm{ri}}\&le;a_2\\ 0& x_{ri}>a_2\end{array}\right.,$

Function two: described H is made in the distribution of osculant ridge shape ₃subordinate function:

${\mathrm{\&mu;}}_3\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_1\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_1<x_{ri}\&le;a_2\\ 1& a_2<x_{ri}\&le;a_3\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_3<x_{ri}\&le;a_4\\ 0& x_i>a_4\end{array}\right.,$

Function three: the distribution of osculant ridge shape is then respectively as the H described in work ₂subordinate function:

${\mathrm{\&mu;}}_2\left(x_{ri}\right)=\left\{\begin{array}{cc}0& x_{ri}\&le;a_3\\ \frac{1}{2}+\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_3+a_4}{2})& a_3<x_{ri}\&le;a_4\\ 1& a_4<x_{ri}\&le;a_5\\ \frac{1}{2}-\frac{1}{2}\sin \frac{\mathrm{\&pi;}}{a_4-a_3}(x_{ri}-\frac{a_4+a_3}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right.,$

Function four: the H described in osculant ridge shape distribution work ₁subordinate function:

${\mathrm{\&mu;}}_1\left(x_{ri}\right)=\left\{\begin{array}{cc}1& x_{ri}\&le;a_5\\ \frac{1}{2}+\frac{1}{2}sin\frac{\mathrm{\&pi;}}{a_2-a_1}(x_{ri}-\frac{a_2+a_1}{2})& a_5<x_{ri}\&le;a_6\\ 0& x_{ri}>a_6\end{array}\right.;$

In four described subordinate functions, a _i(i=1,2 ..., 6) be the fuzzy separation of four kinds of state grades, according to the known a of definition ₁=0.1, a ₂=0.3, a ₃=0.4, a ₄=0.6, a ₅=0.7, a ₆=0.9, and x _ritwo class adjacent states may be belonged at most simultaneously;

The second layer factor layer f determined by described ground floor indicator layer _rits degree of membership can be expressed as P _r(H): $P_r\left(H\right)=\left(\begin{array}{cccc}{\mathrm{\&mu;}}_1\left(x_{r1}\right)& {\mathrm{\&mu;}}_2\left(x_{r1}\right)& \mathrm{...}& {\mathrm{\&mu;}}_4\left(x_{r1}\right)\\ {\mathrm{\&mu;}}_1\left(x_{r2}\right)& {\mathrm{\&mu;}}_1\left(x_{r2}\right)& \mathrm{...}& {\mathrm{\&mu;}}_4\left(x_{r2}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ {\mathrm{\&mu;}}_1\left(x_{rM}\right)& {\mathrm{\&mu;}}_2\left(x_{rM}\right)& \mathrm{...}& {\mathrm{\&mu;}}_4\left(x_{rM}\right)\end{array}\right),$ In conjunction with described in comprehensive weight formula the comprehensive weight value of each state estimation index calculated, can obtain the original confidence level M obtained after r combined factors fuzzy evaluation in total evaluation system thus _r(H): $M_r\left(H\right)=\underset{m=1}{\overset{M}{\&Sigma;}}{w}_{rm}{P}_r\left(H\right);$

The sub-step of a GIS Condition assessment of insulation model merged based on DS evidential reasoning:

By factor layer { f ₁, f ₂, f ₃be defined as three evidence sources, before the fusion to three evidence sources, introduce degree of confidence factor alpha _r(r=1,2,3) characterize different evidence source to the weight of final decision result, the original confidence level M described in supposing _r(H) be factor layer factor f _roriginal basic probability assignment, can r after weight modification new m function m _r(H) be: m _r(H)=α _rm _r(H), and m _r(θ) for revising the uncertain evidence probability assignments after weight; Degree of confidence factor alpha herein _rby revising original weight value α _kcalculate, namely w _maxfor f ₁, f ₂, f ₃maximal value in weight;

There is r kind in definition evidence source, then take this rule of combination to all m function fusion evidence fusion formulas under n m function of the described indicator layer belonging to each evidence source: $m\left(A\right)=(m_1\&CirclePlus;m_2\&CirclePlus;...\&CirclePlus;m_n)\left(A\right)=\frac{1}{K}\underset{A_1\&cap;A_2\&cap;\mathrm{...}{A}_n=A}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right),$ With $K=\underset{A_1\&cap;A_2\&cap;\mathrm{...}{A}_n\&NotEqual;\mathrm{\&phi;}}{\&Sigma;}{m}_1\left(A_1\right)\&CenterDot;m_2\left(A_2\right)...m_2\left(A_n\right)$ Carry out fusion to solve, described evidence fusion formula is the strict AND operation to entire evidence source, gets m _r(H) state evaluation belonging to the maximal value in is final GIS state of insulation grade.