CN104484722A - CIM standard based modeling method for model about power grid disasters influenced by meteorological factors - Google Patents

Abstract

A CIM standard based modeling method for a model about power grid disasters influenced by meteorological factors comprises steps as follows: an improved multiple linear regression algorithm based meteorological disaster factor model is constructed through combination of a multiple linear regression algorithm with the meteorology, that is, the meteorological factors are taken as independent variables, meteorological disasters are taken as dependent variables, correlation coefficients of different independent variables and dependent variables are acquired with the multiple linear regression algorithm, and corresponding multiple linear regression models are established; an original regression model is amended, that is, data in a certain period are acquired, and the original regression model is amended through recalculation of the correlation coefficients with the multiple linear regression algorithm. With the adoption of the model, prediction and analysis of the effects caused by the meteorological factors on the power grid disasters can be well conducted, and the meteorological disasters can be predicted actually, so that economic losses caused by the meteorological factor disasters in power supply and distribution processes can be decreased to the greatest extent.

Description

A kind of meteorological factor based on CIM standard affects electrical network hazard model modeling method

Technical field

The present invention relates to power information administrative skill, particularly relating to a kind of meteorological factor based on CIM standard affects electrical network hazard model modeling method.

Background technology

In the research of " meteorologic factor is to the model of electric power safety accident impact ", for south China somewhere, have collected 15 meteorological element data of this area's electric power accident data of continuous 48 months and correspondence thereof.Secondly first eliminate the multicollinearity of 15 meteorological elements by factor analysis, be extracted temperature factor, rainfall factor, humidity factor, the wind-force factor 4 class principal element, application Logistic returns and establishes meteorologic factor and affect model to electric power accident.The inner link of meteorological condition and electric power accident probed into by model, and demonstrates the accuracy of models fitting with the test samples of 2010, and positive discussion has been carried out in the foundation for electric power accident early warning mechanism.

In the research of " the Central China Power Grid load forecasting method based on meteorological factor is studied ", on the basis of analysis load variations rule various festivals or holidays, meteorological factor is utilized to make prediction variable, use dynamic General Linear to return and the mixed linear regression method that combines of autoregression and nonlinear Artificial Neural Network to carry out Central China Power Grid daily load and Daily treatment cost and day minimum load prediction. to 12 months the independent sample test prediction of totally 365 days show, this objective scheme can meet the needs of traffic scheduling to the precision of prediction of Central China network load.

In the research of " meteorological factor and electric power obstacle relationship assessment ", the 11 years electric power obstacle data provided with Dezhou power administration power-management centre, go out by relative analysis method the infringement that thunder and lightning, strong wind >=17.0m/s and intra day ward >=10.0mm wind and rain weather (these three kinds occur that number of days sum x represents and is referred to as meteorological multi-stress meteorological factor year) causes operation power, thus draw the preliminary funtcional relationship between electric power obstacle (y) and meteorological multi-stress (x).

At present, because shortage is to the investigation collection of history meteorological disaster information and statistical study, to making not set up meteorological disaster Genetic Model, to realize the prediction to meteorological disaster.In meteorological disaster prediction, do not consider that ice and snow congeals, the actual influence to electrical network business such as thunder and lightning and mountain fire, the research being partial to vertical linear, lacks the holistic approach of the analysis of many inducements and the linear regression of diversification more.

Summary of the invention

The object of this invention is to provide a kind of meteorological factor based on CIM standard and affect electrical network hazard model modeling method, it can set up a model accurate, predict electrical network disaster caused by meteorological factor with becoming more meticulous.

The present invention is achieved like this, and a kind of meteorological factor based on CIM standard affects electrical network hazard model modeling method, it is characterized in that, it comprises following method step:

(1), arithmetic of linearity regression is combined structure based on the method for meteorological disaster Genetic Model improving arithmetic of linearity regression with meteorology, the method is using meteorological factor as independent variable, meteorological disaster is as dependent variable, utilize the related coefficient that arithmetic of linearity regression obtains between different independent variable and dependent variable, set up corresponding multiple linear regression model;

(2), the method step of former regression model is revised to the auto modification of related coefficient in (1), it comprise data acquisition in some cycles method step and together with image data in this cycle is put with historical data as new independent variable observed reading, utilize arithmetic of linearity regression to recalculate related coefficient, revise the method step of former regression model.The described cycle is 1 year.

