CN105139099A - LLS-based regional lightning damage feature correlation analysis method - Google Patents

Abstract

An LLS-based regional lightning damage feature correlation analysis method is characterized in that the correlation between the spatial features of cloud-to-ground lightning and the meteorological background extracted from NECP reanalysis data is determined through singular value decomposition based on cloud-to-ground lightning monitoring data of a lightning location system and the NECP reanalysis data, and therefore, factors strongly correlated with the features of regional lightning damage can be analyzed and identified. The method of the invention has the characteristics of easy field application, high-reliability algorithm, and high reference value in auxiliary decision making. The method of the invention can be applied to a provincial power company power dispatching institution which has rich experience in running a lightning location system, provides accurate cloud-to-ground lightning analysis and is equipped with a grid meteorological information system in a power grid. A method for screening meteorological conditions affecting the features of regional lightning damage is provided, and the construction of the capability of the power grid to resist lightning damage is promoted.

Description

A kind of region damage to crops caused by thunder feature correlation analysis method based on LLS

Technical field

The present invention relates to a kind of region damage to crops caused by thunder feature correlation analysis method based on LLS, belong to high voltage electric equipment operation maintenance service technique field.

Background technology

Material Takeoff such as " national Lightning Disaster representative instance compilation " according in recent years, all industries subjecting to Lightning disservice be broadly divided into electric power, broadcasting and TV, chemical industry, computing machine, communication and other etc. 6 classes, wherein the lightning strike accident proportion of power high voltage electric equipment is higher, accounts for 24.7% of sum.With regard to power industry, lightning stroke trip is in first of all kinds of failure accident always.According to incompletely statistics, in China's 66 ~ 500kV transmission line of electricity tripping operation sum in 2003 close to 40.23% is caused by thunderbolt; 220 ~ 500kV lightning stroke trip belonging to State Grid Corporation of China in 2004 accounts for 32.72% of total tripping operation quantity.Although the design standards of primary system middle/high-voltage electrical appliance equipment ability that is high, that resist damage to crops caused by thunder is better than low-voltage electric apparatus, but the defence capability that the design standards only by improving primary equipment strengthens lightning hazards certainly will cause great capital input to be born, be difficult at short notice be achieved, therefore, how to strengthen the analysis and research of damage to crops caused by thunder feature, lifting electrical secondary system damage to crops caused by thunder monitoring and warning analysis ability seems particularly important.

There is damage to crops caused by thunder event and regional climate, natural conditions are in close relations.With regard to the mechanism that damage to crops caused by thunder occurs, the generation of thunder and lightning caused primarily of the Small and Medium Sized synoptic process with multiple weather phenomenon, often with strong wind and heavy rain when damage to crops caused by thunder occurs, sometimes even occurs hail and spout all diastrous weather.Weather background research before and after damage to crops caused by thunder occurs is significant for the analysis and research of damage to crops caused by thunder event, as significant for the elaboration of Electrification mechanism in cloud in backgrounds on the large scale circulation, Convective dynamic parameters, power/heating power trigger mechanism etc., because of but one of major issue of atmospheric electricity and thunder and lightning principle.For the damage to crops caused by thunder period of right time meteorological physical element for analysis, contribute to the early warning technology level improving damage to crops caused by thunder power-off fault, possible thunderbolt subregion scope divide and in the forecast of lightning strike area probability of equipment failure by positive role certain for performance.

Summary of the invention

The object of the invention is, the accuracy of front accident anticipation occurring to improve region damage to crops caused by thunder, enriching high voltage electric equipment lightning protection early warning reference information, the invention discloses a kind of region damage to crops caused by thunder feature correlation analysis method based on LLS.

Technical scheme of the present invention is, the present invention is based on the region damage to crops caused by thunder feature correlation analysis method of LLS, by lightning location system Historical Monitoring data and history Environmental same period forecasting centre NCEP analysis of data (NationalCentersforEnvironmentalPrediction again, NCEP), singular value decomposition method is adopted to dodge the correlation analysis of space distribution and meteorological element spatial analysis with carrying out, dodge the meteorological element classification of Spatial distribution with determining appreciable impact, thus disclose the restriction condition of region damage to crops caused by thunder feature according to meteorological mechanism.

