CN101794118A - Excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm - Google Patents

Abstract

The invention discloses an excitation system parameter identification method based on system decoupling and a sequence-optimization genetic algorithm, relating to the technical field of excitation system parameter identification in electric power systems and comprising the following steps of: decoupling an excitation system and a generator by utilizing a PMU (Phasor Measuring Unit); determining an excitation system model based on decoupling; and identifying parameters in the excitation system model based on the decoupling by utilizing the sequence-optimization genetic algorithm. The invention decreases the complicated degree of the excitation system parameter identification and reduces errors caused by the accurate model and parameters of the generator, and the identified excitation system parameters are accurate and have a certain confidence.

Description

Excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm

Technical field

The invention belongs to the excitation system parameter identification technique field in the electric system, relate in particular to a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm.

Background technology

Big electrical network, high voltage and big unit epoch have been stepped in electric system, therefore system stability are had higher requirement.In Operation of Electric Systems analysis and control, the generator excitation model is as the important component part of electric system electromechanical transient mathematical model, its model parameter is provided with correctness and the confidence level that correctness directly determines Model for Stability Calculation of Power System, and then influences the formulation and the enforcement of system's operation control measure.For this reason, State Grid Corporation of China's science and technology development planning is pointed out, aspect the systematic parameter modelling technique, " further deepen and study ' four big parameters ' measurement and the analytical technology of (generator, excitation, speed regulator and load), improves its accuracy.”

Measured curve identification gained when at present, the excitation controller parameter mostly is according to the generator zero load.The comparatively ripe algorithm of systematic parameter identification roughly is divided into frequency domain method and time domain method.Though their principles are clear, and are simple and easy to do, can only carry out the parameter identification of linear system, the result of institute's identification can't reflect the non-linear dynamic characteristic of system.For this reason, random optimization algorithms such as genetic algorithm are introduced into and are used for the excitation system parameter identification, preferably resolve the problems referred to above.Yet can the parameter of being surveyed during no-load running be applicable to run with load, is perplexing vast work about electric power person's difficult problem always.The extensive application of PMU (Phasor Measurement Unit, phasor measurement unit) provides possibility for the identification of excitation controller on-line parameter.Yet, obtain excitation parameter according to various existing methods at present, be difficult to obtain quantitative confidence level evaluation.

The present invention has introduced a kind of online measured data of PMU of directly utilizing and has carried out parameters identification method, need not generator is injected noise signal, has reduced the complexity of identification greatly.And, also generator and excitation controller are carried out decoupling zero by the PMU measured data, separately the excitation controller parameter is carried out identification, effectively reduce because of generator model and the inaccurate error of introducing of parameter, the excitation controller parameter that makes identification come out in conjunction with sequence-optimization genetic algorithm is more accurate, and has certain degree of confidence.

Summary of the invention

The objective of the invention is to, a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm is provided, be used for the on-line parameter identification of excitation controller.

Technical scheme is that a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm is characterized in that described method comprises:

---utilize phasor measurement unit PMU is with excitation system and generator decoupling zero;

The excitation system model of---determine based on decoupling zero;

The identification of---utilize sequence-optimization genetic algorithm is based on the excitation system Model parameter of decoupling zero.

Described excitation system model based on decoupling zero is:

$\underset{\mathrm{\α}}{\min }{J}_0\left(\mathrm{\α}\right)={\∫}_0^T{\left|\right|z\left(t\right)-\hat{z}\left(t\right)\left|\right|}^2\mathrm{dt}$

$s.t.{\stackrel{\·}{x}}_{\mathrm{iE}}=f(t,x_{\mathrm{iE}}\left(t\right),x_{\mathrm{iEa}}\left(t\right),y\left(t\right),\mathrm{\α})$

$\hat{z}\left(t\right)=z(t;x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\α})$

Wherein,

Be the measuring amount in the identification after the decoupling zero, adopt the mode of measuring field voltage, exciting current and generator outlet voltage, electric current to obtain identification result.