Described multiple linear regression model be congeal for ice and snow respectively, thunder and lightning and mountain fire set up corresponding multiple linear regression model, this multiple linear regression model comprises

1, ice and snow congeals disaster

$y_1=\left\{\begin{array}{c}a+b{x}_1+c{x}_2+d{x}_3+e{x}_4+f{x}_5+g{x}_6+h{x}_7,x_1\≤0\\ 0,x_1>0\end{array}\right.---\left(1\right)$

2, Lightning Disaster

y ₂＝a+bx ₁+cx ₂+dx ₃+ex ₄+fx ₅+gx ₆+hx ₇(2)

3, mountain fire disaster

y ₃＝a+bx ₁+cx ₂+dx ₃+ex ₄+gx ₆+hx ₇,x ₅＝0 (3)

In formula, x ₁to x ₇be respectively temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, its unit be respectively DEG C, kPa, hPa, m/s, mm, mm and MJ/m ²; A to h is respectively corresponding related coefficient, and their value is determined by arithmetic of linearity regression.

In described some cycles the method step of data acquisition be included in data acquisition that ice and snow congeals zero point every day between the disaster emergence period, the data acquisition of each hour and the collection of every 12 hour datas between the mountain fire disaster emergence period between the Lightning Disaster emergence period.

Described image data in this cycle is put with historical data together with as new independent variable observed reading, utilize arithmetic of linearity regression to recalculate related coefficient, the method step revising former regression model is:

If dependent variable y and independent variable x ₁, x ₂..., x _mtotal n group actual observation data:

Assuming that Dependent variable, y and independent variable x ₁, x ₂..., x _mbetween there is linear relationship, its mathematical model is:

y _j＝β ₀+β ₁x _1j+β ₂x _2j+...+β _mx _mj+ε _j(j＝1,2,…,n)

(2-4)

In formula, x ₁, x ₂..., x _mfor the general variance that can observe or the stochastic variable for observing; Y is the stochastic variable that can observe, with x ₁, x ₂..., x _mand become, be put to the test error effect; ε _jfor separate and all obey N (0, σ ²) stochastic variable; According to actual observed value to β ₀, β ₁, β ₂..., β _mand variances sigma ²make an estimate;

Equation of linear regression is as follows:

If y is to x ₁, x ₂..., x _mm unit equation of linear regression be:

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+\·\·\·+b_m{x}_m---(2-5)$

B wherein ₀, b ₁, b ₂..., b _mfor β ₀, β ₁, β ₂..., β _mleast-squares estimation value; I.e. b ₀, b ₁, b ₂..., b _mactual observed value y and regression estimates value should be made sum of square of deviations minimum;

Order $\begin{array}{c}Q=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{(y_j-{\hat{y}}_j)}^2\\ =\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-b_2{x}_{2j}-\·\·\·-b_m{x}_{\mathrm{mj}})}^2\end{array}$

Q is about b ₀, b ₁, b ₂..., b _mm+1 meta-function.

Ask the method for extreme value according to the multivariate function in the differential calculus, if make Q reach minimum, then should have:

$\begin{array}{c}\frac{\∂Q}{{\∂b}_0}=-2\underset{j=1}{\overset{n}{\mathrm{\Σ}}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-\·\·\·-b_m{x}_{\mathrm{mj}})=0\\ \frac{\∂Q}{\∂b_i}=-2\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{ij}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-\·\·\·-b_m{x}_{\mathrm{mj}})=0\end{array}---(2-6)$

Wherein (i=1,2 ..., m)

Through arranging:

Can be obtained by first equation in system of equations (3-12)

$b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\·\·\·b_m{\stackrel{-}{x}}_m---(2-8)$

Namely $b_0=\stackrel{-}{y}-\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{b}_i{\stackrel{-}{x}}_i$

$\stackrel{-}{y}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{y}_j,{\stackrel{-}{x}}_i=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_{\mathrm{ij}}---(2-9)$

If note

$\begin{array}{cc}{\mathrm{SS}}_i=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)}^2,& {\mathrm{SS}}_y=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{(y_j-\stackrel{-}{y})}^2\end{array}$

$\begin{array}{cc}S{P}_{\mathrm{ik}}=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)(x_{\mathrm{kj}}-{\stackrel{-}{x}}_k)={\mathrm{SP}}_{\mathrm{ki}}& {\mathrm{SP}}_{\mathrm{io}}=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)(y_j-\stackrel{-}{y})\end{array}$

(i、k＝1、2、…、m；i≠k)

And will substituting into rear m equation in system of equations (2-7) respectively, can obtain about partial regression coefficient b through arranging ₁, b ₂..., b _mnormal equation group be:

Separate normal equation group (2-10) partial regression coefficient b ₁, b ₂..., b _msolution, and

$b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\·\·\·-b_m{\stackrel{-}{x}}_m---(2-11)$