A kind of region damage to crops caused by thunder feature correlation analysis method based on LLS of the present invention, data are analyzed again according to lightning location system ground sudden strain of a muscle Monitoring Data and NECP, adopt singular value decomposition method to dodge space characteristics definitely and analyze the correlationship between meteorological background that data extract again, the analysis of the factor of the strong correlation of feasible region damage to crops caused by thunder feature and identification by NECP.

Described method performing step is as follows:

(1) to LLS lightning monitoring data prediction, LLS monitoring is dodged with obtaining record is processed into 0.01 ° × 0.01 °, 0.05 ° × 0.05 ° and the 0.1 ° × 0.1 ° of three kinds of spatial and temporal resolution CG lightning density of gridding day by day space matrix by gridding method.

(2) again data prediction is analyzed to NECP, the NECP collecting LLS lightning monitoring data history same period analyzes data 0.1 ° × 0.1 ° Grid data again, determine that NECP analyzes data space span again according to LLS monitoring range, the time, span was in accordance with LLS.

(3) LLS lightning monitoring data are carried out and NECP analyzes data coupling relationship analysis again, SVD method is adopted to implement different conversions to lightning monitoring field from the multinomial variable of meteorological element field, the new variables obtained is one group of time coefficient item, and this and two fields mutual relationship respectively between time coefficient that is corresponding, that occur in pairs is the correlationship of lightning monitoring field and meteorological element field.

(4) the meteorological element correlationship analysis of region damage to crops caused by thunder feature is carried out, by SVD analyze draw before N to time coefficient (by the large minispread of covariance), the gridded data that ground can be dodged over time and the same period meteorological element change this challenge in time and be reduced to N to the mutual relationship between time coefficient.

(5) the meteorological effect factor analysis of correlation test and damage to crops caused by thunder active characteristics is carried out, for known one group of related coefficient, the conspicuousness of the method determination correlationship that T can be adopted to check; Because the historical data of LLS mainly concentrates on the second half year, when considering day Value Data, the vector element number of the time coefficient that SVD obtains can meet the sample hypothesis of T inspection.

The correlationship computing method of described lightning monitoring field and meteorological element field are as follows:

It is C that variance matrix is intersected in lightning monitoring field and the association of meteorological element field _sz=<SZ ^t>, symbol <> represent and are averaging; By SVD, find two orthogonal linear transformation matrix L and R, make there is maximization covariance between two fields, that is:

cov(L ^TS,R ^TZ)＝L ^TC _szR＝MAX

$C_{\mathrm{SZ}}=L\left(\begin{array}{cc}\mathrm{\&Sigma;}& 0\\ 0& 0\end{array}\right)R^T$

$\mathrm{\&Sigma;}=\left(\begin{array}{cccc}{\mathrm{\&sigma;}}_1& 0& ...& 0\\ 0& {\mathrm{\&sigma;}}_2& ...& 0\\ ...& ...& ...& ...\\ 0& 0& ...& {\mathrm{\&sigma;}}_r\end{array}\right)$

Wherein, Σ=diag (σ ₁, σ ₂..., σ _r) (r≤min{x, z}), and σ ₁>=σ ₂>=...>=σ _r> 0; σ _i(i=1,2 ..., r) be called the singular value of C;

If A=L ^ts, B=R ^tz, wherein A is called the S time coefficient matrix of actual measurement field, and B is called the time coefficient matrix of simulation yard Z; The kth column vector l of L and R _kand r _k(k=1,2 ...., r) be called the left and right singular vector of kth, namely kth is to spatial mode.

Theoretical by linear algebra, L and R meeting above-mentioned condition can be solved uniquely.

$\begin{array}{cc}L=\left(\begin{array}{cccc}{l}_{11}& l_{12}& ...& l_{1\mathrm{Ns}}\\ l_{21}& l_{22}& ...& l_{2\mathrm{Ns}}\\ ...& ...& ...& ...\\ l_{\mathrm{Ns}1}& l_{\mathrm{Ns}2}& ...& l_{\mathrm{NsNs}}\end{array}\right)& R=\left(\begin{array}{cccc}{r}_{11}& r_{12}& ...& r_{1\mathrm{Nz}}\\ r_{21}& r_{22}& ...& r_{2\mathrm{Nz}}\\ ...& ...& ...& ...\\ r_{\mathrm{Nz}1}& r_{\mathrm{Nz}2}& ...& r_{\mathrm{NzNz}}\end{array}\right)\end{array}$

Described N realizes being calculated as to the coefficient R of the mutual relationship between time coefficient:

Front N is to the kth column vector l of time coefficient _kand r _k, k≤N

$R_k=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})\&CenterDot;(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})]}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})}^2\&CenterDot;\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})}^2}}$

Wherein, i=min (N _s, N _z).