Describedly utilize the sequence-optimization genetic algorithm identification specifically to comprise based on the excitation system Model parameter of decoupling zero:

Step 1: the inequality constrain condition based on the excitation parameter identification model of decoupling zero is added in the fitness function;

Step 2: determine required calculating total individual number 2n, genetic algorithm each for individual number 2m and algebraically g;

Step 3: the decision variable coding to problem forms chromosome;

Step 4: determine range of variables according to the inequality constrain condition, generate first generation population at random;

Step 5: each individuality to when pre-group, calculate its individual goal functional value and fitness function value F _Fitness

Step 6: the individuality for preceding m-p, carry out interlace operation; Wherein, m is a target function value, and p is an integer, i.e. m/2 or (m+1)/2;

Step 7: the individuality for preceding m-p+1 to m, carry out mutation operation; Real coding is made a variation, form offspring individual;

Step 8: generate m individuality at random, the next generation of adding system;

Step 9: each individuality to when pre-group, calculate its target function value m and fitness function value F _Fitness

Step 10: whether check target function value less than setting value, if less than, then jump to step 5; Otherwise, finish.

Described inequality constrain condition is:

$P={\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">M</figure-callout>}}_1\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}({\left[\max \right\{0,{\mathrm{\&alpha;}}_i-{\mathrm{\&alpha;}}_i^{\mathrm{Max}}\left\}\right]}^2+{\left[\max \right\{0,{\mathrm{\&alpha;}}_i^{\mathrm{Min}}-{\mathrm{\&alpha;}}_i\left\}\right]}^2),$

Wherein, α _iRepresent that calculative each parameter, N represent to have the number of the parameter of bound.

In the described step 2, n=ln (1-P _Se)/ln (1-k%), P _SeBe to realize having at least one to separate and be arranged in solution space all separate the probability that performance accounts for the required stochastic sampling number of times of preceding k%; M gets the value between the 100-1000, algebraically g=n/m.

Described objective function is: J (α).

Described fitness function is: F _Fitness=J (α)+P.

Described interlace operation is specifically: for two parameter alpha ₁, α ₂, producing the random number between [0,1] earlier, the result who obtains then to intersect is:

$\begin{array}{c}{\mathrm{\&alpha;}}_1^{\mathrm{new}}=r{\mathrm{\&alpha;}}_1+(1-r){\mathrm{\&alpha;}}_2\\ {\mathrm{\&alpha;}}_2^{\mathrm{new}}=(1-r){\mathrm{\&alpha;}}_1+r{\mathrm{\&alpha;}}_2\end{array}.$

Described mutation operation is specifically: if r ₁Be random number between [0,1], definition parent α _kVariation be:

${{\mathrm{\&alpha;}}^{\&prime;}}_k=\left\{\begin{array}{cc}{\mathrm{\&alpha;}}_k+(\mathrm{UB}-{\mathrm{\&alpha;}}_k)f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1<0.5\\ {\mathrm{\&alpha;}}_k-({\mathrm{\&alpha;}}_k-\mathrm{LB})f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1\&GreaterEqual;0.5\end{array}\right.,$

Wherein, UB and LB are respectively variable α _kBound, t is for when the algebraically of evolution.F (k) is: $f\left(k\right)=r\&CenterDot;{(1-\frac{k}{n})}^3.$

The present invention directly utilizes the online measured data of PMU to carry out parameter identification, has reduced the complexity of identification; By the PMU measured data generator and excitation controller are carried out decoupling zero, reduced because of generator model and the inaccurate error of introducing of parameter, carry out parameter recognition in conjunction with sequence-optimization genetic algorithm, make that the excitation controller parameter of identification is more accurate, and have certain degree of confidence.