So obtain m unit equation of linear regression

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+\·\·\·+b_m{x}_m---(2-12)$

The figure of m unit equation of linear regression is a plane of m+1 dimension space, is called regression plane; b ₀be called regression constant item, work as x ₁=x ₂=...=x _mwhen=0, at b ₀when being of practical significance, b ₀represent the initial value of y; b _i(i=1,2 ..., m) be called that Dependent variable, y is to independent variable x _ipartial regression coefficient, represent except independent variable x _iwhen all the other m-1 independent variable in addition all immobilizes, independent variable x _ioften change a unit, the unit amount of Dependent variable, y mean change, works as b _iduring >0, independent variable x _ioften increase a unit, Dependent variable, y on average increases b _iindividual unit; Work as b _iduring <0, independent variable x _ioften increase a unit, Dependent variable, y decreased average b _iindividual unit;

If will $b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{\stackrel{-}{x}}_m$ Substitute into above formula, then

$\hat{y}=\stackrel{-}{y}+b_1(x_1-{\stackrel{-}{x}}_1)+b_2(x_2-{\stackrel{-}{x}}_2)+\&CenterDot;\&CenterDot;\&CenterDot;+b_m(x_m-{\stackrel{-}{x}}_m)---(2-13)$

(2-13) formula is also that y is to x ₁, x ₂..., x _mm unit equation of linear regression;

For normal equation group (2-10), note

$b=\left[\begin{array}{c}{b}_1\\ b_2\\ .\\ .\\ .\\ b_m\end{array}\right],$ $B=\left[\begin{array}{c}{\mathrm{SP}}_{10}\\ {\mathrm{SP}}_{20}\\ .\\ .\\ .\\ {\mathrm{SP}}_{m0}\end{array}\right]$

Then normal equation group (2-10) available matrix representation is

Namely

Ab＝B (2-15)

Wherein A is the matrix of coefficients of normal equation group, b is partial regression coefficient matrix, B is constant term matrix;

If the inverse matrix of coefficient matrices A is C matrix, i.e. A ^-1=C, then

Wherein: C entry of a matrix element c _ij(i, j=1,2 ..., _m) be called Gauss's multiplier, be in multiple linear regression analysis required for significance test;

Matrix equation (2-16) is solved, has:

b＝A ^-1B

b＝CB

That is:

About partial regression coefficient b ₁, b ₂..., b _msolution can be expressed as:

b _i＝c _i1SP ₁₀+c _i2SP ₂₀+…+c _imSP _m0(i＝1、2、…、m)

Or $b_i=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{c}_{\mathrm{ij}}s{p}_{j0}---(2-18)$

And $b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{\stackrel{-}{x}}_m.$

Described data (temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, Ge Tiao supply line number of lightning strokes, mountain fire number of times, icing number of times, total tripping operation number of times, other number of times) value mode, accurate value instrument need be adopted and be weighted on average basis, by the integration algorithm of many factors (based on the meteorological disaster Genetic Model of arithmetic of linearity regression, power supply component failure rate algorithm), finally carry out being worth and (ice and snow occurs congeal, the grade of thunder and lightning and mountain fire disaster, the failure rate of Ge Tiao supply line under Hazard Meteorological factor under different Meteorological Grade) output.

Beneficial effect of the present invention is: the algorithm provided in model of the present invention is in the rank of advanced units in industry, the Predicting and analysis of meteorological factor to electrical network disaster can be carried out well, predict the generation of meteorological disaster practically, farthest reduce in process of distributing electricity due to economic loss that meteorologic factor disaster is brought.The value mode of each entry value in this research, need adopt accurate value instrument and be weighted on average basis, by the integration algorithm of many factors, finally carry out the output be worth, relative in disposable value, accuracy rate is higher, realizes the risk control become more meticulous.

Accompanying drawing explanation

Fig. 1 is the line failure rate assessment models FB(flow block) under typical meteorological disaster.

Fig. 2 is the process flow diagram of meteorological disaster Genetic Model.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described further.

Principle of the present invention is: for the defect of traditional meteorological Disaster cause model, utilize multiple linear regression model less demanding to sample, modeling approach is clear, the features such as convergence is good, combine structure based on the meteorological disaster Genetic Model improving arithmetic of linearity regression by arithmetic of linearity regression with meteorology.This model is using meteorological factor as independent variable, and meteorological disaster, as dependent variable, utilizes the related coefficient that arithmetic of linearity regression obtains between different independent variable and dependent variable, sets up out corresponding multiple linear regression model.Then former regression model is revised to the auto modification of related coefficient, the method step of data acquisition in some cycles and together with image data in this cycle is put with historical data as new independent variable observed reading, utilize arithmetic of linearity regression to recalculate related coefficient, revise the method step of former regression model.

