CN102280877A - Method for identifying parameter of poor branch of power system through a plurality of measured sections - Google Patents

Abstract

The invention relates to a method for identifying the parameter of a poor branch of a power system based on a multi-measured section Lagrange multiplier method, and belongs to the technical fields of power system scheduling automation and power grid simulation. The method comprises the following steps of: reading in a plurality of measured sections from a historical database one by one; establishing a state estimation model of each section, solving, and calculating the regularization Lagrange multiplier of each section; and accumulating the regularization Lagrange multiplier vector of each section to obtain a suspicious degree index of a branch parameter, namely the regularization Lagrange multiplier vectors of the plurality of sections. The method is convenient to implement; meanwhile, the accuracy of the original poor parameter identification program can be greatly improved.

Description

A kind of volume is surveyed the bad branch road parameter identification method of electric power system of section

Technical field

The invention belongs to dispatching automation of electric power systems and grid simulation technical field, particularly a kind of bad branch road parameter identification method of electric power system of surveying the section method of Lagrange multipliers based on volume.

Background technology

What guarantee grid branch parameter (comprising resistance, reactance, direct-to-ground capacitance etc.) accurately is major issue in the electric power system modeling.Yet in the electric power system modeling process of reality, may there be the inaccurate wrong branch road parameter of some numerical value, these wrong branch road parameters can have a strong impact on precision and the credibility that power system analysis is calculated, and therefore just need the discrimination method of the bad branch road parameter of research electric power system.

The bad branch road parameter of electric power system generally acknowledges that relatively the effective identification method is a method of Lagrange multipliers at present.The main flow process of its method is to be 0 to join in the state estimation Optimization Model as equality constraint with electric power system branch road parameter error, promptly sets up following Power system state estimation Optimization Model:

$\min J\left(x\right)=\frac{1}{2}{r}^T\mathrm{Wr}$

s.t.c(x，p _e)＝0

p _e＝0

Wherein x is a system state variables, comprises the amplitude and the phase angle of node voltage; R is for measuring residual vector, r=z-h (x).Z and h are respectively the real-time measurement values and measure function.W is for measuring weight diagonal matrix, p _eBe electric power system branch road parameter error vector, c (x, p _eEquality constraint is injected in)=0 expression zero.Above state estimation Optimization Model is found the solution, can be obtained the state estimation result

And corresponding Calculate the covariance matrix of residual error afterwards respectively, the covariance matrix of the Lagrange multiplier vector sum Lagrange multiplier of branch road parameter correspondence.Wherein covariance matrix Cov (r) computing formula of residual error is:

$\mathrm{Cov}\left(r\right)=W^{-1}-H_x{\left(H_x^{T}W{H}_x\right)}^{-1}{H}_x^T$

H wherein _xFor measuring the Jacobian matrix to state variable, subscript T represents transposition.

The computing formula of Lagrange multiplier is:

$\mathrm{\λ}=H_p^T\mathrm{Wr}$

H wherein _pFor measuring Jacobian matrix to the branch road parameter.

The computing formula of Lagrange multiplier covariance matrix Cov (λ) is:

$\mathrm{Cov}\left(\mathrm{\λ}\right)=H_P^T\mathrm{WCov}\left(r\right)W{H}_P$

Utilize above result of calculation to calculate regularization residual vector r afterwards _NRegularization Lagrange multiplier vector λ _N, computing formula is:

$r_{\mathrm{Ni}}=\frac{{\hat{r}}_i}{\sqrt{\mathrm{Cov}{\left(r\right)}_{\mathrm{ii}}}}$

${\mathrm{\λ}}_{\mathrm{Ni}}=\frac{{\mathrm{\λ}}_i}{\sqrt{\mathrm{Cov}{\left(\mathrm{\λ}\right)}_{\mathrm{ii}}}}$

And obtain r _NAnd λ _NIn maximum, if max (r _N, λ _N)＜c then thinks not have bad remote measurement and bad grid branch parameter, calculates and finishes; Wherein c is the artificial threshold value of setting, and generally gets 3.0.Otherwise, if max (r _N)＞max (λ _N), then reject max (r _N) corresponding measurement, carry out aforementioned calculation again; If max (r _N)＜max (λ _N), then think max (λ _N) corresponding branch road parameter is wrong, utilizes the branch road parameter of electric power system branch road method for parameter estimation correction correspondence, carries out aforementioned calculation afterwards again.