The invention has the beneficial effects as follows, the present invention has considered electric network thunder and lightning positioning system historical data and the correlationship of data is observed, analyzed to all kinds of meteorological element, meteorological background for damage to crops caused by thunder feature is analyzed targetedly, draws the concrete all kinds of physical quantitys affecting damage to crops caused by thunder feature.In view of the practical level of Utilities Electric Co.s at different levels electrical network Meteorological Information System, data information richness improve year by year, it is convenient that the present invention has rig-site utilization, and algorithm reliability is high, the feature that aid decision making reference value is large.

The present invention can be used for electric network thunder and lightning positioning system operating experience enrich, dodge and analyze accurately and have the net province one-level Utilities Electric Co. power dispatching station of electrical network Meteorological Information System, the discriminating method of the meteorology influence condition of region damage to crops caused by thunder feature is provided, promotes the construction of electric network thunderstorm damage defensive ability/resistance ability.

Accompanying drawing explanation

Fig. 1 sums up according to main technical flows of the present invention the algorithm flow schematic diagram drawn.

Embodiment

Algorithm basis of the present invention is svd (SingularValueDecomposition, be called for short SVD), by improving the correlationship qualitative analysis of two Physical Quantity Field in former technical method, realize the analysis and diagnosis characterizing correlationship between the CG lightning density parameter of damage to crops caused by thunder feature and meteorological element field.Input information involved in the present invention comprises LLS lightning monitoring data and analyzes data again with the NECP of LLS lightning monitoring data same periods, when region meteorological observation in electrical network Meteorological Information System (ground, high-altitude) data is complete, also can adopt the meteorological measuring of domestic meteorological department.

The technology used in the present invention performing step comprises:

Step one: LLS lightning monitoring data prediction

LLS monitoring is dodged with obtaining record is processed into 0.01 ° × 0.01 °, 0.05 ° × 0.05 ° and the 0.1 ° × 0.1 ° of three kinds of spatial and temporal resolution CG lightning density of gridding day by day space matrix by gridding method.

Build with grid number and monitor the two-dimensional matrix that number of days is dimension, being generally expressed as:

$S=\left(\begin{array}{cccc}{x}_{11}& x_{12}& ...& x_{1t}\\ x_{21}& x_{22}& ...& x_{2t}\\ ...& ...& ...& ...\\ x_{N_{S}1}& x_{N_{S}2}& ...& x_{N_{S}t}\end{array}\right)$

Wherein, N _srepresent the grid sum of the different resolution in regional extent, the natural number of days in the period is analyzed in t representative, and each matrix element is through variance standardization.

Step 2: NECP analyzes data prediction again

The NECP collecting LLS lightning monitoring data history same period analyzes data 0.1 ° × 0.1 ° Grid data again, and determine that NECP analyzes data space span again according to LLS monitoring range, the time, span was in accordance with LLS.

NECP again analysis of data matrix is generally expressed as:

$Z=\left(\begin{array}{cccc}{y}_{11}& y_{12}& ...& y_{1t}\\ y_{21}& y_{22}& ...& y_{2t}\\ ...& ...& ...& ...\\ y_{N_{z}1}& x_{N_{z}2}& ...& x_{N_{z}t}\end{array}\right)$

Wherein, N _zrepresent the lattice point quantity in regional extent, each matrix element is equally through variance standardization.

Step 3: LLS lightning monitoring data and NECP analyze data coupling relationship analysis again

For ease of describing, by based on the ripe data of LLS, by the two-dimentional real matrix of dodging to gridding day by day constructed by data referred to as lightning monitoring field (lightningmonitoringfield), the meteorological element two dimension real matrix of period corresponding in NCEP again analysis of data, corresponding region is called meteorological element field (meteorologicalelementsfield).