Described data (temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, Ge Tiao supply line number of lightning strokes, mountain fire number of times, icing number of times, total tripping operation number of times, other number of times) value mode, accurate value instrument need be adopted and be weighted on average basis, by the integration algorithm of many factors (based on the meteorological disaster Genetic Model of arithmetic of linearity regression, power supply component failure rate algorithm), finally carry out being worth and (ice and snow occurs congeal, the grade of thunder and lightning and mountain fire disaster, the failure rate of Ge Tiao supply line under Hazard Meteorological factor under different Meteorological Grade) output.

Congeal with ice and snow, that thunder and lightning and mountain fire set up corresponding multivariate regression model embodiment be as follows:

Congeal for ice and snow respectively, thunder and lightning and mountain fire set up corresponding multivariate regression model, with reference to former research with determine through test of many times as follows:

1, ice and snow congeals disaster

$y_1=\left\{\begin{array}{c}a+b{x}_1+c{x}_2+d{x}_3+e{x}_4+f{x}_5+g{x}_6+h{x}_7,x_1\&le;0\\ 0,x_1>0\end{array}\right.---\left(1\right)$

2, Lightning Disaster

y ₂＝a+bx ₁+cx ₂+dx ₃+ex ₄+fx ₅+gx ₆+hx ₇(2)

3, mountain fire disaster

y ₃＝a+bx ₁+cx ₂+dx ₃+ex ₄+gx ₆+hx ₇,x ₅＝0 (3)

In formula, x ₁to x ₇be respectively temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, its unit be respectively DEG C, kPa, hPa, m/s, mm, mm and MJ/m ².A to h is respectively corresponding related coefficient, and their value will be determined by arithmetic of linearity regression.

As shown in Figure 1, disaster loss grade that can be different according to model partition, its division methods is: when generation ice and snow congeals disaster, to congeal the grade of disaster based on the ice and snow measurable ice and snow of hazard model that congeals, in conjunction with the grade classification of snow and rain and the historical failure rate statistics of every bar circuit, draw the fault level model of current line under ice and snow congeals disaster, then add up the model state of the prediction of output; When there is Lightning Disaster and mountain fire disaster, its model prediction principle is the same.

Due to along with passage of time, the weather data obtained increases, and accuracy is also higher, therefore three kinds of Disaster cause models should continuous its related coefficient of auto modification.In the cycle being auto modification with 1 year, as new independent variable observed reading together with putting with historical data by the data of new a year, utilize arithmetic of linearity regression to recalculate related coefficient, revise former regression model.Because the time difference that three kinds of disasters continue when occurring is comparatively large, therefore the time point data that its model auto modification is used are also different, as follows respectively:

1, ice and snow congeals disaster

Ice and snow congeals the data gathered at zero point every day between the disaster emergence period;

2, Lightning Disaster

The data that between the Lightning Disaster emergence period, each hour gathers, i.e. integral point data;

3, mountain fire disaster

The data that between the mountain fire disaster emergence period, every 12 hours gather.

If dependent variable y and independent variable x ₁, x ₂..., x _mtotal n group actual observation data:

Assuming that dependent variable y and independent variable x ₁, x ₂..., x _mbetween there is linear relationship, its mathematical model is:

y _j＝β ₀+β ₁x _1j+β ₂x _2j+...+β _mx _mj+ε _j(j＝1,2,…,n)

(2-4)

In formula, x ₁, x ₂..., x _mfor the general variance (or the stochastic variable for observing) that can observe; Y is the stochastic variable that can observe, with x ₁, x ₂..., x _mand become, be put to the test error effect; ε _jfor separate and all obey N (0, σ ²) stochastic variable.We can according to actual observed value to β ₀, β ₁, β ₂..., β _mand variances sigma ²make an estimate.

Equation of linear regression is as follows:

If y is to x ₁, x ₂..., x _mm unit equation of linear regression be:

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+b_m{x}_m---(2-5)$

B wherein ₀, b ₁, b ₂..., b _mfor β ₀, β ₁, β ₂..., β _mleast-squares estimation value.I.e. b ₀, b ₁, b ₂..., b _mactual observed value y and regression estimates value should be made sum of square of deviations minimum.

Order $Q=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-{\hat{y}}_j)}^2$

$=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{x}_{\mathrm{mj}})}^2$

Q is about b ₀, b ₁, b ₂..., b _mm+1 meta-function.