There are two important deficiencies in above method flow, and at first, present method of Lagrange multipliers all only utilizes the measurement information of discontinuity surface when single, promptly single measurement section; In single measurement section, very limited with the measurement number of branch road parameter strong correlation, these limited measurements can't embody the expectation of branch road parameter on statistical significance-be the actual value of parameter well.Secondly, above method flow carries out the identification of bad remote measurement and bad branch road parameter simultaneously, need carry out finding the solution of state estimation model repeatedly, and amount of calculation is very big.Therefore the state estimation result who needs research how to go out based on the state estimation computed in software of present commerce uses a plurality of measurement sections to pick out wrong branch road parameter in the electrical network easily.

Summary of the invention

The objective of the invention is for overcoming the weak point of prior art, a kind of bad branch road parameter identification method of electric power system of surveying the section method of Lagrange multipliers based on volume is proposed, this method can obtain more credible more rational identification result, and has very strong on-the-spot practical ability.

A kind of bad branch road parameter identification method of electric power system based on volume survey section method of Lagrange multipliers that the present invention proposes is characterized in that this method comprises: read in a plurality of measurement sections from historical data base one by one; Set up the state estimation model of each section and find the solution, calculate the regularization Lagrange multiplier of each section; And the regularization Lagrange multiplier vector of each section added up, obtain the suspicious level index of this branch road parameter of regularization Lagrange multiplier vector of multibreak.

This method specifically may further comprise the steps:

(1) volume of each the bar branch road parameter in the initialization electric power system is surveyed section regularization Lagrange multiplier

, with the suspicious level index of regularization Lagrange multiplier vector as this branch road parameter, initialization measures section sequence number m=1, and sets the measurement section sum m that participates in parameter identification _∑, common desirable m _∑=96;

(2) from the historical database server of electric power system, read in m and measure section, and set up electric power system least square state estimation model;

$\min J_m\left(x_m\right)=\frac{1}{2}{r}^T\mathrm{Wr}$

s.t c _m(x _m)＝0

Wherein, subscript m represents to measure the section sequence number, and r is for measuring residual vector, and subscript T represents transposition, r=z _m-h _m(x _m); z _mBe m real-time measurement values vector that measures in the section, h _m(x _m) be m measurement equation vector that measures in the section, x _mBe m electric network state variable that measures section, this variable comprises the voltage magnitude and the phase angle of all nodes; W is for measuring weight diagonal matrix, c _m(x _m)=the 0th do not articulate zero injection equality constraint of the load and the node of generator;

(3) the least square state estimation model of step (2) is found the solution, obtain m electric network state variable x that measures section _mEstimated value

(4) calculate m Lagrange multiplier vector λ that measures section _m, λ _mBe to calculate the suspicious level index of branch road parameter

A necessary intermediate variable, its computing formula is:

${\mathrm{\λ}}_m=H_P^{T}W\hat{r}$

H wherein _PFor measuring Jacobian matrix to the branch road parameter;

For

The time residual vector;

(5) calculate Lagrange multiplier vector λ _mCorresponding covariance matrix Cov (λ), computing formula is:

$\mathrm{Cov}\left(\mathrm{\λ}\right)=H_P^T\mathrm{WCov}\left(r\right)W{H}_P$

Wherein Cov (r) is for measuring the covariance matrix of residual error, and its computing formula is:

$\mathrm{Cov}\left(r\right)=W^{-1}-H_x{\left(H_x^{T}W{H}_x\right)}^{-1}{H}_x^T$

H wherein _xFor measuring the Jacobian matrix to state variable, subscript T represents transposition.

(6), obtain the regularization Lagrange multiplier vector of m section with the Lagrange multiplier regularization

, the regularization Lagrange multiplier of k branch road parameter correspondence

Computing formula be:

${\mathrm{\λ}}_{m,k}^N=\frac{{\mathrm{\λ}}_{m,k}}{\sqrt{{\mathrm{Cov}\left(\mathrm{\λ}\right)}_{\mathrm{kk}}}}$

λ wherein _{M, k}Be the Lagrange multiplier of k branch road parameter correspondence of m section, Cov (λ) _KkK diagonal element for Lagrange multiplier covariance matrix Cov (λ);

(7) m regularization Lagrange multiplier that measures section is added to multibreak regularization Lagrange multiplier

On, promptly

(8) if m=m _∑, then carry out step (9); If m＜m _∑, then make m=m+1, return step (2)

(9) for i branch road parameter, if

Think that then corresponding branch road parameter is suspicious branch road parameter.Wherein ε is the artificial suspicious branch road parameter threshold value of setting, common desirable 3.0.