The coupled relation of lightning monitoring field and meteorological element field is the basis analyzing damage to crops caused by thunder feature and weather environment background correlativity, the multinomial variable of SVD method to two fields is adopted to implement different conversions, the new variables obtained is one group of time coefficient item, and this and two fields mutual relationship respectively between time coefficient that is corresponding, that occur in pairs is the correlationship of lightning monitoring field and meteorological element field.

Computing method are summarized as follows:

Association's intersection variance matrix of two fields is C _sz=<SZ ^t>, symbol <> represent and are averaging.By SVD, find two orthogonal linear transformation matrix L and R, make there is maximization covariance between two fields, namely

cov(L ^TS,R ^TZ)＝L ^TC _szR＝MAX

$C_{\mathrm{SZ}}=L\left(\begin{array}{cc}\mathrm{\&Sigma;}& 0\\ 0& 0\end{array}\right)R^T$

$\mathrm{\&Sigma;}=\left(\begin{array}{cccc}{\mathrm{\&sigma;}}_1& 0& ...& 0\\ 0& {\mathrm{\&sigma;}}_2& ...& 0\\ ...& ...& ...& ...\\ 0& 0& ...& {\mathrm{\&sigma;}}_r\end{array}\right)$

Wherein, Σ=diag (σ ₁, σ ₂..., σ _r) (r≤min{x, z}), and σ ₁>=σ ₂>=...>=σ _r> 0.σ _i(i=1,2 ..., r) be called the singular value of C.

If A=L ^ts, B=R ^tz, wherein A is called the S time coefficient matrix of actual measurement field, and B is called the time coefficient matrix of simulation yard Z.The kth column vector l of L and R _kand r _k(k=1,2 ...., r) be called the left and right singular vector of kth, namely kth is to spatial mode.Theoretical by linear algebra, L and R meeting above-mentioned condition can be solved uniquely.

$\begin{array}{cc}L=\left(\begin{array}{cccc}{l}_{11}& l_{12}& ...& l_{1\mathrm{Ns}}\\ l_{21}& l_{22}& ...& l_{2\mathrm{Ns}}\\ ...& ...& ...& ...\\ l_{\mathrm{Ns}1}& l_{\mathrm{Ns}2}& ...& l_{\mathrm{NsNs}}\end{array}\right)& R=\left(\begin{array}{cccc}{r}_{11}& r_{12}& ...& r_{1\mathrm{Nz}}\\ r_{21}& r_{22}& ...& r_{2\mathrm{Nz}}\\ ...& ...& ...& ...\\ r_{\mathrm{Nz}1}& r_{\mathrm{Nz}2}& ...& r_{\mathrm{NzNz}}\end{array}\right)\end{array}$

Step 4: the meteorological element correlationship of region damage to crops caused by thunder feature is analyzed

Before being drawn by SVD analysis, N is to time coefficient (by the large minispread of covariance), the gridded data that ground can be dodged over time and the same period meteorological element change this challenge in time and be reduced to N to the mutual relationship between time coefficient, adopt coefficient R to realize calculating in the present invention.

Front N is to the kth column vector l of time coefficient _kand r _k, k≤N

$R_k=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})\&CenterDot;(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})]}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})}^2\&CenterDot;\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})}^2}}$

Wherein, i=min (N _s, N _z)

Step 5: the meteorological effect factor analysis of correlation test and damage to crops caused by thunder active characteristics

For known one group of related coefficient, the method that T can be adopted to check (formula slightly) determines the conspicuousness of correlationship.Because the historical data of LLS mainly concentrates on the second half year, when considering day Value Data, the vector element number of the time coefficient that SVD obtains can meet the sample hypothesis of T inspection.

Through T inspection, be with all kinds of meteorological element fields of the movable significant correlation of damage to crops caused by thunder the main environment variable affecting damage to crops caused by thunder active characteristics, may be used for the analysis of damage to crops caused by thunder active procedure, damage to crops caused by thunder early warning and other electric system lightning protection engineerings.