Ask the method for extreme value according to the multivariate function in the differential calculus, if make Q reach minimum, then should have:

$\begin{array}{c}\frac{\&PartialD;Q}{{\&PartialD;b}_0}=-2\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{x}_{\mathrm{mj}})=0\\ \frac{\&PartialD;Q}{\&PartialD;b_i}=-2\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ij}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{x}_{\mathrm{mj}})=0\end{array}---(2-6)$

Wherein (i=1,2 ..., m)

Through arranging:

Can be obtained by first equation in system of equations (3-12)

$b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;b_m{\stackrel{-}{x}}_m---(2-8)$

Namely $b_0=\stackrel{-}{y}-\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{b}_i{\stackrel{-}{x}}_i$

$\stackrel{-}{y}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{y}_j,{\stackrel{-}{x}}_i=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ij}}---(2-9)$

If note

$\begin{array}{cc}{\mathrm{SS}}_i=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)}^2,& {\mathrm{SS}}_y=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-\stackrel{-}{y})}^2\end{array}$

$\begin{array}{cc}S{P}_{\mathrm{ik}}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)(x_{\mathrm{kj}}-{\stackrel{-}{x}}_k)={\mathrm{SP}}_{\mathrm{ki}}& {\mathrm{SP}}_{\mathrm{io}}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(x_{\mathrm{ij}}-{\stackrel{-}{x}}_i)(y_j-\stackrel{-}{y})\end{array}$

(i、k＝1、2、…、m；i≠k)

And will substituting into rear m equation in system of equations (2-7) respectively, can obtain about partial regression coefficient b through arranging ₁, b ₂..., b _mnormal equation group be:

Separate normal equation group (2-10) partial regression coefficient b ₁, b ₂..., b _msolution, and

$b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{\stackrel{-}{x}}_m---(2-11)$

So obtain m unit equation of linear regression

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+b_m{x}_m---(2-12)$

The figure of m unit equation of linear regression is a plane of m+1 dimension space, is called regression plane; b ₀be called regression constant item, work as x ₁=x ₂=...=x _mwhen=0, at b ₀when being of practical significance, b ₀represent the initial value of y; b _i(i=1,2 ..., m) be called that Dependent variable, y is to independent variable x _ipartial regression coefficient, represent except independent variable x _iwhen all the other m-1 independent variable in addition all immobilizes, independent variable x _ioften change a unit, the unit amount of Dependent variable, y mean change, exactly, works as b _iduring >0, independent variable x _ioften increase a unit, Dependent variable, y on average increases b _iindividual unit; Work as b _iduring <0, independent variable x _ioften increase a unit, Dependent variable, y decreased average b _iindividual unit.

If will $b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{\stackrel{-}{x}}_m$ Substitute into above formula, then

$\hat{y}=\stackrel{-}{y}+b_1(x_1-{\stackrel{-}{x}}_1)+b_2(x_2-{\stackrel{-}{x}}_2)+\&CenterDot;\&CenterDot;\&CenterDot;+b_m(x_m-{\stackrel{-}{x}}_m)---(2-13)$

(2-13) formula is also that y is to x ₁, x ₂..., x _mm unit equation of linear regression.

For normal equation group (2-10), note

$b=\left[\begin{array}{c}{b}_1\\ b_2\\ .\\ .\\ .\\ b_m\end{array}\right],$ $B=\left[\begin{array}{c}{\mathrm{SP}}_{10}\\ {\mathrm{SP}}_{20}\\ .\\ .\\ .\\ {\mathrm{SP}}_{m0}\end{array}\right]$

Then normal equation group (2-10) available matrix representation is

Namely

Ab＝B (2-15)

Wherein A is the matrix of coefficients of normal equation group, b is partial regression coefficient matrix (column vector), B is constant term matrix (column vector).

If the inverse matrix of coefficient matrices A is C matrix, i.e. A ^-1=C, then

Wherein: C entry of a matrix element c _ij(i, j=1,2 ..., _m) be called Gauss's multiplier, be in multiple linear regression analysis required for significance test.

About the inverse matrix A asking coefficient matrices A ^-1method have multiple, as the elementary transform method etc. of row (or row).