Advantage of the present invention is:

1, the method for Lagrange multipliers parameter identification of single section has stronger theoretical foundation and practical experience, and multibreak method of Lagrange multipliers also has very strong on-the-spot practical ability as an improvement of the method for Lagrange multipliers of single section.

2, multibreak method of Lagrange multipliers can overcome the problem of the method for Lagrange multipliers measurement redundancy deficiency of single section, can obtain than the more credible more rational identification result of the method for Lagrange multipliers parameter identification method of single section.

3, the branch road parameter identification method of this patent directly carries out based on the state estimation result, has avoided repeat mode to estimate the amount of calculation problem of bringing, and has very high computational efficiency.

Embodiment

A kind of bad branch road parameter identification method of electric power system based on volume survey section method of Lagrange multipliers that the present invention proposes is described in detail as follows in conjunction with the embodiments:

A kind of bad branch road parameter identification method of electric power system based on volume survey section method of Lagrange multipliers that the present invention proposes is characterized in that this method may further comprise the steps:

(1) different with traditional Lagrange multiplier branch road parameter identification method of introducing in the background technology, the method for this patent adopts a plurality of measurement sections to carry out the identification of bad branch road parameter.At first the volume of each the bar branch road parameter in the initialization electric power system is surveyed section regularization Lagrange multiplier

, with the suspicious level index of regularization Lagrange multiplier vector as this branch road parameter, initialization measures section sequence number m=1, and sets the measurement section sum m that participates in parameter identification _∑, common desirable m _∑=96;

(2) from the historical database server of electric power system, read in m and measure section, and set up electric power system least square state estimation model;

$\min J_m\left(x_m\right)=\frac{1}{2}{r}^T\mathrm{Wr}$

s.t c _m(x _m)＝0

Wherein, subscript m represents to measure the section sequence number, and r is for measuring residual vector, and subscript T represents transposition, r=z _m-h _m(x _m); z _mBe m real-time measurement values vector that measures in the section, h _m(x _m) be m measurement equation vector that measures in the section, x _mBe m electric network state variable that measures section, this variable comprises the voltage magnitude and the phase angle of all nodes; W is for measuring weight diagonal matrix, c _m(x _m)=the 0th do not articulate zero injection equality constraint of the load and the node of generator;

(3) the least square state estimation model of step (2) is found the solution, obtain m electric network state variable x that measures section _mEstimated value

(4) calculate m Lagrange multiplier vector λ that measures section _m, λ _mBe to calculate the suspicious level index of branch road parameter

A necessary intermediate variable, its computing formula is:

${\mathrm{\λ}}_m=H_P^{T}W\hat{r}$

H wherein _PFor measuring Jacobian matrix to the branch road parameter;

For

The time residual vector;

As can be seen, traditional Lagrange multiplier branch road parameter identification method of introducing in this paper method and the background technology is different, it is not the identification of carrying out bad remote measurement and bad branch road parameter simultaneously, but carry out bad remote measurement identification earlier, the bad branch road parameter of identification more afterwards, the state estimation of so just having avoided repeating is calculated, and has improved computational efficiency.

(5) calculate Lagrange multiplier vector λ _mCorresponding covariance matrix Cov (λ), computing formula is:

$\mathrm{Cov}\left(\mathrm{\λ}\right)=H_P^T\mathrm{WCov}\left(r\right)W{H}_P$

Wherein Cov (r) is for measuring the covariance matrix of residual error, and its computing formula is:

$\mathrm{Cov}\left(r\right)=W^{-1}-H_x{\left(H_x^{T}W{H}_x\right)}^{-1}{H}_x^T$

H wherein _xFor measuring the Jacobian matrix to state variable, subscript T represents transposition.

