CN103279803A - Load modeling method and system based on comprehensive information theory and modern interior point theory - Google Patents

Abstract

The invention provides a load modeling method and a loading modeling system based on a comprehensive information theory and a modern interior point theory. The method comprises the following steps that the load data of a plurality of transformer substations is counted; the load characteristic index of each transformer substation is calculated according to the counted load data; each transformer substation is classified by a clustering method according to the load characteristic indexes obtained through calculation, and in addition, the typical transformer substation is selected; a synthetic load model is built for the selected typical transformer substation; and the modern interior point theory is utilized for identifying the parameters to be solved of the synthetic load model, and an optimal synthetic load model is built. The load modeling method and the loading modeling system have the advantages that precise parameter identification results can be effectively and fast obtained in the identification of the load model parameters, so the accuracy of the built load model is improved.

Description

Based on the touch upon load modeling method and system of modern interior-point theory of integrated information

Technical field

The present invention relates to a kind of power technology field, relate in particular to a kind of based on the touch upon load modeling method and system of modern interior-point theory of integrated information.

Background technology

Along with the continuous expansion of electrical network scale, its complexity is also more and more higher, and the dynamic stability of electrical network and voltage stability problem are more outstanding, and load model is also remarkable further to the influence of power system digital simulation and safe and stable operation thereof.

As one of electric system three big ingredients, load is numerous consumers and user's comprehensive object, because it is non-linear, time variation, change is structural, the constituent otherness causes property difference, each consumer response characteristic the unknown under voltage and frequency change excitation significantly etc., makes its modeling much more difficult than single structures such as generator, excitation system fixing element and system modelling.Studies show that load model is stable to the transient stability of electric system, voltage, trend calculating, low-frequency oscillation etc. have influence in various degree, any pessimism or optimistic load model all can cause the result of systematic analysis inaccurate, influence Voltage Stability Analysis result's confidence level greatly.Simultaneously, the dynamic load proportion may cause system frequency vibration and voltage fluctuation.Generally speaking, load model inaccurate can cause electric system to be disturbed, make its dynamic perfromance especially the dynamic perfromance of weak link be affected, thereby cause the safe and stable operation of whole electric power networks to be had a strong impact on.Simultaneously, along with making constant progress of society, the electricity consumption industry increases gradually, and the type of consumer is also varied, and this has also caused characteristics such as the randomness of loading, time variation remarkable all the more.Yet existing load modeling method is according to trade classification, and the inconsistent possibility of class internal loading Changing Pattern causes some industry internal loading to differ greatly, and can't accurately reflect the actual load situation.Therefore, how effectively load to be classified, set up the load model of realistic part throttle characteristics, it is significant to improve its accuracy and reliability.

Another key of load modeling is the selection of identification of Model Parameters algorithm.Find the solution that the most frequently used method of parameter identification has traditional optimization algorithm such as least square method and be the algorithm of the heuristic artificial intelligence of representative with genetic algorithm, particle cluster algorithm.It is high that yet traditional algorithms such as least square method require initial value, and the algorithm poor robustness is absorbed in Local Extremum easily, causes iterations many, even enters endless loop.Intelligent algorithms such as genetic algorithm can overcome the defective of traditional discrimination method, but introduced a large amount of calculated amount, caused computing time long, be difficult to handle and optimize for the dimension problem of higher, and algorithm is limited to the exploring ability in new space, converges to locally optimal solution easily.As seen, the technology of existing identification of Model Parameters and method all are difficult to obtain precise parameters, thereby have reduced the accuracy of load model.

Therefore, how to avoid the shortcoming of existing identification algorithm, thereby the accuracy that improves the precision raising load model of parameter identification is a key issue.

Summary of the invention

The present invention is excessive for the parameter identification algorithm calculated amount that exists in the construction load model that solves the prior art existence, poor robustness, the coarse characteristics of parameter identification result provide a kind of based on the touch upon load modeling method and system of modern interior-point theory of integrated information.

Provided by the invention a kind of based on the touch upon load modeling method of modern interior-point theory of integrated information, comprising:

Add up the load data of a plurality of transformer stations;

According to the load data of statistics gained, calculate the part throttle characteristics index of each transformer station;

According to the part throttle characteristics index of described calculating gained, adopt clustering procedure that described each transformer station is classified, and choose typical transformer station;

For the typical transformer station that chooses sets up integrated load model;

Utilize the parameter to be asked of the described integrated load model of modern interior-point theory identification, set up the optimal synthesis load model.

Wherein, described for the typical transformer station that chooses sets up integrated load model, comprising:

Selected typical transformer station is set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations respectively;

To static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model, set up integrated load model.

Wherein, utilize the parameter to be asked of the described integrated load model of modern interior-point theory identification, set up the optimal synthesis load model, comprising:

Utilize the modern interior-point theory parameter to be asked of the described static load model of identification, dynamic load model, power distribution network load model and reactive-load compensation load model respectively, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

With the described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that get access to wait ask the optimal value of parameter generation return in described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model, so that static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized respectively;

Static load model, dynamic load model, power distribution network load model and reactive-load compensation load model after optimizing are sued for peace, set up the optimal synthesis load model.

Wherein, the load data of a plurality of transformer stations of described statistics comprises:

Add up r transformer station's multiple spot every day charge value.

Wherein, according to the load data of statistics gained, calculate the part throttle characteristics index of each transformer station, comprising:

According to r transformer station's multiple spot every day charge value of statistics, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

Wherein, according to the part throttle characteristics index of described calculating gained, adopt clustering procedure that described each transformer station is classified, comprising:

Choose the daily load curve data as the proper vector of cluster;

According to described proper vector, utilize the fuzzy C-means clustering method that the part throttle characteristics index is classified;

Choose the typical transformer station in each classification.

Wherein, choose the daily load curve data as the proper vector of cluster, comprising:

Choose each transformer station's meritorious data of typical case's month workaday sampled point as the cluster feature vector, and the daily load curve data of regular working day are as proper vector;

Each sampling number certificate to the constitutive characteristic vector is carried out normalized;

Concrete, the note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, to each sampling number of constitutive characteristic vector according to carrying out normalized be: x _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

Wherein, described integrated load model expression formula is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

Wherein, do not consider that the expression formula of integrated load model is under the situation of power distribution network load model and reactive-load compensation load model:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right..$

Wherein, the expression formula of described optimal synthesis load model is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\ωe}}_d}^{\′}\right)+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\ωe}}_d}^{\′})+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\end{array}\right.;$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

Accordingly, it is a kind of based on the touch upon load modeling system of modern interior-point theory of integrated information that the present invention also provides, and comprising:

The load data statistical module is for the load data of a plurality of transformer stations of statistics;

The characteristic index computing module is used for the load data according to described load data statistical module counts gained, calculates the part throttle characteristics index of each transformer station;

Transformer station's sort module is used for the part throttle characteristics index according to described characteristic index computing module calculating gained, adopts clustering procedure that described each transformer station is classified, and chooses typical transformer station;

Load model is set up module, and the typical transformer station that is used to described transformer station sort module to choose sets up integrated load model;

Load model is optimized module, utilizes the described load model of modern interior-point theory identification to set up the parameter to be asked of the integrated load model of module foundation, sets up the optimal synthesis load model.

Wherein, described load model is set up module, comprising:

Submodel is set up the unit, is used for selected typical transformer station is set up respectively static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations;

Load model is set up the unit, is used for described submodel is set up static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model of setting up the unit, sets up integrated load model.

Wherein, described load model is optimized module, comprising:

The parameter identification unit, be used for utilizing modern interior-point theory respectively the described submodel of identification set up the parameter to be asked of static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model set up the unit, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

Submodel is optimized the unit, the optimal value of waiting to ask parameter that is used for static load model, dynamic load model, power distribution network load model and reactive-load compensation load model that described parameter identification unit is got access to respectively generation return described submodel and set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that the unit is set up, so that described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized;

Unified model is optimized the unit, is used for setting up the optimal synthesis load model to suing for peace through static load model, dynamic load model, power distribution network load model and reactive-load compensation load model after the submodel optimization unit optimization.

Wherein, described load data statistical module specifically is used for r transformer station's multiple spot every day charge value of statistics.

Wherein, described characteristic index computing module specifically is used for: according to r transformer station's multiple spot every day charge value of described load data statistical module counts, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

Wherein, described transformer station sort module comprises:

Proper vector is chosen the unit, is used for choosing the daily load curve data as the proper vector of cluster;

The characteristic index taxon for the proper vector of choosing unit selection according to described proper vector, utilizes the fuzzy C-means clustering method that the part throttle characteristics index is classified;

Typical case transformer station chooses the unit, be used for choosing described characteristic index taxon sub-category electric typical transformer station.