Description of drawings

Fig. 1 is a model synoptic diagram before generator and the electrical network decoupling zero;

Fig. 2 is based on the excitation parameter identification model synoptic diagram of decoupling zero;

Fig. 3 is the ultimate principle figure of decoupling zero excitation parameter identification;

Fig. 4 utilizes sequence-optimization genetic algorithm to carry out excitation system parameter identification schematic flow sheet;

Fig. 5 is the omission that provides of the embodiment of the invention crosses and encourages restriction and low FV model synoptic diagram of encouraging restriction;

Fig. 6 is the set end voltage synoptic diagram that embodiment of the invention actual measurement obtains;

Fig. 7 is that embodiment of the invention actual measurement obtains field voltage response synoptic diagram;

Fig. 8 be under embodiment of the invention excitation system real response and the identified parameters response ratio than result schematic diagram.

Embodiment

Below in conjunction with accompanying drawing, preferred embodiment is elaborated.Should be emphasized that following explanation only is exemplary, rather than in order to limit the scope of the invention and to use.

Fig. 1 is a model synoptic diagram before generator and the electrical network decoupling zero.Among Fig. 1, G is a generator, E _FdBe the field voltage of generator, I _t, θ _IBe the electric current and the phase angle of machine end, U _t, θ _uVoltage and phase angle thereof for the machine end.As shown in Figure 1, the closed-loop system that contains excitation can be described below:

${\stackrel{\&CenterDot;}{x}}_{\mathrm{iE}}=f_1(x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})$

${\stackrel{\&CenterDot;}{x}}_{\mathrm{iEa}}=f_2(x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})---\left(1\right)$

0＝g(x _iE，x _iEa，y，α)

Wherein, x _IEBe the state variable of the excitation system that needs identification, x _IEaBe other state variable, y is the algebraically variable in the system, and α is the parameter that needs identification.

The purpose of excitation parameter identification is exactly according to the measuring amount that obtains

With theoretical value z (t)=z (t; x _IE, x _IEa, y, α) difference, (wherein

With x _IE, x _IEa, y is relevant) directly obtain parameter alpha.This problem can be expressed as:

$\underset{\mathrm{\&alpha;}}{\min }{J}_0\left(\mathrm{\&alpha;}\right)={\&Integral;}_0^T{\left|\right|z\left(t\right)-\hat{z}\left(t\right)\left|\right|}^2\mathrm{dt}$

${\stackrel{\&CenterDot;}{x}}_{\mathrm{iE}}=f_1(x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})$

$s.t.{\stackrel{\&CenterDot;}{x}}_{\mathrm{iEa}}=f_2(x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})---\left(2\right)$

0＝g(x _iE，x _iEa，y，α)

z(t)＝z(t；x _iE，x _iEa，y，α)

Measuring amount in the identification wherein Range of choice very big, can select local signal, also can select the signal in a distant place.

The pairing block mold of excitation parameter identification, its dimension is very high, for the generator that contains more than 100, does not consider dynamic elements such as other loads and speed regulator, and its dimension is greater than 600 dimensions, and this causes identification institute elapsed time excessive.In view of this, the present invention adopts the decoupling zero scheme, and generator is obtained set end voltage and electric current by phasor measurement unit PMU.

Fig. 2 is based on the excitation parameter identification model synoptic diagram of decoupling zero.In Fig. 2, utilize PMU to obtain field voltage, can be with excitation system and generator decoupling zero.Wherein, U _RefBe excitation system reference voltage, U _SFor generator PSS output signal, ∑ are the diagrammatic representation of " summation " sign of operation.Under decoupling zero, excitation system can be described as:

${\stackrel{\&CenterDot;}{x}}_{\mathrm{iE}}=f_1(t,x_{\mathrm{iE}}\left(t\right),x_{\mathrm{iEa}}\left(t\right),y\left(t\right),\mathrm{\&alpha;})---\left(3\right)$

X wherein _IEa(t), y (t) can obtain by measuring directly.