Matrix equation (2-16) is solved, has:

b＝A ^-1B

b＝CB

That is:

About partial regression coefficient b ₁, b ₂..., b _msolution can be expressed as:

b _i＝c _i1SP ₁₀+c _i2SP ₂₀+…+c _imSP _m0(i＝1、2、…、m)

Or $b_i=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{c}_{\mathrm{ij}}s{p}_{j0}---(2-18)$

And $b_0=\stackrel{-}{y}-b_1{\stackrel{-}{x}}_1-b_2{\stackrel{-}{x}}_2-\&CenterDot;\&CenterDot;\&CenterDot;-b_m{\stackrel{-}{x}}_m.$

According to the model of above-mentioned continuous auto modification related coefficient, input the integral point meteorological factor data x of each subregion at present period ₁to x ₇calculate meteorological disaster dependent variable y value, can according to shown in Fig. 2, if y < 0.5,0.5 < y < 0.7,0.7 < y < 0.9 and y > 0.9, meteorological disaster corresponding general, yellow, orange and red respectively, in conjunction with the grade of actual snow and rain, the electrical network disaster that meteorological factor causes can be predicted exactly.

Claims (7)

1. the meteorological factor based on CIM standard affects an electrical network hazard model modeling method, and it is characterized in that, it comprises following method step:

(1), arithmetic of linearity regression is combined structure based on the method for meteorological disaster Genetic Model improving arithmetic of linearity regression with meteorology, the method is using meteorological factor as independent variable, meteorological disaster is as dependent variable, utilize the related coefficient that arithmetic of linearity regression obtains between different independent variable and dependent variable, set up corresponding multiple linear regression model;

(2), the method step of former regression model is revised to the auto modification of related coefficient in (1), it comprise data acquisition in some cycles method step and together with image data in this cycle is put with historical data as new independent variable observed reading, utilize arithmetic of linearity regression to recalculate related coefficient, revise the method step of former regression model.

2. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 1, it is characterized in that, described multiple linear regression model be congeal for ice and snow respectively, thunder and lightning and mountain fire set up corresponding multiple linear regression model, this multiple linear regression model comprises

1, ice and snow congeals disaster

$y_1=\left\{\begin{array}{c}a+b{x}_1+{\mathrm{cx}}_2+{\mathrm{dx}}_3+e{x}_4+{\mathrm{fx}}_5+{\mathrm{gx}}_6+{\mathrm{hx}}_7,x_1\&le;0\\ 0,x_1>0\end{array}\right.---\left(1\right)$

2, Lightning Disaster

y ₂＝a+bx ₁+cx ₂+dx ₃+ex ₄+fx ₅+gx ₆+hx ₇(2)

3, mountain fire disaster

y ₃＝a+bx ₁+cx ₂+dx ₃+ex ₄+gx ₆+hx ₇,x ₅＝0 (3)

In formula, x ₁to x ₇be respectively temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, its unit be respectively DEG C, kPa, hPa, m/s, mm, mm and MJ/m ²; A to h is respectively corresponding related coefficient, and their value is determined by arithmetic of linearity regression.

3. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 2, it is characterized in that, the described cycle is 1 year.

4. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 2 or claim 3, it is characterized in that, in described some cycles the method step of data acquisition be included in data acquisition that ice and snow congeals zero point every day between the disaster emergence period, the data acquisition of each hour and the collection of every 12 hour datas between the mountain fire disaster emergence period between the Lightning Disaster emergence period.

5. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 4, it is characterized in that, described image data in this cycle is put with historical data together with as new independent variable observed reading, utilize arithmetic of linearity regression to recalculate related coefficient, the method step revising former regression model is:

If dependent variable y and independent variable x ₁, x ₂..., x _mtotal n group actual observation data:

Assuming that Dependent variable, y and independent variable x ₁, x ₂..., x _mbetween there is linear relationship, its mathematical model is:

y _j＝β ₀+β ₁x _1j+β ₂x _2j+...+β _mx _mj+ε _j(j＝1,2,…,n)

(2-4)

In formula, x ₁, x ₂..., x _mfor the general variance that can observe or the stochastic variable for observing; Y is the stochastic variable that can observe, with x ₁, x ₂..., x _mand become, be put to the test error effect; ε _jfor separate and all obey N (0, σ ²) stochastic variable; According to actual observed value to β ₀, β ₁, β ₂..., β _mand variances sigma ²make an estimate;

Equation of linear regression is as follows:

If y is to x ₁, x ₂..., x _mm unit equation of linear regression be:

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+...+b_m{x}_m---(2-5)$

B wherein ₀, b ₁, b ₂..., b _mfor β ₀, β ₁, β ₂..., β _mleast-squares estimation value; I.e. b ₀, b ₁, b ₂..., b _mactual observed value y and regression estimates value should be made sum of square of deviations minimum;

Order $\begin{array}{c}Q=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-{\hat{y}}_j)}^2\\ =\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-...-b_m{x}_{\mathrm{mj}})}^2\end{array}$

Q is about b ₀, b ₁, b ₂..., b _mm+1 meta-function.