(6), obtain the regularization Lagrange multiplier vector of m section with the Lagrange multiplier regularization

, the regularization Lagrange multiplier of k branch road parameter correspondence

Computing formula be:

${\mathrm{\λ}}_{m,k}^N=\frac{{\mathrm{\λ}}_{m,k}}{\sqrt{{\mathrm{Cov}\left(\mathrm{\λ}\right)}_{\mathrm{kk}}}}$

λ wherein _{M, k}Be the Lagrange multiplier of k branch road parameter correspondence of m section, Cov (λ) _KkK diagonal element for Lagrange multiplier covariance matrix Cov (λ);

(7) m regularization Lagrange multiplier that measures section is added to multibreak regularization Lagrange multiplier

On, promptly

(8) if m=m _∑, then carry out step (9); If m＜m _∑, then make m=m+1, return step (2)

(9) for i branch road parameter, if Think that then corresponding branch road parameter is suspicious branch road parameter.Wherein ε is the artificial suspicious branch road parameter threshold value of setting, common desirable 3.0.

Claims (2)

1. the bad branch road parameter identification method of electric power system based on volume survey section method of Lagrange multipliers is characterized in that this method comprises: read in a plurality of measurement sections from historical data base one by one; Set up the state estimation model of each section and find the solution, calculate the regularization Lagrange multiplier of each section; And the regularization Lagrange multiplier vector of each section added up, obtain the suspicious level index of this branch road parameter of regularization Lagrange multiplier vector of multibreak.

2. method according to claim 1 is characterized in that this method specifically may further comprise the steps:

(1) volume of each the bar branch road parameter in the initialization electric power system is surveyed section regularization Lagrange multiplier

, with the suspicious level index of regularization Lagrange multiplier vector as this branch road parameter, initialization measures section sequence number m=1, and sets the measurement section sum m that participates in parameter identification _∑, common desirable m _∑=96;

(2) from the historical database server of electric power system, read in m and measure section, and set up electric power system least square state estimation model;

$\min J_m\left(x_m\right)=\frac{1}{2}{r}^T\mathrm{Wr}$

s.t c _m(x _m)＝0

Wherein, subscript m represents to measure the section sequence number, and r is for measuring residual vector, and subscript T represents transposition, r=z _m-h _m(x _m); z _mBe m real-time measurement values vector that measures in the section, h _m(x _m) be m measurement equation vector that measures in the section, x _mBe m electric network state variable that measures section, this variable comprises the voltage magnitude and the phase angle of all nodes; W is for measuring weight diagonal matrix, c _m(x _m)=the 0th do not articulate zero injection equality constraint of the load and the node of generator;

(3) the least square state estimation model of step (2) is found the solution, obtain m electric network state variable x that measures section _mEstimated value

(4) calculate m Lagrange multiplier vector λ that measures section _m, λ _mBe to calculate the suspicious level index of branch road parameter A necessary intermediate variable, its computing formula is:

${\mathrm{\λ}}_m=H_P^{T}W\hat{r}$

H wherein _PFor measuring Jacobian matrix to the branch road parameter;

For

The time residual vector;

(5) calculate Lagrange multiplier vector λ _mCorresponding covariance matrix Cov (λ), computing formula is:

$\mathrm{Cov}\left(\mathrm{\λ}\right)=H_P^T\mathrm{WCov}\left(r\right)W{H}_P$

Wherein Cov (r) is for measuring the covariance matrix of residual error, and its computing formula is:

$\mathrm{Cov}\left(r\right)=W^{-1}-H_x{\left(H_x^{T}W{H}_x\right)}^{-1}{H}_x^T$

H wherein _xFor measuring the Jacobian matrix to state variable, subscript T represents transposition.

(6), obtain the regularization Lagrange multiplier vector of m section with the Lagrange multiplier regularization

, the regularization Lagrange multiplier of k branch road parameter correspondence

Computing formula be:

${\mathrm{\λ}}_{m,k}^N=\frac{{\mathrm{\λ}}_{m,k}}{\sqrt{{\mathrm{Cov}\left(\mathrm{\λ}\right)}_{\mathrm{kk}}}}$

λ wherein _{M, k}Be the Lagrange multiplier of k branch road parameter correspondence of m section, Cov (λ) _KkK diagonal element for Lagrange multiplier covariance matrix Cov (λ);

(7) m regularization Lagrange multiplier that measures section is added to multibreak regularization Lagrange multiplier

On, promptly

(8) if m=m _∑, then carry out step (9); If m＜m _∑, then make m=m+1, return step (2)

(9) for i branch road parameter, if

Think that then corresponding branch road parameter is suspicious branch road parameter.Wherein ε is the artificial suspicious branch road parameter threshold value of setting, common desirable 3.0.