Compare with above-mentioned block mold (formula (2)), this model does not need to consider the dynamic of other generators, only needs to measure associated state and gets final product.For any excitation system, only need to measure the generator outlet voltage and current, can be with excitation and system decoupling.After the decoupling zero, be example, only consider the quadravalence model, simplified calculating greatly with the FV model.

Excitation parameter identification based on decoupling zero can be described below

$\underset{\mathrm{\&alpha;}}{\min }{J}_0\left(\mathrm{\&alpha;}\right)={\&Integral;}_0^T{\left|\right|z\left(t\right)-\hat{z}\left(t\right)\left|\right|}^2\mathrm{dt}$

$s.t.{\stackrel{\&CenterDot;}{x}}_{\mathrm{iE}}=f(t,x_{\mathrm{iE}}\left(t\right),x_{\mathrm{iEa}}\left(t\right),y\left(t\right),\mathrm{\&alpha;})---\left(4\right)$

$\hat{z}\left(t\right)=z(t;x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})$

Compare measuring amount in the identification after the decoupling zero with block mold

Range of choice diminish, its variable can only be selected to depend on interface voltage and electric current, and the amount of each state of excitation.Therefore, only need to adopt and measure field voltage, exciting current, the mode of generator outlet voltage and current can obtain identification result.

Fig. 3 is the ultimate principle figure of decoupling zero excitation parameter identification.At the excitation parameter identification model after Fig. 2 decoupling zero, can adopt discrimination method shown in Figure 3 to carry out parameter identification.I.e. response by input-output system under the different parameters relatively, with response near the parameter of real system as identification result.Particularly, the input quantity input system that is obtained during exactly with disturbance compares with the output of generator model under the output of real system and the preset parameter, seeks the parameter near real response.In the parameter identification process, adopt sequence-optimization genetic algorithm.

Preface is optimized concretism: establish search volume S and have N _SIndividual feasible solution, G is the set of separating of k% before being accounted for by performance, size is N _G, k%=N is promptly arranged _G/ N _SFor N _SVery big optimization problem can not be to wherein each be separated and carries out emulation, obtain optimum solution and be unusual difficulty.Therefore will have to take the second best here, it is with P that target is softened _SeProbability find one to separate in G at least.If the hits that reaching this target needs is n, then easily know P _Se=1-(1-k%) ⁿThereby, have

n＝ln(1-P _se)/ln(1-k％) (5)

Outstanding versatility of separating with genetic algorithm in conjunction with preface optimization provides certain confidence level can obtain sequence-optimization genetic algorithm.Fig. 4 utilizes sequence-optimization genetic algorithm to carry out excitation system parameter identification schematic flow sheet, wherein:

Step 1: the inequality constrain condition based on the excitation parameter identification model of decoupling zero is added in the fitness function.

The inequality constrain condition is:

$P={\mathrm{<figure-callout\; id="1"\; label="M"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">M</figure-callout>}}_1\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}({\left[\max \right\{0,{\mathrm{\&alpha;}}_i-{\mathrm{\&alpha;}}_i^{\mathrm{Max}}\left\}\right]}^2+{\left[\max \right\{0,{\mathrm{\&alpha;}}_i^{\mathrm{Min}}-{\mathrm{\&alpha;}}_i\left\}\right]}^2),$

Wherein, α _iRepresent that calculative each parameter, N represent to have the number of the parameter of bound.

Step 2: the number of individuals 2m and the algebraically g in each generation in determine to acquire a certain degree individual number 2n of needed calculating and the genetic algorithm.

N=ln (1-P _Se)/ln (1-k%), P _SeBe to realize having at least one to separate and be arranged in solution space all separate the probability that performance accounts for the required stochastic sampling number of times of preceding k%.The individual number in per generation is made as 2m in the genetic algorithm, suggestion m～[100,1000].Then the evolutionary generation of whole genetic algorithm is g=2m/2n=m/n.