Ask the method for extreme value according to the multivariate function in the differential calculus, if make Q reach minimum, then should have:

$\frac{\&PartialD;Q}{\&PartialD;b_0}=-2\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-...-b_m{x}_{\mathrm{mj}})=0$

$\frac{\&PartialD;Q}{\&PartialD;b_i}=-2\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ij}}(y_j-b_0-b_1{x}_{1j}-b_2{x}_{2j}-...-b_m{x}_{\mathrm{mj}})=0---(2-6)$

Wherein (i=1,2 ..., m)

Through arranging:

Can be obtained by first equation in system of equations (3-12)

$b_0=\stackrel{\&OverBar;}{y}-b_1{\stackrel{\&OverBar;}{x}}_1-b_2{\stackrel{\&OverBar;}{x}}_2-...-b_m{\stackrel{\&OverBar;}{x}}_m---(2-8)$

Namely $b_0=\stackrel{\&OverBar;}{y}-\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{b}_i{\stackrel{\&OverBar;}{x}}_i$

$\stackrel{\&OverBar;}{y}=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{y}_j,{\stackrel{\&OverBar;}{x}}_i=\frac{1}{n}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{x}_{\mathrm{ij}}---(2-9)$

If note

$\begin{array}{cc}{\mathrm{SS}}_i=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(x_{\mathrm{ij}}-{\stackrel{\&OverBar;}{x}}_i)}^2,& {\mathrm{SS}}_y=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{(y_j-\stackrel{\&OverBar;}{y})}^2\end{array}$

$\begin{array}{cc}{\mathrm{SP}}_{\mathrm{ik}}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(x_{\mathrm{ij}}-{\stackrel{\&OverBar;}{x}}_i)(x_{\mathrm{kj}}-{\stackrel{\&OverBar;}{x}}_k)={\mathrm{SP}}_{\mathrm{ki}}& {\mathrm{SP}}_{\mathrm{io}}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(x_{\mathrm{ij}}-{\stackrel{\&OverBar;}{x}}_i)(y_j-\stackrel{\&OverBar;}{y})\end{array}$

(i、k＝1、2、…、m；i≠k)

And will substituting into rear m equation in system of equations (2-7) respectively, can obtain about partial regression coefficient b through arranging ₁, b ₂..., b _mnormal equation group be:

Separate normal equation group (2-10) partial regression coefficient b ₁, b ₂..., b _msolution, and

$b_0=\stackrel{\&OverBar;}{y}-b_1{\stackrel{\&OverBar;}{x}}_1-b_2{\stackrel{\&OverBar;}{x}}_2-...-b_m{\stackrel{\&OverBar;}{x}}_m---(2-11)$

So obtain m unit equation of linear regression

$\hat{y}=b_0+b_1{x}_1+b_2{x}_2+...+b_m{x}_m---(2-12)$

The figure of m unit equation of linear regression is a plane of m+1 dimension space, is called regression plane; b ₀be called regression constant item, work as x ₁=x ₂=...=x _mwhen=0, at b ₀when being of practical significance, b ₀represent the initial value of y; b _i(i=1,2 ..., m) be called that Dependent variable, y is to independent variable x _ipartial regression coefficient, represent except independent variable x _iwhen all the other m-1 independent variable in addition all immobilizes, independent variable x _ioften change a unit, the unit amount of Dependent variable, y mean change, works as b _iduring >0, independent variable x _ioften increase a unit, Dependent variable, y on average increases b _iindividual unit; Work as b _iduring <0, independent variable x _ioften increase a unit, Dependent variable, y decreased average b _iindividual unit;

If will $b_0=\stackrel{\&OverBar;}{y}-b_1{\stackrel{\&OverBar;}{x}}_1-b_2{\stackrel{\&OverBar;}{x}}_2-...-b_m{\stackrel{\&OverBar;}{x}}_m$ Substitute into above formula, then

$\hat{y}=\stackrel{\&OverBar;}{y}+b_1(x_1-{\hat{x}}_1)+b_2(x_2-{\hat{x}}_2)+...+b_m(x_m-{\stackrel{\&OverBar;}{x}}_m)---(2-13)$ (2-13) formula is also that y is to x ₁, x ₂..., x _mm unit equation of linear regression;