Wherein, described proper vector is chosen the unit, comprising:

Proper vector is chosen subelement, be used for choosing each transformer station's meritorious data of typical case's month workaday sampled point as the cluster feature vector, and the daily load curve data of regular working day is as proper vector;

The normalized subelement is used for each sampling number certificate of constitutive characteristic vector is carried out normalized;

Concrete, the note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, the normalized subelement to each sampling number of constitutive characteristic vector according to carrying out normalized is: x _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

Wherein, described load model is set up the integrated load model expression formula of setting up the unit and is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

Wherein, do not consider under the situation of power distribution network load model and reactive-load compensation load model that the expression formula that described load model is set up the integrated load model of setting up the unit is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right..$

Wherein, the expression formula of the optimal synthesis load model of described unified model optimization unit foundation is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\ωe}}_d}^{\′}\right)+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\ωe}}_d}^{\′})+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\end{array}\right.;$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

Provided by the invention a kind of based on the touch upon load modeling method and system of modern interior-point theory of integrated information, adopt modern interior-point theory to carry out parameter identification.Modern interior-point theory is suitable for finding the solution advantages such as continuous large-scale nonlinear optimization problem, robustness is good, processing ill-conditioning problem ability is strong, is applicable to planning field, particularly nonlinear programming problem.Load model parameters is being carried out in the identification, can obtain the precise parameters identification result fast and effectively, thereby improving the accuracy of the load model of setting up.

Description of drawings

In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, to do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art below, apparently, accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 is provided by the invention a kind of based on the touch upon schematic flow sheet of load modeling method embodiment one of modern interior-point theory of integrated information;

Fig. 2 is provided by the invention a kind of based on the touch upon schematic flow sheet of load modeling method embodiment two of modern interior-point theory of integrated information;

Fig. 3 is provided by the invention a kind of based on the touch upon structural representation of load modeling system embodiment one of modern interior-point theory of integrated information;

Fig. 4 is provided by the invention a kind of based on the touch upon structural representation of load modeling system embodiment two of modern interior-point theory of integrated information;

Fig. 5 is provided by the invention a kind of based on the touch upon structural representation of load modeling system embodiment three of modern interior-point theory of integrated information;

Fig. 6 is provided by the invention a kind of based on the touch upon structural representation of load modeling system embodiment four of modern interior-point theory of integrated information;

Fig. 7 is provided by the invention a kind of based on the touch upon structural representation of load modeling system embodiment five of modern interior-point theory of integrated information;

Fig. 8 is for provided by the invention a kind of based on the touch upon structural drawing of the integrated load model that the load modeling system of modern interior-point theory builds of integrated information.

Embodiment

The present invention is excessive for the parameter identification algorithm calculated amount that exists in the construction load model that solves the prior art existence, poor robustness, the coarse characteristics of parameter identification result provide a kind of based on the touch upon load modeling method and system of modern interior-point theory of integrated information.

Referring to Fig. 1, provided by the invention a kind of based on the touch upon schematic flow sheet of load modeling method embodiment one of modern interior-point theory of integrated information, this method comprises:

Step 100 is added up the load data of a plurality of transformer stations;

Concrete, can add up r transformer station's multiple spot every day charge value.

Step 101 according to the load data of statistics gained, is calculated the part throttle characteristics index of each transformer station;

Concrete, according to r transformer station's multiple spot every day charge value of statistics, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

Step 102 according to the part throttle characteristics index of described calculating gained, adopts clustering procedure that described each transformer station is classified, and chooses typical transformer station;

Concrete, choose the daily load curve data as the proper vector of cluster; According to described proper vector, utilize the fuzzy C-means clustering method that the part throttle characteristics index is classified; Choose the typical transformer station in each classification.

Step 103 is for the typical transformer station that chooses sets up integrated load model;

Step 104 is utilized the parameter to be asked of the described integrated load model of modern interior-point theory identification, sets up the optimal synthesis load model.

Referring to Fig. 2, provided by the invention a kind of based on the touch upon schematic flow sheet of load modeling method embodiment two of modern interior-point theory of integrated information, this method comprises:

The typical transformer station that present embodiment will be described as choosing sets up integrated load model and to the flow process that it is optimized, comprises the steps:

Step 200 is set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations respectively to selected typical transformer station;

Step 201 to static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model, is set up integrated load model.

The integrated load model P that the present invention sets up _SLM+ jQ _SLMForm, it embodies formula and is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

Wherein, do not consider that the expression formula of integrated load model is under the situation of power distribution network load model and reactive-load compensation load model:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right..$

Wherein, described for the typical transformer station that chooses sets up integrated load model, also comprise:

Step 202, utilize the modern interior-point theory parameter to be asked of the described static load model of identification, dynamic load model, power distribution network load model and reactive-load compensation load model respectively, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

Step 203, with the described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that get access to wait ask the optimal value of parameter generation return in described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model, so that static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized respectively;

Step 204 is sued for peace to static load model, dynamic load model, power distribution network load model and reactive-load compensation load model after optimizing, sets up the optimal synthesis load model.

The expression formula of described optimal synthesis load model is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\ωe}}_d}^{\′}\right)+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\ωe}}_d}^{\′})+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\end{array}\right.;$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

Accordingly, the present invention also provides a kind of based on the touch upon load modeling system of modern interior-point theory of integrated information.

Referring to Fig. 3, be the structural representation of the load modeling system embodiment one of the modern interior-point theory of touching upon based on integrated information, this system comprises:

Load data statistical module 10 is for the load data of a plurality of transformer stations of statistics;

Concrete, described load data statistical module 10 concrete r transformer station's multiple spot every day charge values of statistics that are used for.

Characteristic index computing module 11 is used for the load data according to described load data statistical module 10 statistics gained, calculates the part throttle characteristics index of each transformer station;

Concrete, described characteristic index computing module 11 specifically is used for: according to r transformer station's multiple spot every day charge value of described load data statistical module counts, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

Transformer station's sort module 12 is used for the part throttle characteristics index according to described characteristic index computing module 11 calculating gained, adopts clustering procedure that described each transformer station is classified, and chooses typical transformer station;

Load model is set up module 13, and the typical transformer station that is used to described transformer station sort module 12 to choose sets up integrated load model;

Load model is optimized module 14, utilizes the described load model of modern interior-point theory identification to set up the parameter to be asked of the integrated load model of module 13 foundation, sets up the optimal synthesis load model.

Referring to Fig. 4, be the structural representation of the load modeling system embodiment two of the modern interior-point theory of touching upon based on integrated information.

Present embodiment two will be described the The Nomenclature Composition and Structure of Complexes of transformer station's sort module 12, comprise:

Proper vector is chosen unit 120, is used for choosing the daily load curve data as the proper vector of cluster;

Characteristic index taxon 121 is used for choosing the proper vector that unit 120 is chosen according to described proper vector, utilizes the fuzzy C-means clustering method that the part throttle characteristics index is classified;

Typical case transformer station chooses unit 122, is used for choosing 121 sub-category electric typical transformer stations of described characteristic index taxon.

Referring to Fig. 5, be the structural representation of the load modeling system embodiment three of the modern interior-point theory of touching upon based on integrated information.

Present embodiment three will be described the composition that proper vector is chosen unit 120, comprise:

Proper vector is chosen subelement 1200, be used for choosing each transformer station's meritorious data of typical case's month workaday sampled point as the cluster feature vector, and the daily load curve data of regular working day is as proper vector;

Normalized subelement 1201 is used for each sampling number certificate of constitutive characteristic vector is carried out normalized;

Concrete, the note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, the normalized subelement to each sampling number of constitutive characteristic vector according to carrying out normalized is: x _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

Referring to Fig. 6, be the structural representation of the load modeling system embodiment four of the modern interior-point theory of touching upon based on integrated information.

Present embodiment four will be described the The Nomenclature Composition and Structure of Complexes that described load model is set up module, comprise:

Submodel is set up unit 130, is used for selected typical transformer station is set up respectively static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations;

Load model is set up unit 131, is used for described submodel is set up static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model of setting up unit 130, sets up integrated load model.

Wherein, described load model is set up the integrated load model expression formula of setting up unit 131 and is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

Wherein, do not consider under the situation of power distribution network load model and reactive-load compensation load model that the expression formula that described load model is set up the integrated load model of setting up unit 131 is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right..$

Referring to Fig. 7, be the structural representation of the load modeling system embodiment five of the modern interior-point theory of touching upon based on integrated information.

Present embodiment five will be described the The Nomenclature Composition and Structure of Complexes that load model is optimized module, comprise:

Parameter identification unit 140, be used for utilizing modern interior-point theory respectively the described submodel of identification set up the parameter to be asked of static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model set up unit 130, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

Submodel is optimized unit 141, the optimal value of waiting to ask parameter that is used for static load model, dynamic load model, power distribution network load model and reactive-load compensation load model that described parameter identification unit 140 is got access to respectively generation return described submodel and set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that the unit is set up, so that described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized;

Unified model is optimized unit 142, is used for static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model optimized through submodel after unit 141 is optimized are sued for peace, and sets up the optimal synthesis load model.

Wherein, the expression formula of the optimal synthesis load model of described unified model optimization unit 142 foundation is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\ωe}}_d}^{\′}\right)+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\ωe}}_d}^{\′})+\mathrm{\ω}0.129(u_q-{{\mathrm{\ωe}}_q}^{\′})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\ω}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\ωe}}_q}^{\′})-\mathrm{\ω}0.129(u_d-{{\mathrm{\ωe}}_d}^{\′})]\end{array};\right.$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

Below will introduce provided by the invention a kind of based on the touch upon specific embodiment of load modeling method and system of modern interior-point theory of integrated information in detail.