Step 3: the decision variable coding to problem forms chromosome.

Step 4: determine range of variables according to the inequality constrain condition, generate first generation population at random.

Step 5: each individuality to when pre-group, calculate its target function value and fitness function value F _FitnessTarget function value is: J (α)

Fitness function is: F _Fitness=J (α)+P.

Step 6: the individuality for preceding m-p, carry out interlace operation; Wherein, m is a target function value, and p is an integer, i.e. m/2 or (m+1)/2.

Interlace operation is specifically: for two parameter alpha ₁, α ₂, producing the random number between [0,1] earlier, the result who obtains then to intersect is:

$\begin{array}{c}{\mathrm{\&alpha;}}_1^{\mathrm{new}}=r{\mathrm{\&alpha;}}_1+(1-r){\mathrm{\&alpha;}}_2\\ {\mathrm{\&alpha;}}_2^{\mathrm{new}}=(1-r){\mathrm{\&alpha;}}_1+r{\mathrm{\&alpha;}}_2\end{array}.$

Step 7:, carry out mutation operation for preceding m-p+1 to m individuality; Real coding is made a variation, form offspring individual.

Mutation operation is specifically: if r ₁Be random number between [0,1], definition parent α _kVariation be:

${{\mathrm{\&alpha;}}^{\&prime;}}_k=\left\{\begin{array}{cc}{\mathrm{\&alpha;}}_k+(\mathrm{UB}-{\mathrm{\&alpha;}}_k)f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1<0.5\\ {\mathrm{\&alpha;}}_k-({\mathrm{\&alpha;}}_k-\mathrm{LB})f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1\&GreaterEqual;0.5\end{array}\right.,$

Wherein, UB and LB are respectively variable α _kBound, t is for when the algebraically of evolution.F (k) is: $f\left(k\right)=r\&CenterDot;{(1-\frac{k}{n})}^3.$

Step 8: generate m individuality at random, the next generation of adding system.

Step 9: each individuality to when pre-group, calculate its target function value and fitness function value F _Fitness

Step 10: whether check target function value less than setting value, if less than, then jump to step 5; Otherwise, finish.

Embodiment

Fig. 5 is the omission that provides of the embodiment of the invention crosses and encourages restriction and low FV model synoptic diagram of encouraging restriction.This figure is example with certain power plant from the excitation system of shunt dynamo.Wherein import V _ERRBe the difference of generator terminal voltage and reference voltage, output E _FDBe generator excitation voltage, K _VBe the proportional integral or the pure integration selection factor, I _FDBe exciter current of generator.For FV model, parameter T ₁, T ₂, T ₃, T ₄, K _A, T _A, K _F, T _FFor needing the parameter of identification.That is α=[T, ₁, T ₂, T ₃, T ₄, K _a, T _a] ^TFor above-mentioned model, under decoupling zero, measuring amount is the extreme voltage in three phases of generator, electric current, and the field voltage of generator and exciting current.

Fig. 6 is the set end voltage synoptic diagram that embodiment of the invention actual measurement obtains.

Fig. 7 is that embodiment of the invention actual measurement obtains field voltage response synoptic diagram.

The purpose of excitation parameter identification is in this example, decision α=[T ₁, T ₂, T ₃, T ₄, K _a, T _a] ^T, make to respond for the virtual field voltage that set end voltage shown in Figure 6 changes at the FV model, approach the response of actual field voltage shown in Figure 7 as far as possible.

Utilize sequence-optimization genetic algorithm to carry out excitation system parameter alpha=[T ₁, T ₂, T ₃, T ₄, K _a, T _a] ^TIdentification, its process is:

A: generate first generation individuality at random: repeat n time.In the present embodiment, get n=ln (1-P _Se)/ln (1-k%), m=120 produces a six-vector α=[T at random ₁, T ₂, T ₃, T ₄, K _a, T _a] ^T

B: calculate identified parameters α=[T ₁, T ₂, T ₃, T ₄, K _a, T _a] ^TWhen getting 2m value respectively, the value of the objective function that each is individual.