For normal equation group (2-10), note

$A=\left[\begin{array}{cccc}{\mathrm{SS}}_1& {\mathrm{SP}}_{12}& ...& {\mathrm{SP}}_{1m}\\ {\mathrm{SP}}_{21}& {\mathrm{SS}}_2& ...& {\mathrm{SP}}_{2m}\\ .& .& & .\\ .& .& ...& .\\ .& .& & .\\ {\mathrm{SP}}_{m1}& {\mathrm{SP}}_{m2}& ...& {\mathrm{SS}}_m\end{array}\right],b=\left[\begin{array}{c}{b}_1\\ b_2\\ .\\ .\\ .\\ b_m\end{array}\right],B=\left[\begin{array}{c}{\mathrm{SP}}_{10}\\ {\mathrm{SP}}_{20}\\ .\\ .\\ .\\ {\mathrm{SP}}_{m0}\end{array}\right]$

Then normal equation group (2-10) available matrix representation is

$\left[\begin{array}{cccc}{\mathrm{SS}}_1& {\mathrm{SP}}_{12}& ...& {\mathrm{SP}}_{1m}\\ {\mathrm{SP}}_{21}& {\mathrm{SS}}_2& ...& {\mathrm{SP}}_{2m}\\ .& .& & .\\ .& .& ...& .\\ .& .& & .\\ {\mathrm{SP}}_{m1}& {\mathrm{SP}}_{m2}& ...& {\mathrm{SS}}_m\end{array}\right]\left[\begin{array}{c}{b}_1\\ b_2\\ .\\ .\\ .\\ b_m\end{array}\right]=\left[\begin{array}{c}{\mathrm{SP}}_{10}\\ {\mathrm{SP}}_{20}\\ .\\ .\\ .\\ {\mathrm{SP}}_{m0}\end{array}\right]---(2-14)$

Namely

Ab＝B (2-15)

Wherein A is the matrix of coefficients of normal equation group, b is partial regression coefficient matrix, B is constant term matrix;

If the inverse matrix of coefficient matrices A is C matrix, i.e. A ^-1=C, then

$C=A^{-1}={\left[\begin{array}{cccc}{\mathrm{SS}}_1& {\mathrm{SP}}_{12}& ...& {\mathrm{SP}}_{1m}\\ {\mathrm{SP}}_{21}& {\mathrm{SS}}_2& ...& {\mathrm{SP}}_{2m}\\ .& .& & .\\ .& .& ...& .\\ .& .& & .\\ {\mathrm{SP}}_{m1}& {\mathrm{SP}}_{m2}& ...& {\mathrm{SS}}_m\end{array}\right]}^{-1}=\left[\begin{array}{cccc}{c}_{11}& c_{12}& ...& c_{1m}\\ c_{21}& c_{22}& ...& c_{2m}\\ .& .& & .\\ .& .& ...& .\\ .& .& & .\\ c_{m1}& c_{m2}& ...& c_{\mathrm{mm}}\end{array}\right]---(2-16)$

Wherein: C entry of a matrix element c _ij(i, j=1,2 ..., m) be called Gauss's multiplier, be in multiple linear regression analysis required for significance test;

Matrix equation (2-16) is solved, has:

b＝A ^-1B

b＝CB

That is:

$\left[\begin{array}{c}{b}_1\\ b_2\\ .\\ .\\ .\\ b_m\end{array}\right]=\left[\begin{array}{cccc}{c}_{11}& c_{12}& ...& c_{1m}\\ c_{21}& c_{22}& ...& c_{2m}\\ .& .& & .\\ .& .& ...& .\\ .& .& & .\\ c_{m1}& c_{m2}& ...& c_{\mathrm{mm}}\end{array}\right]\left[\begin{array}{c}{\mathrm{SP}}_{10}\\ S{P}_{20}\\ .\\ .\\ .\\ {\mathrm{SP}}_{m0}\end{array}\right]---(2-17)$

About partial regression coefficient b ₁, b ₂..., b _msolution can be expressed as:

b _i＝c _i1SP ₁₀+c _i2SP ₂₀+…+c _imSP _m0(i＝1、2、…、m)

Or $b_i=\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{c}_{\mathrm{ij}}{\mathrm{sp}}_{j0}---(2-18)$

And $b_0=\stackrel{\&OverBar;}{y}-b_1{\stackrel{\&OverBar;}{x}}_1-b_2{\stackrel{\&OverBar;}{x}}_2-...-b_m{\stackrel{\&OverBar;}{x}}_m.$

6. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 1, it is characterized in that, the data of described collection comprise temperature, air pressure, humidity, wind speed, evaporation capacity, precipitation and sunshine, Ge Tiao supply line number of lightning strokes, mountain fire number of times, icing number of times and the number of times that always trips.

7. affect electrical network hazard model modeling method based on the meteorological factor of CIM standard as claimed in claim 6, it is characterized in that, described data carry out weighted mean by after the collection of value instrument.