Provided by the inventionly a kind ofly touch upon based on integrated information that to carry out the method flow of load modeling as follows for the load modeling system of modern interior-point theory:

Step 1 is collected r transformer station N point load 1 year every day value, and the load value of every day is expressed as y ₁, y ₂..., y _i..., y ₉₆, y wherein _iRepresent load value.In the present embodiment, be 96 to be example explanation with the N value, other numerical value principles are similar, repeat no more.

Step 2 according to 96 the load data of collecting every day, is calculated each transformer station's part throttle characteristics index, comprises day maximum (little) load (P _Max/ P _Min), per day load (P _Av), daily load rate (K _d), day ratio of minimum load to maximum load (β _d), day peak-valley difference (H), day peak valley rate (α _d), month peak load (P _M-max), month minimum load (P _M-min), the monthly average daily load

Monthly average daily load rate (K _Av), monthly load factor (K _M), month ratio of minimum load to maximum load (β _Min), month maximum peak-valley difference (H _Max), month maximum peak valley rate (α _Max), monthly average day peak-valley difference (H _Av), monthly average day peak valley rate (α _Av)

Step 3 according to the part throttle characteristics index, is classified r transformer station.

The concrete grammar of cluster is as follows:

R is the number of transformer station, and c is the number of classification, and that month that define power consumption maximum in a year be the typical moon.

x _k=(x _K1..., x _KN1, x _KN1+1..., x _KN2, x _KN2+1..., x _KN3, x _KN3+1..., x _KN4, x _KN4+1, x _KN4+2, x _KN4+3, x _KN4+4, x _KN4+5, x _KN4+6, x _KN5, x _KN5+1..., x _KN6, x _KN6+1..., x _KN7, x _KN7+1, x _KN7+2, x _KN7+3, x _KN7+4) be the proper vector of describing k transformer station, wherein x _K1, x _K2,, x _KN1The load of common η * 96 point of expression typical case month η regular working day, x _KN1+1..., x _KN2The day peak load of expression typical case month η regular working day, x _KN2+1..., x _KN3The day minimum load of expression typical case month η regular working day, x _KN3+1..., x _KN4The day peak-valley difference of expression typical case month η regular working day, x _KN4+1The moon peak load of the expression typical case moon, x _KN4+2The moon minimum load of the expression typical case moon, x _KN4+3The monthly average daily load of the expression typical case moon, x _KN4+4The moon maximum peak-valley difference of the expression typical case moon, x _KN4+5The monthly average day peak-valley difference of the expression typical case moon, x _KN4+6..., x _KN5The day peak valley rate of expression typical case month η regular working day, x _KN5+1..., x _KN6The daily load rate of expression typical case month η regular working day, x _KN6+1..., x _KN7The day ratio of minimum load to maximum load of expression typical case month η regular working day, x _KN7+1The monthly average daily load rate of the expression typical case moon, x _KN7+2The moon ratio of minimum load to maximum load of the expression typical case moon, x _KN7+3The moon maximum peak valley rate of the expression typical case moon, x _KN7+4The monthly average day peak valley rate of the expression typical case moon.

The present invention adopts the method for cluster to classify, and the basic thought of cluster is as follows:

(1) sets up the model of cluster.

The objective function of definition cluster is the weighted error quadratic sum of transformer station's proper vector and its cluster centre, that is:

$\left\{\begin{array}{c}\min \left\{J_m(U,P)\right\}=\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}\min \left\{\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^m{\left(d_{\mathrm{ik}}\right)}^2\right\},m\&Element;[1,\&infin;)\\ s.t.U\&Element;M_{\mathrm{fc}}\\ M_{\mathrm{fc}}=\{U\&Element;R^{\mathrm{cr}}|{\mathrm{\&mu;}}_{\mathrm{ik}}\&Element;\left[\mathrm{0,1}\right],\&ForAll;i,k;\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ik}}=1,\&ForAll;k;0<\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}<r,\&ForAll;i\}\end{array}\right.---\left(1\right)$

Wherein,

J _m(U, P): the function representation formula that expression is relevant with the cluster centre matrix with dividing matrix;

Matrix is divided in expression, and the row of dividing matrix U represents classification, and row represent transformer station, its element μ _IkRepresent that k transformer station is under the jurisdiction of the degree of membership of i class.Press maximum membership grade principle: the x of transformer station _kMaximum membership degree max (μ _1k, μ _2k..., μ _Ck) capable at i, just belong to the i class, thereby can determine classification under the transformer station.

P=(p ₁, p ₂..., p _c) ^TExpression cluster centre matrix, its line display classification, row are identical with the row expression and significance of transformer station's proper vector.Its expression formula is:

${P=(p_1,p_2,\&CenterDot;\&CenterDot;\&CenterDot;,p_c)}^T=\left[\begin{array}{ccccc}{p}_{11}& p_{12}& {\mathrm{<figure-callout\; id="13"\; label="p"\; filenames="HDA00003115573700031.png,HDA00003115573700042.png"\; state="\{\{state\}\}">p</figure-callout>}}_{13}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{1,N7+4}\\ p_{21}& p_{22}& p_{23}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{2,N7+4}\\ p_{31}& p_{32}& p_{33}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{3,N7+4}\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ p_{c1}& p_{c2}& p_{c3}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{c,N7+4}\end{array}\right]---\left(2\right)$

In the formula, the transposition of T representing matrix, p _i=(p _I1, p _I2..., p _{I, N7+4}), i=1,2 ..., c represents the cluster centre of i class, c represents cluster numbers.

M _FcRefer to μ _IkField of definition, R ^CrThe real number space that refers to c * r rank;

M represents weighted index;

d _IkThe expression x of transformer station _kCluster centre p with the i class _iDistance between two vectors, its computing formula is:

(d _ik) ²=||x _k-p _i|| _A=(x _k-p _i) ^TA(x _k-p _i) （3）

In the formula, T represents transposition, and A is the symmetric positive definite matrix on s * s rank, and when getting unit matrix, formula (19) is corresponding to Euclidean distance;

(2) find the solution above-mentioned Clustering Model.

By finding the solution formula (1), can obtain dividing the formula of finding the solution (8) of matrix element and the formula of finding the solution (9) of cluster centre vector, concrete method for solving is:

1) to the following Lagrangian function of making of objective function structure:

$F=\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^m{\left(d_{\mathrm{ik}}\right)}^2+\mathrm{\&lambda;}(\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}-1)---\left(4\right)$

2) to parameter lambda, μ _Ik, p _iAsk local derviation respectively,

$\frac{\&PartialD;F}{\&PartialD;\mathrm{\&lambda;}}=(\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}-1)=0---\left(5\right)$

$\frac{\&PartialD;F}{\&PartialD;{\mathrm{\&mu;}}_{\mathrm{ik}}}=[m{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^{m-1}{\left(d_{\mathrm{ik}}\right)}^2-\mathrm{\&lambda;}]=0---\left(6\right)$

$\frac{\&PartialD;F}{\&PartialD;p_i}=\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^m\frac{\&PartialD;}{\&PartialD;p_i}\left[{(x_k-p_i)}^{T}A(x_k-p_i)\right]=0---\left(7\right)$

3) can to get the necessary condition of extreme value of (1) formula as follows in solving equation (5)-(7):

${\mathrm{\&mu;}}_{\mathrm{ik}}=\frac{1}{{\underset{j=1}{\overset{c}{\mathrm{\&Sigma;}}}\left(\frac{d_{\mathrm{ik}}}{d_{\mathrm{jk}}}\right)}^{\frac{2}{m-1}}}---\left(8\right)$

$p_i=\frac{\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^m{x}_k}{\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}\right)}^m}---\left(9\right)$

Generalized case, Weighting exponent m is taken as 2.

(3) cluster classification number determines.

The present invention intends adopting cluster validity function P ' (U; C) as the method for judging the cluster optimal number, be specially: the scope according to actual conditions cluster numbers c is generally 2～6 classes, so c gets 5 following specific implementation step process of 6 repetitions from 2 and obtains 5 cluster validity functional values, judge that again the cluster numbers of getting maximum functional value correspondence is best cluster numbers, and the division matrix of output under this cluster numbers obtains final classification results.

Cluster validity function P ' (U; C) computing formula is as follows:

$P^{\&prime;}(U;c)=\frac{{\min }_{i=1}^c\left(\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}\right)}{{\max }_{i=1}^c\left(\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}\right)}[P(U;c)+1-\frac{\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}^2{\left|\right|x_k-p_i\left|\right|}^2}{J_0}]---\left(10\right)$

Wherein, r is transformer station's number;

x _kBe transformer station's proper vector, k=1,2 ..., r;

C is clusters number;

p _iBe i cluster centre, i=1,2 ..., c; U is the degree of membership matrix;

$P(U;c)=\frac{1}{c}\underset{i=1}{\overset{c}{\mathrm{\&Sigma;}}}(\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}^2/\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ik}}^2)$ Be the possibility division factor;

Center for transformer station's proper vector.