C: 2m identified parameters is worth pairing target function value ordering, chooses the less m of a target function value value.

D: carry out interlace operation, for two parameter alpha ₁, α ₂, producing the random number between [0,1] earlier, the result who obtains then to intersect is:

$\begin{array}{c}{\mathrm{\&alpha;}}_1^{\mathrm{new}}=r{\mathrm{\&alpha;}}_1+(1-r){\mathrm{\&alpha;}}_2\\ {\mathrm{\&alpha;}}_2^{\mathrm{new}}=(1-r){\mathrm{\&alpha;}}_1+r{\mathrm{\&alpha;}}_2\end{array}.$

E: carry out mutation operation, if r ₁Be random number between [0,1], definition parent α _kVariation be:

${{\mathrm{\&alpha;}}^{\&prime;}}_k=\left\{\begin{array}{cc}{\mathrm{\&alpha;}}_k+(\mathrm{UB}-{\mathrm{\&alpha;}}_k)f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1<0.5\\ {\mathrm{\&alpha;}}_k-({\mathrm{\&alpha;}}_k-\mathrm{LB})f\left(k\right)& {\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000038096300021.png,HSA00000038096300031.png"\; state="\{\{state\}\}">r</figure-callout>}}_1\&GreaterEqual;0.5\end{array}\right.$

Here UB and LB are respectively variable α _kBound, t is for when the algebraically of evolution.f(k)： $f\left(k\right)=r\&CenterDot;{(1-\frac{k}{n})}^3.$

F: at random the follow-on m of adding system of Sheng Chenging individual.Be that Repeated m is inferior, produce a six-vector α=[T at random ₁, T ₂, T ₃, T ₄, K _a, T _a] ^T

G: carry out fitness and calculate.To each individuality of current colony, calculate its target function value and fitness function value F _Fitness

H: check whether satisfy convergence criterion, whether promptly check target function value,, jumped to for second step and continue to calculate as not satisfying less than setting value; If satisfy, then finish.

Through said process, can obtain based on the virtual output response of FV model and the comparative result of real response data.Fig. 8 be under embodiment of the invention excitation system real response and the identified parameters response ratio than result schematic diagram.It is pointed out that said method is more conservative, can guarantee to obtain to have the suboptimal solution of certain confidence level.In fact, because total intersection factor of genetic algorithm, the actual confidence level of separating that obtains will be much larger than the confidence level of setting.

The present invention directly utilizes the online measured data of PMU to carry out parameter identification, need not generator is injected noise signal, has reduced the complexity of identification greatly; And, also generator and excitation controller are carried out decoupling zero by the PMU measured data, separately the excitation controller parameter is carried out identification, effectively reduce because of generator model and the inaccurate error of introducing of parameter, the excitation controller parameter that makes identification come out in conjunction with sequence-optimization genetic algorithm is more accurate, and has certain degree of confidence.

The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (9)

1. excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm is characterized in that described method comprises:

---utilize phasor measurement unit PMU is with excitation system and generator decoupling zero;

The excitation system model of---determine based on decoupling zero;

The identification of---utilize sequence-optimization genetic algorithm is based on the excitation system Model parameter of decoupling zero.

2. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 1 is characterized in that described excitation system model based on decoupling zero is:

$\underset{\mathrm{\&alpha;}}{\min }{J}_0\left(\mathrm{\&alpha;}\right)={\&Integral;}_0^T{\left|\right|z\left(t\right)-\hat{z}\left(t\right)\left|\right|}^2\mathrm{dt}$

$s.t.{\stackrel{\&CenterDot;}{x}}_{\mathrm{iE}}=f(t,x_{\mathrm{iE}}\left(t\right),x_{\mathrm{iEa}}\left(t\right),y\left(t\right),\mathrm{\&alpha;})$

$\hat{z}\left(t\right)=z(t;x_{\mathrm{iE}},x_{\mathrm{iEa}},y,\mathrm{\&alpha;})$

Wherein, Be the measuring amount in the identification after the decoupling zero, adopt the mode of measuring field voltage, exciting current and generator outlet voltage, electric current to obtain identification result.

3. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 2 is characterized in that describedly utilizing the sequence-optimization genetic algorithm identification specifically to comprise based on the excitation system Model parameter of decoupling zero:

Step 1: the inequality constrain condition based on the excitation parameter identification model of decoupling zero is added in the fitness function;

Step 2: determine required calculating total individual number 2n, genetic algorithm each for individual number 2m and algebraically g;

Step 3: the decision variable coding to problem forms chromosome;

Step 4: determine range of variables according to the inequality constrain condition, generate first generation population at random;

Step 5: each individuality to when pre-group, calculate its individual goal functional value and fitness function value F _Fitness

Step 6: the individuality for preceding m-p, carry out interlace operation; Wherein, m is a target function value, and p is an integer, i.e. m/2 or (m+1)/2;

Step 7: the individuality for preceding m-p+1 to m, carry out mutation operation; Real coding is made a variation, form offspring individual;

Step 8: generate m individuality at random, the next generation of adding system;

Step 9: each individuality to when pre-group, calculate its target function value m and fitness function value F _Fitness

Step 10: whether check target function value less than setting value, if less than, then jump to step 5; Otherwise, finish.

4. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that described inequality constrain condition is:

$P=M_1\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}({\left[\max \right\{0,{\mathrm{\&alpha;}}_i-{\mathrm{\&alpha;}}_i^{\mathrm{Max}}\left\}\right]}^2+{\left[\max \right\{0,{\mathrm{\&alpha;}}_i^{\mathrm{Min}}-{\mathrm{\&alpha;}}_i\}}^2),$

Wherein, α _iRepresent that calculative each parameter, N represent to have the number of the parameter of bound.

5. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that in the described step 2 n=ln (1-P _Se)/ln (1-k%), P _SeBe to realize having at least one to separate and be arranged in solution space all separate the probability that performance accounts for the required stochastic sampling number of times of preceding k%; M gets the value between the 100-1000, algebraically g=n/m.

6. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that described objective function is: J (α).

7. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that described fitness function is: F _Fitness=J (α)+P.

8. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that described interlace operation specifically: for two parameter alpha ₁, α ₂, producing the random number between [0,1] earlier, the result who obtains then to intersect is:

${\mathrm{\&alpha;}}_1^{\mathrm{new}}=r{\mathrm{\&alpha;}}_1+(1-r){\mathrm{\&alpha;}}_2.$

${\mathrm{\&alpha;}}_2^{\mathrm{new}}=(1-r){\mathrm{\&alpha;}}_1+r{\mathrm{\&alpha;}}_2$

9. a kind of excitation system parameter identification method based on system decoupling and sequence-optimization genetic algorithm according to claim 3 is characterized in that described mutation operation specifically: if r ₁Be random number between [0,1], definition parent α _kVariation be:

${{\mathrm{\&alpha;}}^{\&prime;}}_k=\left\{\begin{array}{cc}{\mathrm{\&alpha;}}_k+(\mathrm{UB}-{\mathrm{\&alpha;}}_k)f\left(k\right)& r_1<0.5\\ {\mathrm{\&alpha;}}_k-({\mathrm{\&alpha;}}_k-\mathrm{LB})f\left(k\right)& r_1\&GreaterEqual;0.5\end{array}\right.,$

Wherein, UB and LB are respectively variable α _kBound, t is for when the algebraically of evolution.F (k) is:

$f\left(k\right)=r\&CenterDot;{(1-\frac{k}{n})}^3.$

Images