Claims (4)

1. the region damage to crops caused by thunder feature correlation analysis method based on LLS, it is characterized in that, described method dodges Monitoring Data according to lightning location system ground and NECP analyzes data again, adopt singular value decomposition method to dodge space characteristics definitely and analyze the correlationship between meteorological background that data extract again, the analysis of the factor of the strong correlation of feasible region damage to crops caused by thunder feature and identification by NECP.

2. a kind of region damage to crops caused by thunder feature correlation analysis method based on LLS according to claim 1, it is characterized in that, described method performing step is as follows:

(1) to LLS lightning monitoring data prediction, LLS monitoring is dodged with obtaining record is processed into 0.01 ° × 0.01 °, 0.05 ° × 0.05 ° and the 0.1 ° × 0.1 ° of three kinds of spatial and temporal resolution CG lightning density of gridding day by day space matrix by gridding method;

(2) again data prediction is analyzed to NECP, the NECP collecting LLS lightning monitoring data history same period analyzes data 0.1 ° × 0.1 ° Grid data again, determine that NECP analyzes data space span again according to LLS monitoring range, the time, span was in accordance with LLS;

(3) LLS lightning monitoring data are carried out and NECP analyzes data coupling relationship analysis again, SVD method is adopted to implement different conversions to lightning monitoring field from the multinomial variable of meteorological element field, the new variables obtained is one group of time coefficient item, and this and two fields mutual relationship respectively between time coefficient that is corresponding, that occur in pairs is the correlationship of lightning monitoring field and meteorological element field;

(4) the meteorological element correlationship analysis of region damage to crops caused by thunder feature is carried out, by SVD analyze draw before N to time coefficient (by the large minispread of covariance), the gridded data that ground can be dodged over time and the same period meteorological element change this challenge in time and be reduced to N to the mutual relationship between time coefficient;

(5) the meteorological effect factor analysis of correlation test and damage to crops caused by thunder active characteristics is carried out, for known one group of related coefficient, the conspicuousness of the method determination correlationship that T can be adopted to check; Because the historical data of LLS mainly concentrates on the second half year, when considering day Value Data, the vector element number of the time coefficient that SVD obtains can meet the sample hypothesis of T inspection.

3. a kind of region damage to crops caused by thunder feature correlation analysis method based on LLS according to claim 2, it is characterized in that, the correlationship computing method of described lightning monitoring field and meteorological element field are as follows:

It is C that variance matrix is intersected in lightning monitoring field and the association of meteorological element field _sz=<SZ ^t>, symbol <> represent and are averaging; By SVD, find two orthogonal linear transformation matrix L and R, make there is maximization covariance between two fields, that is:

cov(L ^TS,R ^TZ)＝L ^TC _szR＝MAX

$C_{\mathrm{SZ}}=L\left(\begin{array}{cc}\mathrm{\&Sigma;}& 0\\ 0& 0\end{array}\right)R^T$

$\mathrm{\&Sigma;}=\left(\begin{array}{cccc}{\mathrm{\&sigma;}}_1& 0& ...& 0\\ 0& {\mathrm{\&sigma;}}_2& ...& 0\\ ...& ...& ...& ...\\ 0& 0& ...& {\mathrm{\&sigma;}}_r\end{array}\right)$

Wherein, Σ=diag (σ ₁, σ ₂..., σ _r) (r≤min{x, z}), and σ ₁>=σ ₂>=...>=σ _r> 0; σ _i(i=1,2 ..., r) be called the singular value of C;

If A=L ^ts, B=R ^tz, wherein A is called the S time coefficient matrix of actual measurement field, and B is called the time coefficient matrix of simulation yard Z; The kth column vector l of L and R _kand r _k(k=1,2 ...., r) be called the left and right singular vector of kth, namely kth is to spatial mode.

4. a kind of region damage to crops caused by thunder feature correlation analysis method based on LLS according to claim 2, is characterized in that, described N realizes being calculated as to the coefficient R of the mutual relationship between time coefficient:

Front N is to the kth column vector l of time coefficient _kand r _k, k≤N

$R_k=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}[(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})\&CenterDot;(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})]}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(l_{\mathrm{ik}}-\stackrel{\&OverBar;}{l})}^2\&CenterDot;\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{(r_{\mathrm{ik}}-\stackrel{\&OverBar;}{r})}^2}}$

Wherein, i=min (N _s, N _z).