For all transformer station's proper vectors arrive V ₀Apart from sum;

Below be the specific implementation step:

1) initialization: a given cluster classification is counted c=2 earlier, sets iteration stopping threshold values ε=10 ^-6, consequence counter, initialization cluster centre P ⁽⁰⁾, iterations b is set to 0.Initial p ⁽⁰⁾The arithmetic mean of desirable transformer station proper vector;

2) calculate distance matrix D ^(b)

D ^(b)The distance matrix that forms when being the b time iteration, its element are by Euclidean distance d ^IkConstitute, calculate distance matrix with (3) formula, its form is as follows:

3) calculate the division matrix U ^(b)

U ^(b)The division matrix that forms when being the b time iteration, its element are by degree of membership μ _IkConstitute.There are two kinds of situations in the computing method of degree of membership wherein:

For the k of transformer station, each distances of clustering centers of it and c class is not 0, then can calculate degree of membership by (8) formula and obtain formula (11).

${\mathrm{\&mu;}}_{\mathrm{ik}}^{\left(b\right)}=\{\underset{j=1}{\overset{c}{\mathrm{\&Sigma;}}}{\left[{\left(\frac{d_{\mathrm{ik}}^{\left(b\right)}}{d_{\mathrm{jk}}^{\left(b\right)}}\right)}^{\frac{2}{m-1}}\right]}^{-1}---\left(11\right)$

If certain the cluster centre j in the k of transformer station and the c class apart from d _JkEqual 0, the k of transformer station is 1 to the degree of membership of j class so, is 0 to the degree of membership of all the other classes, namely has

${\mathrm{\&mu;}}_{\mathrm{jk}}^{\left(b\right)}=1,$ And to i ≠ j, ${\mathrm{\&mu;}}_{\mathrm{ik}}^{\left(b\right)}=0---\left(12\right)$

Wherein, i, j belong to set 1,2 ..., c};

4) upgrade cluster centre matrix P ^(b+1), P ^(b+1)The cluster centre matrix that forms when representing the b+1 time iteration.Its formula of finding the solution is as (13):

$p_i^{b+1}=\frac{\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}^{\left(b\right)}\right)}^m\&times;x_k}{\underset{k=1}{\overset{r}{\mathrm{\&Sigma;}}}{\left({\mathrm{\&mu;}}_{\mathrm{ik}}^{\left(b\right)}\right)}^m}$ $i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,c---\left(13\right)$

5) if || P ^(b)-P ^(b+1)||＜ε, then algorithm stops and exporting and divide matrix U and cluster centre matrix P, otherwise P ^(b)=P ^(b+1)Turn to (2).

6) be updated to cluster validity function P ' (U dividing matrix and cluster centre matrix; C) (formula 10) evaluation.

7) result and P ' (U are divided in output; C) functional value, the form of dividing the result is:

I class={ l ₁, l ₂..., l _k} ^c

II class={ χ ₁, χ ₂..., χ _k} ^c

.

.

.

C class={ ζ ₁, ζ ₂..., ζ _k} ^c

L wherein ₁, l ₂..., l _k, χ ₁, χ ₂..., χ _k, ζ ₁, ζ ₂..., ζ _kBe integer, represent the numbering of transformer station respectively, subscript c(c=1,2 ..., 5) and c result of expression.

8) judge whether c equals 6, if then stop; Cluster numbers c gets 2,3,4,5 respectively, and 6 have repeated above-mentioned 8 steps 5 times, obtain 5 results, judges P ' (U; When c) getting maximal value, corresponding division result is as final optimum division result's output.

Final output result is:

I class={ l ₁, l ₂..., l _k} ^z

II class={ χ ₁, χ ₂..., χ _k} ^z

.

.

.

C class={ ζ ₁, ζ ₂..., ζ _k} ^z

Wherein, subscript z ∈ { 1,2,3,4,5}.

${P=(p_1,p_2,\&CenterDot;\&CenterDot;\&CenterDot;,p_c)}^T=\left[\begin{array}{ccccc}{p}_{11}& p_{12}& {\mathrm{<figure-callout\; id="13"\; label="p"\; filenames="HDA00003115573700031.png,HDA00003115573700042.png"\; state="\{\{state\}\}">p</figure-callout>}}_{13}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{1,N7+4}\\ p_{21}& p_{22}& p_{23}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{2,N7+4}\\ p_{31}& p_{32}& p_{33}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{3,N7+4}\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ p_{c1}& p_{c2}& p_{c3}& \&CenterDot;\&CenterDot;\&CenterDot;& p_{c,N7+4}\end{array}\right];$

Otherwise c=c+1 changeed for (2) step to be calculated.

Step 4 can obtain transformer station's classification results according to step 3, select all kinds of in typical transformer station carry out load modeling.

According to cluster centre matrix (formula 3) as can be known: row represents classification in the matrix, and row represent transformer station.The maximal value of every row in the selection matrix, the transformer station of this value representative is as the typical transformer station in this classification.Below will set up the integrated load model (SLM) of all kinds of transformer stations to selected typical transformer station respectively, and Model parameter will be carried out identification, determine final mask;

The concrete grammar of determining integrated load model is as follows:

Integrated load model is to be obtained by static load model, dynamic load model, power distribution network load model and four partial summations of reactive-load compensation load model.Complete integrated load model structural drawing as shown in Figure 8.

Among Fig. 8, Us is the actual load busbar voltage, and UL is virtual busbar voltage.Part between actual load bus and the virtual bus is equivalent power distribution network impedance.RD, XD are respectively power distribution network substitutional resistance and reactance.P, Q are active power and the reactive power of actual load bus output, and PL, QL are through load active power and reactive power behind the distribution network.Symbol M represents induction motor part in the integrated load model, and ZIP represents the static load part, and C represents the reactive-load compensation part.

As shown in Figure 8, integrated load model is P _SLM+jQ _SLMForm.Wherein, P _SLMExpression load active power, Q _SLMExpression reactive load power, the formula of embodying is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.---\left(14\right)$

In the formula, PSLM represents integrated load model active power, and PZIP represents static load model active power, PM represents dynamic load model active power, PD represents power distribution network load model active power, and PC represents reactive-load compensation load model active power part, PC=0 herein.In like manner as can be known, QSLM, QZIP, QM, QD, QC represent the reactance capacity of integrated load model, static load model, dynamic load model, power distribution network load model and reactive-load compensation load model respectively.

Below will introduce concrete definite method of four department patterns respectively.

1) static load model

The static load model has reflected the rule that load is meritorious, reactive power slowly changes with frequency and voltage, its curve approximation straight line or smooth curve, and available algebraic equation is represented.The static load model is P _ZIP+jQ _ZIPForm.P _ZIP, Q _ZIPThe form that embodies mainly contain two kinds of power function and polynomial expressions.

Definite method concrete steps of static load model are as follows:

1. set up the static load model:

When static load change in voltage scope less than+10% situation under, P _ZIP, Q _ZIPComputing formula following (suc as formula 15):

$\left\{\begin{array}{c}{P}_{\mathrm{ZIP}}=P_0{\left(\frac{U}{U_0}\right)}^{\mathrm{pv}}{\left(\frac{f}{f_0}\right)}^{\mathrm{pf}}\\ Q_{\mathrm{ZIP}}=Q_0{\left(\frac{U}{U_0}\right)}^{\mathrm{qv}}{\left(\frac{f}{f_0}\right)}^{\mathrm{qf}}\end{array}\right.---\left(15\right)$

Wherein, pv, qv are the voltage characteristic index of load active power and reactive power; Pf, qf are the frequency characteristic index of load meritorious and reactive power; Pv, qv, pf, qf are parameter to be asked.

P ₀, Q ₀, U ₀, f ₀The load active power of calculation of tidal current, reactive power, load busbar voltage amplitude and frequency when being respectively system stability can be calculated acquisition by BPA software.

U, f are transformer station's load bus voltage of 96 one day, frequency measured value, and this value can be provided by transformer station's SCADA software.

PZIP, QZIP are respectively the output result of model.

When the static load voltage range changes greatly, set up multinomial model (suc as formula 16)

$\left\{\begin{array}{c}{P}_{\mathrm{ZIP}}=P_0[p_1{\left(\frac{U}{U_0}\right)}^2+p_2\left(\frac{U}{U_0}\right)+p_3]\\ Q_{\mathrm{ZIP}}=Q_0[q_1{\left(\frac{U}{U_0}\right)}^2+q_2\left(\frac{U}{U_0}\right)+q_3]\end{array}\right.---\left(16\right)$

In the formula, the zero degree item is equivalent to firm power load (P); Once item is equivalent to steady current load (I); Quadratic term is equivalent to constant impedance load (Z).P1, p2, p3 represent the constant impedance part that total load is heavy respectively, steady current part, the active power of firm power part; Q1, q2, q3 represent the constant impedance part that total load is heavy respectively, steady current part, the reactive power of firm power part; P1, p2, p3, q1, q2, q3 are parameter to be asked.

P0, Q0, load active power, reactive power and the load busbar voltage amplitude of calculation of tidal current when U0 is respectively system stability can be calculated by BPA software and obtain.

Concerning multinomial model, model be input as voltage magnitude U, frequency f, this value can be obtained by SCADA system of transformer station; Output quantity is load active power PZIP, reactive load power QZIP.

2. with the theoretical identification model of interior some parameter to be asked:

Step1: the nonlinear programming problem of setting up the static load Model Distinguish

The main task of identification of Model Parameters is to seek one group of best parameter vector, makes the target function value minimum of error.Among the present invention, the objective definition function is the quadratic sum of the error of voltage calculating and measured value, and then objective function can be expressed as:

$\left\{\begin{array}{c}\min \underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}{(U_{\mathrm{cal}}-U_{\mathrm{mea}})}^2\\ s.t.\underset{\&OverBar;}{\mathrm{\&theta;}}\&le;\mathrm{\&theta;}\&le;\stackrel{\&OverBar;}{\mathrm{\&theta;}}\end{array}\right.---\left(17\right)$

Wherein, Ucal is the model calculated value, and Umea is the model measurement amount, and θ is identified parameters.

Step2: adopt Modern Interior Point Optimization Algorithm to find the solution objective function.

Active power with multinomial model partly is example, and then the objective function of static load model is:

$\min f_P\left(x\right)=\underset{k=1}{\overset{96}{\mathrm{\&Sigma;}}}{(P_{\mathrm{ZIP}}-P)}^2$

$s.t.h\left(x\right)=p_1+p_2+p_3-1=0---\left(18\right)$

Wherein, x=[p ₁, p ₂, p ₃] ^TBe parameter vector to be identified, T representing matrix transposition.

The concrete steps that interior some theory is found the solution objective function are as follows:

The initialization model parameter arranges iteration precision ε, iterations k;

The structure Lagrangian function, and find the solution the KKT condition of objective function, obtain:

L _P=f(x)-y ^Th(x)

$\left\{\begin{array}{c}{L}_{x_P}=\&dtri;f\left(x\right)-y^T\&dtri;h\left(x\right)=\begin{array}{c}\left[\begin{array}{c}2(P_{\mathrm{ZIP}}-P)*{\left(\frac{U}{U_0}\right)}^2\\ 2(P_{\mathrm{ZIP}}-P)*\left(\frac{U}{U_0}\right)\\ 2(P_{\mathrm{ZIP}}-P)*E\end{array}\right]-y^T\left[\begin{array}{c}{\left(\frac{U}{U_0}\right)}^2\\ \left(\frac{U}{U_0}\right)\\ 1\end{array}\right]\end{array}\\ L_y=h\left(x\right)=[p_1+p_2+p_3-1]\end{array}\right.---\left(19\right)$

E is the unit column vector in the formula, and y is Lagrange multiplier.

Definition according to differential is rewritten as matrix form with following formula, and expression formula is:

$\left[\begin{array}{cc}H& \&dtri;h\left(x\right)\\ \&dtri;h{\left(x\right)}^T& 0\end{array}\right]\left[\begin{array}{c}\mathrm{\&Delta;x}\\ \mathrm{\&Delta;y}\end{array}\right]=-\left[\begin{array}{c}-L_x\\ h\left(x\right)\end{array}\right]---\left(20\right)$

In the formula: $H=-\left[\begin{array}{ccc}{\left(\frac{U}{U_0}\right)}^2*{\left(\frac{U}{U_0}\right)}^2& {\left(\frac{U}{U_0}\right)}^2*\left(\frac{U}{U_0}\right)& {\left(\frac{U}{U_0}\right)}^2*E\\ {\left(\frac{U}{U_0}\right)}^2*\left(\frac{U}{U_0}\right)& {\left(\frac{U}{U_0}\right)}^2*E& \left(\frac{U}{U_0}\right)*E\\ {\left(\frac{U}{U_0}\right)}^2*E& \left(\frac{U}{U_0}\right)*E& 2\end{array}\right],$ $\mathrm{\&Delta;x}=\left[\begin{array}{c}\mathrm{\&Delta;}{p}_1\\ \mathrm{\&Delta;}{p}_2\\ \mathrm{\&Delta;}{p}_3\end{array}\right]$

Solving equation (20) can obtain the correction of the k time iteration, and is as follows:

x ^(k+1)=x ^(k)+△x

y ^(k+1)=y ^(k)+△y （21）

Judge whether target function value satisfies iteration precision, and whether iterations exceeds set maximal value.

If do not satisfy precision and iterations does not exceed, then return step B and continue to calculate; If satisfy precision and iterations does not exceed, then export the optimum solution of x; If precision does not satisfy and exceeds the iterations maximal value, then function does not have optimum solution, withdraws from circulation.

3. decide the static load model:

By parameter identification, can obtain waiting asking ginseng p1, p2, p3, q1, q2, q3(or pv, qv, pf, qf) optimal value.At this moment, these optimal value back substitutions are arrived in the static load model (formula 14,15) the static load model that can obtain determining.To can be obtained PZIP, the QZIP in the corresponding moment by BPA software calculating acquisition P0, Q0, the above-mentioned static model of U0 substitution by certain U, f constantly of SCADA system of transformer station acquisition.

2) dynamic load model

The rule that dynamic load model reaction bus load is meritorious, reactive power changes with the variation of busbar voltage, frequency and time, the general available differential equation or difference equation are represented, because its principal ingredient is induction motor, therefore, induction motor model commonly used is represented the dynamic load model.

Definite method concrete steps of dynamic load model are as follows:

1. set up the dynamic load model:

The expression formula of three rank induction motor load model PM, QM is as follows:

$\left\{\begin{array}{c}{P}_M=u_d{i}_d+u_q{i}_q\\ Q_M=u_q{i}_d-u_d{i}_q\end{array}\right.---\left(22\right)$

In the formula, PM, QM represent active power, the reactive power output of dynamic model respectively, ud, uq represent the d axle (d-axis) of induction motor voltage, the coordinate components of q axle (handing over axle) respectively, it is the model input quantity, id, iq are the d axle (d-axis) of load current, the coordinate components of q axle (handing over axle), and the solution formula of id, iq is as follows:

$\left\{\begin{array}{c}{i}_d=\frac{1}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}[R_s(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})+{\mathrm{\&omega;x}}^{\&prime;}(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ i_q=\frac{1}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}[R_s(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-{\mathrm{\&omega;x}}^{\&prime;}(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\end{array}\right.---\left(23\right)$

In the formula, ω is system's actual angular frequency;

Rs represents the resistance of stator;

The reactance of x ' expression transient state, its expression formula is:

x'=x _s+x _mx _s/(x _m+x _s)

Wherein, xr is the rotor reactance, and xm is excitatory reactance.

e _d', e _q' represent the coordinate components of the d axle (d-axis), q axle (handing over axle) of induction motor transient potential, e respectively _d', e _q' acquisition be to obtain by the differential equation of finding the solution transient potential, its expression formula is as follows:

$\frac{{{\mathrm{de}}_d}^{\&prime;}}{\mathrm{dt}}=-\frac{1}{{T_0}^{\&prime;}}{e_d}^{\&prime;}+\frac{x_s-x^{\&prime;}}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}(R_s(-{{\mathrm{\&omega;e}}_q}^{\&prime;})-{\mathrm{\&omega;x}}^{\&prime;}(u-{{\mathrm{\&omega;e}}_d}^{\&prime;}))+{\mathrm{\&omega;}}_B(\mathrm{\&omega;}-{\mathrm{\&omega;}}_r)e_q$

$\frac{{{\mathrm{de}}_q}^{\&prime;}}{\mathrm{dt}}=-\frac{1}{{T_0}^{\&prime;}}{e_q}^{\&prime;}+\frac{x_s-x^{\&prime;}}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}(R_s(u-{{\mathrm{\&omega;e}}_d}^{\&prime;})+{\mathrm{\&omega;x}}^{\&prime;}(-{{\mathrm{\&omega;e}}_q}^{\&prime;}))+{\mathrm{\&omega;}}_B(\mathrm{\&omega;}-{\mathrm{\&omega;}}_r){e_d}^{\&prime;}$

$\frac{{\mathrm{d\&omega;}}_r}{\mathrm{dt}}=\frac{1}{H}\frac{{e_d}^{\&prime;}}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}{R}_s(u-{{\mathrm{\&omega;e}}_d}^{\&prime;})+{\mathrm{\&omega;x}}^{\&prime;}\left({{\mathrm{\&omega;e}}_q}^{\&prime;}\right)+\frac{{e_q}^{\&prime;}}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}[R_s\left({-{\mathrm{\&omega;e}}_q}^{\&prime;}\right)-{\mathrm{\&omega;x}}^{\&prime;}(u-{{\mathrm{\&omega;e}}_d}^{\&prime;})]-T_L{\mathrm{\&omega;}}_r^n---\left(24\right)$

In the formula, ω _rThe expression rotor velocity; ω _BBe the system synchronization angular frequency; Xs represents the reactance of stator; T ₀' expression stator open circuit transient state time constant; H represents the motor inertia time constant; TL represents load factor; N represents the power of the moment of resistance relevant with rotating speed.

In the dynamic load model, Rs, xs, T ₀', H, TL, n, xr, xm be parameter to be identified, ud, uq, ω, ω _BBe given value.

2. the theoretical identification model of some parameter to be asked in adopting:

Step1: the nonlinear programming problem of setting up the dynamic load Model Distinguish

Similar to the static load identification of Model Parameters, set up nonlinear model:

Active power with three rank induction motor model partly is example, and then the objective function of static load model is:

$\min f_P\left(x\right)=\underset{k=1}{\overset{96}{\mathrm{\&Sigma;}}}{(P_M-P)}^2$ $s.t.h\left(x\right)=\left\{\begin{array}{c}{R}_s(u_d-e_{d0}^{\&prime;})+x^{\&prime;}(u_q-e_{q0}^{\&prime;})-(R_s^2+x^{\&prime;2})i_{d0}=0\\ R_s(u_q-e_{q0}^{\&prime;})-x^{\&prime;}(u_d-e_{d0}^{\&prime;})-(R_s^2+x^{\&prime;2})i_{q0}=0\\ e_{d0}^{\&prime;}++i_{q0}(x_s-x^{\&prime;})-(1-{\mathrm{\&omega;}}_0){T_0}^{\&prime;}{\mathrm{\&omega;}}_B{e}_{q0}^{\&prime;}=0\\ e_{q0}^{\&prime;}+i_{d0}(x_s-x^{\&prime;})+(1-{\mathrm{\&omega;}}_0){T_0}^{\&prime;}{\mathrm{\&omega;}}_B{e}_{d0}^{\&prime;}=0\\ T_L{{\mathrm{\&omega;}}_0}^n-e_{d0}^{\&prime;}{i}_{d0}-e_{q0}^{\&prime;}{i}_{q0}=0\end{array}\right.$

$g\left(x\right):0.08<x_s<0.2---\left(25\right)$

Wherein, x=[R _s, x', T ₀', H, n, T _L] ^TBe parameter vector to be identified, T representing matrix transposition.

Step2: adopt Modern Interior Point Optimization Algorithm to find the solution objective function.

The concrete steps that interior some theory is found the solution objective function are as follows:

The initialization model parameter arranges iteration precision ε, iterations k;

The structure Lagrangian function, and find the solution the KKT condition of objective function, introduce slack variable l, v, make inequality constrain be converted into the equivalent form of value: to obtain:

$L_P=f\left(x\right)-y^{T}h\left(x\right)-z^T(g\left(x\right)-l-0.08)-w^T(g\left(x\right)+v-0.2)$

$\left\{\begin{array}{c}{L}_{x_P}\&dtri;f\left(x\right)-\&dtri;h\left(x\right)y-\&dtri;g\left(x\right)z-\&dtri;g\left(x\right)w\\ L_y=h\left(x\right)\\ L_z=g\left(x\right)-l-0.08\\ L_w=g\left(x\right)-v-0.2\\ L_l^{\mathrm{\&mu;}}=\mathrm{LZe}-\mathrm{\&mu;e}=0\\ L_v^{\mathrm{\&mu;}}=\mathrm{VWe}+\mathrm{\&mu;e}=0\\ (l,v,z)\&GreaterEqual;0,w\&le;0\end{array}\right.---\left(26\right)$

E is the unit column vector in the formula, and y is Lagrange multiplier.

Definition according to differential is rewritten as matrix form with following formula, and expression formula is:

$\left[\begin{array}{cccccc}H& \&dtri;h\left(x\right)& \&dtri;g\left(x\right)& \&dtri;g\left(x\right)& 0& 0\\ {\&dtri;}^{T}h\left(x\right)& 0& 0& 0& 0& 0\\ {\&dtri;}^{T}g\left(x\right)& 0& 0& 0& -I& 0\\ {\&dtri;}^{T}g\left(x\right)& 0& 0& 0& 0& I\\ 0& 0& L& 0& Z& 0\\ 0& 0& 0& U& 0& W\end{array}\right]\left[\begin{array}{c}\mathrm{\&Delta;x}\\ \mathrm{\&Delta;y}\\ \mathrm{\&Delta;z}\\ \mathrm{\&Delta;w}\\ \mathrm{\&Delta;l}\\ \mathrm{\&Delta;v}\end{array}\right]=-\left[\begin{array}{c}-L_x\\ L_y\\ L_z\\ L_w\\ L_l^{\mathrm{\&mu;}}\\ L_v^{\mathrm{\&mu;}}\end{array}\right]---\left(27\right)$

In the formula: $H=-[{\&dtri;}_x^{2}f\left(x\right)-{\&dtri;}_x^2\mathrm{\&theta;}(z+w)].$

Solving equation (27) can obtain the correction of the k time iteration, and is as follows:

$\left[\begin{array}{c}{x}^{(k+1)}=x^{\left(k\right)}+{\mathrm{\&alpha;}}_p\mathrm{\&Delta;x}\\ l^{(k+1)}=l^{\left(k\right)}+{\mathrm{\&alpha;}}_p\mathrm{\&Delta;l}\\ u^{(k+1)}=u^{\left(k\right)}+{\mathrm{\&alpha;}}_p\mathrm{\&Delta;u}\\ z^{(k+1)}=z^{\left(k\right)}+{\mathrm{\&alpha;}}_d\mathrm{\&Delta;z}\\ w^{(k+1)}=w^{\left(k\right)}+{\mathrm{\&alpha;}}_d\mathrm{\&Delta;w}\end{array}\right]---\left(28\right)$

In the formula, α _pAnd α _dBe step-length:

$\left.\begin{array}{c}{\mathrm{\&alpha;}}_p=0.9995\min \{\min (\frac{-l}{{\mathrm{\&Delta;l}}_x},{\mathrm{\&Delta;l}}_x<0;\frac{-u}{{\mathrm{\&Delta;u}}_x},{\mathrm{\&Delta;u}}_x<0),1\}\\ {\mathrm{\&alpha;}}_d=0.9995\min \{\min (\frac{-z}{\mathrm{\&Delta;}{z}_x},{\mathrm{\&Delta;z}}_x<0;\frac{-w}{{\mathrm{\&Delta;w}}_x},{\mathrm{\&Delta;w}}_x<0),1\}\end{array}\right\}(i=\mathrm{1,2},...,r)---\left(29\right)$

Judge whether target function value satisfies iteration precision, and whether iterations exceeds set maximal value.

If do not satisfy precision and iterations does not exceed, then return step B and continue to calculate; If satisfy precision and iterations does not exceed, then export the optimum solution of x; If precision does not satisfy and exceeds the iterations maximal value, then function does not have optimum solution, withdraws from circulation.

3. decide the dynamic load model:

By parameter identification, can obtain waiting to ask ginseng Rs, xs, T ₀', the optimal value of H, TL, n, xr, xm.At this moment, these optimal value back substitutions are arrived in the dynamic load model (formula 21,22,23) the static load model that can obtain determining.Will be by certain U, f, ω, ω constantly of SCADA system of transformer station acquisition _BThe above-mentioned dynamic model of substitution can be obtained corresponding PM, QM constantly.

3) power distribution network load model

The power distribution network load model is made up of active power part and reactance capacity equally, and the expression formula of model PD, QD is as follows:

$\left\{\begin{array}{c}{P}_D=\frac{P^2+Q^2}{U_S^2}{R}_D\\ Q_D=\frac{P^2+Q^2}{U_S^2}f{X}_D\end{array}\right.---\left(30\right)$

In the formula, PD, QD are the output result of power distribution network load model, and P, Q are active power and the reactive power (as Fig. 1) of the output of actual load bus, and Us is the actual load busbar voltage, and RD, XD are respectively power distribution network substitutional resistance and reactance, and f is the power distribution network frequency.In parameter identification, RD, XD are parameter to be asked, and all the other are known quantity.

The general comprehensive method of statistics that adopts of determining of RD, XD obtains.According to the regional power grid structure situation, statistics power distribution network line length, resistance, reactance by line length, circuit model are calculated the power distribution network circuit obtain RD or XD thus.In general, the ratio RD/XD of the two changes less, can be taken as representative value.If can't adopt the statistics overall approach in the actual engineering, then determine RD, XD according to representative value.Substitution formula (30) can be determined the power distribution network load model.

4) reactive-load compensation load model

The reactive-load compensation load model has characterized the reactive-load compensation part in the power distribution network.Generally represent with reactive-load compensation capacitor.Its model tormulation formula is:

$Q_C=-\frac{U_L^2}{X_C}=-\frac{U_L^2}{X_{C0}}f$

（31）

In the formula, QC is the output result of model; UL is virtual busbar voltage; F is system frequency, and XC is the capacitive reactance of compensation condenser; Capacitive reactance capacitor when XC0 is normal the operation, XC0=XCf.In the reactive-load compensation model, XC0 is parameter to be identified, and all the other are known quantity.

The reactive-load compensation load model can be similar to ZIP static load model representation, at this moment, reactive-load compensation and static load can be merged, and the reactance capacity that is about in reactive-load compensation load and the static load model merges.The expression formula of reactance capacity is as follows:

$Q_{\mathrm{ZIP}}=Q_0{\left(\frac{U}{U_0}\right)}^{\mathrm{qv}}{\left(\frac{f}{f_0}\right)}^{\mathrm{qf}}---\left(32\right)$

Or $Q_{\mathrm{ZIP}}=Q_0[q_1{\left(\frac{U}{U_0}\right)}^2+q_2\left(\frac{U}{U_0}\right)+q_3]---\left(33\right)$

Point is theoretical in adopting carries out identification to static load model reactance capacity parameter.By parameter identification, can obtain waiting asking ginseng qv, qf(or q1, q2, q3) optimal value.At this moment, these optimal value back substitutions are arrived in formula (32), (33) the static load model reactance capacity that comprises the reactive-load compensation part that can obtain determining.To can be obtained PZIP, the QZIP in the corresponding moment by BPA software calculating acquisition P0, Q0, the above-mentioned static model of U0 substitution by certain U, f constantly of SCADA system of transformer station acquisition.

So far, with the load model summation that above-mentioned four parts are determined, the integrated load model of the consideration frequency characteristic that can obtain to determine is shown in (34).

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.---\left(34\right)$

Below again with concrete data instance, illustrate provided by the inventionly based on the touch upon implementation process of load modeling method and system of modern interior-point theory of integrated information, it mainly is divided into following four parts:

(1) load classification and modeling data

Add up certain city each transformer station's multiple spot every day charge value in recent years; Each transformer station's daily power consumption in recent years; Transformer station's overview of each electric pressure sequence (comprise each electric pressure outlet power load equipment ratio transformer platform number, capacity, the reactive compensation capacity of substation of confession); The whole network power supply area network chart; The reactive-load compensation overview; Each electrical substation monitoring point data (meritorious, idle, voltage, frequency) in recent years; The load structure of 110kV outlet Various Seasonal and part throttle characteristics data (comprise user's name, with electric device constitute, the composition of load kind, load).

For the reason of being convenient to category division, consider load is divided into 5 big classes.Industrial load refers to that industrial load number percent is greater than 75% load, agricultural load refers to that agricultural load number percent is greater than 75% load, Commercial Load refers to that Commercial Load number percent is greater than 75% load, resident load refers to appliance load number percent greater than 75% load, and the load in not being included in is classified as other loads.

(2) calculated load characteristic index

By above-mentioned collection data, calculate day maximum (little) load (P _Max/ P _Min), per day load (P _Av), daily load rate (K _d), day ratio of minimum load to maximum load (β _d), day peak-valley difference (H), day peak valley rate (α _d), month maximum (little) load (P _Max/ P _Min), the monthly average daily load

Deng, obtain following result:

Certain becomes the daily load characteristic index table 1

Certain becomes a month part throttle characteristics index table 2

(3) based on the load classification of part throttle characteristics index

On the basis of the redundant index proposition of finishing part throttle characteristics and selection, utilize the fuzzy C-means clustering method that the part throttle characteristics index is classified.Choose the daily load curve data as the proper vector of cluster, 32 transformer stations in certain city are classified.In order to make classification results more accurate, each transformer station's typical case's month workaday sampled point is gained merit data as the cluster feature vector, and the daily load curve data of 23 days regular working days are as proper vector.The note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, each sampling number certificate of constitutive characteristic vector is carried out normalized, x is then arranged _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

For 32 cluster numbers c of transformer station should be between 2 to 6 scope value (c 〉=2).When c value from 2 to 6 changed, the variation of formula (25) cluster validity functional value was as shown in table 5:

Table 3 cluster numbers and corresponding cluster validity functional value

c	2	3	4	5	6
						P'	0.7830	0.2370	0.3621	0.3913	0.4219

Therefore, according to last table result, when c got 2, the value maximum that cluster validity function is obtained should be divided into two classes.

As space is limited, the division matrix U of best cluster result is divided into U ₁, U ₂, U ₃List for three:

$U_1=\left[\begin{array}{ccccccccccc}0.315& 0.139& 0.744& 0.586& 0.289& 0.224& 0.133& 0.695& 0.476& 0.636& 0.143\\ 0.685& 0.861& 0.256& 0.414& 0.711& 0.776& 0.867& 0.305& 0.524& 0.364& 0.857\end{array}\right]$

$U_2=\left[\begin{array}{ccccccccccc}0.234& 0.165& 0.537& 0.545& 0.142& 0.147& 0.147& 0.087& 0.453& 0.158& 0.263\\ 0.766& 0.835& 0.463& 0.455& 0.858& 0.853& 0.853& 0.913& 0.547& 0.842& 0.737\end{array}\right]$

$U_3=\left[\begin{array}{cccccccccc}0.760& 0.648& 0.835& 0.803& 0.358& 0.353& 0.671& 0.620& 0.214& 0.785\\ 0.240& 0.352& 0.165& 0.197& 0.642& 0.647& 0.329& 0.380& 0.786& 0.215\end{array}\right]$

U=[U ₁，U ₂，U ₃]

By maximum membership grade principle, it is as follows to get net result:

I class={ x _k| k=3,4,8,10,14,15,23,24,25,26,29,30,32}

II class={ x _k| k=1,2,5,6,7,9,11,12,13,16,17,18,19,20,21,22,27,28,31}

(4) integrated load model determines

According to above-mentioned classification situation, choose typical transformer station, it is carried out the integrated load model modeling.

Set up integrated load model, shown in (35):

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.---\left(35\right)$

Owing to only calculate the load model of certain load bus herein, therefore do not consider power distribution network load model and reactive-load compensation load model.Then former integrated load model can be reduced to:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right.---\left(36\right)$

1) expression formula of static load model is:

$\left\{\begin{array}{c}{P}_{\mathrm{ZIP}}=p_1{\left(U\right)}^2+p_2\left(U\right)+p_3\\ Q_{\mathrm{ZIP}}=q_1{\left(U\right)}^2+q_2\left(U\right)+q_3\end{array}\right.---\left(37\right)$

In multinomial model, voltage magnitude U is the model input, and load active power P, reactive power Q are model output, p ₁, p ₂, p ₃, q ₁, q ₂, q ₃Be respectively active power, reactive power coefficient, i.e. parameter to be identified.

According to the implementation step in the summary of the invention as can be known, the objective function of identification static load model parameter is:

$\left\{\begin{array}{c}\min \underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}{(P_{\mathrm{ZIP}}-P)}^2\\ s.t.p_1+p_2{+p}_3-1=0\end{array}\right.---\left(38\right)$

Find the solution the optimum solution of objective function, in the time of can winning time iteration, the KKT condition of model:

$L_{\mathrm{ZIPx}}=\left[\begin{array}{c}8.3648\\ 0.0391\\ 0.002\end{array}\right]$

With the KKT conditional variant, writing matrix form, the update equation when obtaining iteration for the first time, following formula during for iteration for the first time the update equation matrix of coefficients and the identified parameters variable quantity after the first iteration:

$J=\left[\begin{array}{cccc}-4.1932& -0.0196& -0.0001& 0\\ -0.0196& 0.0001& 0& 0\\ -0.0001& 0& -0.0023& 0\\ 0& 0& 0& -0.0174\end{array}\right]$

${\mathrm{\&Delta;x}}_{\mathrm{ZIP}}=\left[\begin{array}{c}-2.007\\ 2.6280\\ -5.621\end{array}\right]$

Judge whether target function value satisfies iteration precision, and whether iterations exceeds set maximal value.Bring in constant renewal in model KKT condition, update equation matrix of coefficients and identified parameters variable quantity, up to obtaining optimum solution, withdraw from circulation.

Table 5 static load parameter identification result

p 1	p 2	p 3	q 1	q 2	q 3
						0.019	-0.03	2.138	0.0293	-0.049	2.903

2) dynamic load model tormulation formula is as follows:

$\left\{\begin{array}{c}{P}_M=u_d{i}_d+u_q{i}_q\\ Q_M=u_q{i}_d-u_d{i}_q\end{array}\right.---\left(39\right)$

i _d, i _qSolution formula as follows:

$\left\{\begin{array}{c}{i}_d=\frac{1}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}[R_s(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})+{\mathrm{\&omega;x}}^{\&prime;}(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ i_q=\frac{1}{R_s^2+{\left({\mathrm{\&omega;x}}^{\&prime;}\right)}^2}[R_s(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-{\mathrm{\&omega;x}}^{\&prime;}(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\end{array}\right.---\left(40\right)$

The objective function of identification dynamic load model parameter is:

$\min f_P\left(x\right)=\underset{k=1}{\overset{96}{\mathrm{\&Sigma;}}}{(P_M-P)}^2$

$s.t.h\left(x\right)=\left\{\begin{array}{c}{R}_s(u_d-e_{d0}^{\&prime;})+x^{\&prime;}(u_q-e_{q0}^{\&prime;})-(R_s^2+x^{\&prime;2})i_{d0}=0\\ R_s(u_q-e_{q0}^{\&prime;})-x^{\&prime;}(u_d-e_{d0}^{\&prime;})-(R_s^2+x^{\&prime;2})i_{q0}=0\\ e_{d0}^{\&prime;}++i_{q0}(x_s-x^{\&prime;})-(1-{\mathrm{\&omega;}}_0){T_0}^{\&prime;}{\mathrm{\&omega;}}_B{e}_{q0}^{\&prime;}=0\\ e_{q0}^{\&prime;}+i_{d0}(x_s-x^{\&prime;})+(1-{\mathrm{\&omega;}}_0){T_0}^{\&prime;}{\mathrm{\&omega;}}_B{e}_{d0}^{\&prime;}=0\\ T_L{{\mathrm{\&omega;}}_0}^n-e_{d0}^{\&prime;}{i}_{d0}-e_{q0}^{\&prime;}{i}_{q0}=0\end{array}\right.---\left(41\right)$

$g\left(x\right):0.08<x_s<0.2$

Find the solution the optimum solution of objective function, in the time of can winning time iteration, the KKT condition of model:

$L_{\mathrm{Mx}}=\left[\begin{array}{c}1.0745\\ 7.9865\\ 0\\ 0\\ 0.502\\ 1.9317\\ 0\\ 0.8933\\ -0.706\\ -1.3774\end{array}\right]$

With the KKT conditional variant, the writing matrix form, the update equation when obtaining iteration for the first time, following formula is the identified parameters variable quantity after the first iteration:

${\mathrm{\&Delta;x}}_M=\left[\begin{array}{c}-0.0004\\ 0.0158\\ 0.0002\\ 0.0005\\ 0.0359\\ 0.1447\\ 1.3898\\ 0.9633\\ -2.013\\ 0.5818\end{array}\right]$

Judge whether target function value satisfies iteration precision, and whether iterations exceeds set maximal value.Bring in constant renewal in model KKT condition, update equation matrix of coefficients and identified parameters variable quantity, up to obtaining optimum solution, withdraw from circulation.

Table 6 induction motor model parameter identification result

X s	X r	R s	R r	X m	H	A	B	C	T 0'
										0.1023	0.0508	1.0661	0.0362	0.0362	0.1095	1.6737	0.7718	-2.1091	0.3102

After obtaining the optimal value of model parameter, with the integrated load model that its substitution formula (36) can obtain determining, its expression formula is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\&omega;e}}_d}^{\&prime;}\right)+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\end{array}\right.---\left(42\right)$

The present invention adopts modern interior-point theory to carry out parameter identification.Modern interior-point theory is suitable for finding the solution advantages such as continuous large-scale nonlinear optimization problem, robustness is good, processing ill-conditioning problem ability is strong, is applicable to planning field, particularly nonlinear programming problem.Load model parameters is being carried out in the identification, can obtain the precise parameters identification result fast and effectively, thereby improving the accuracy of the load model of setting up.

One of ordinary skill in the art will appreciate that all or part of flow process that realizes in above-described embodiment method, be to instruct relevant hardware to finish by computer program, described program can be stored in the computer read/write memory medium, this program can comprise the flow process as the embodiment of above-mentioned each side method when carrying out.Wherein, described storage medium can be magnetic disc, CD, read-only storage memory body (Read-Only Memory, ROM) or at random store memory body (Random Access Memory, RAM) etc.

Above content be in conjunction with concrete preferred implementation to further describing that the present invention does, can not assert that concrete enforcement of the present invention is confined to these explanations.For the general technical staff of the technical field of the invention, without departing from the inventive concept of the premise, can also make some simple deduction or replace, all should be considered as belonging to protection scope of the present invention.

Claims (20)

1. one kind based on the touch upon load modeling method of modern interior-point theory of integrated information, it is characterized in that, comprising:

Add up the load data of a plurality of transformer stations;

According to the load data of statistics gained, calculate the part throttle characteristics index of each transformer station;

According to the part throttle characteristics index of described calculating gained, adopt clustering procedure that described each transformer station is classified, and choose typical transformer station;

For the typical transformer station that chooses sets up integrated load model;

Utilize the parameter to be asked of the described integrated load model of modern interior-point theory identification, set up the optimal synthesis load model.

2. as claimed in claim 1ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, for the typical transformer station that chooses sets up integrated load model, comprising:

Selected typical transformer station is set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations respectively;

To static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model, set up integrated load model.

3. as claimed in claim 2ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, utilize the parameter to be asked of the described integrated load model of modern interior-point theory identification, set up the optimal synthesis load model, comprising:

Utilize the modern interior-point theory parameter to be asked of the described static load model of identification, dynamic load model, power distribution network load model and reactive-load compensation load model respectively, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

With the described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that get access to wait ask the optimal value of parameter generation return in described static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model, so that static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized respectively;

Static load model, dynamic load model, power distribution network load model and reactive-load compensation load model after optimizing are sued for peace, set up the optimal synthesis load model.

4. as claimed in claim 3 based on the touch upon load modeling method of modern interior-point theory of integrated information, it is characterized in that the load data of a plurality of transformer stations of described statistics comprises:

Add up r transformer station's multiple spot every day charge value.

5. as claimed in claim 4ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, according to the load data of statistics gained, calculate the part throttle characteristics index of each transformer station, comprising:

According to r transformer station's multiple spot every day charge value of statistics, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

6. as claimed in claim 5ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, according to the part throttle characteristics index of described calculating gained, adopt clustering procedure that described each transformer station is classified, comprising:

Choose the daily load curve data as the proper vector of cluster;

According to described proper vector, utilize the fuzzy C-means clustering method that the part throttle characteristics index is classified;

Choose the typical transformer station in each classification.

7. as claimed in claim 6ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, choose the daily load curve data as the proper vector of cluster, comprising:

Choose each transformer station's meritorious data of typical case's month workaday sampled point as the cluster feature vector, and the daily load curve data of regular working day are as proper vector;

Each sampling number certificate to the constitutive characteristic vector is carried out normalized;

Concrete, the note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, to each sampling number of constitutive characteristic vector according to carrying out normalized be: x _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

8. as claimed in claim 7 based on the touch upon load modeling method of modern interior-point theory of integrated information, it is characterized in that described integrated load model expression formula is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D+P_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

9. as claimed in claim 8ly it is characterized in that based on the touch upon load modeling method of modern interior-point theory of integrated information, do not consider that the expression formula of integrated load model is under the situation of power distribution network load model and reactive-load compensation load model:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}.\right.$

10. as claimed in claim 8 based on the touch upon load modeling method of modern interior-point theory of integrated information, it is characterized in that the expression formula of described optimal synthesis load model is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\&omega;e}}_d}^{\&prime;}\right)+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\end{array}\right.;$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

11. one kind based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that, comprising:

The load data statistical module is for the load data of a plurality of transformer stations of statistics;

The characteristic index computing module is used for the load data according to described load data statistical module counts gained, calculates the part throttle characteristics index of each transformer station;

Transformer station's sort module is used for the part throttle characteristics index according to described characteristic index computing module calculating gained, adopts clustering procedure that described each transformer station is classified, and chooses typical transformer station;

Load model is set up module, and the typical transformer station that is used to described transformer station sort module to choose sets up integrated load model;

Load model is optimized module, utilizes the described load model of modern interior-point theory identification to set up the parameter to be asked of the integrated load model of module foundation, sets up the optimal synthesis load model.

12. as claimed in claim 11 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described load model is set up module, comprising:

Submodel is set up the unit, is used for selected typical transformer station is set up respectively static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model of all kinds of transformer stations;

Load model is set up the unit, is used for described submodel is set up static load model, dynamic load model, power distribution network load model and the summation of reactive-load compensation load model of setting up the unit, sets up integrated load model.

13. as claimed in claim 12 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described load model is optimized module, comprising:

The parameter identification unit, be used for utilizing modern interior-point theory respectively the described submodel of identification set up the parameter to be asked of static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model set up the unit, obtain the optimal value of waiting to ask parameter of described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model;

Submodel is optimized the unit, the optimal value of waiting to ask parameter that is used for static load model, dynamic load model, power distribution network load model and reactive-load compensation load model that described parameter identification unit is got access to respectively generation return described submodel and set up static load model, dynamic load model, power distribution network load model and the reactive-load compensation load model that the unit is set up, so that described static load model, dynamic load model, power distribution network load model and reactive-load compensation load model are optimized;

Unified model is optimized the unit, is used for setting up the optimal synthesis load model to suing for peace through static load model, dynamic load model, power distribution network load model and reactive-load compensation load model after the submodel optimization unit optimization.

14. as claimed in claim 13 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described load data statistical module specifically is used for r transformer station's multiple spot every day charge value of statistics.

15. it is as claimed in claim 14 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that, described characteristic index computing module specifically is used for: according to r transformer station's multiple spot every day charge value of described load data statistical module counts, calculate a day peak load P _MaxAnd/or minimum load P _Min, per day load P _Av, daily load rate K _d, day ratio of minimum load to maximum load β _d, day peak-valley difference H, day peak valley rate α _d, month peak load P _MaxAnd/or month minimum load P _Min, the monthly average daily load

16. as claimed in claim 15 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described transformer station sort module comprises:

Proper vector is chosen the unit, is used for choosing the daily load curve data as the proper vector of cluster;

The characteristic index taxon for the proper vector of choosing unit selection according to described proper vector, utilizes the fuzzy C-means clustering method that the part throttle characteristics index is classified;

Typical case transformer station chooses the unit, be used for choosing described characteristic index taxon sub-category electric typical transformer station.

17. as claimed in claim 16 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described proper vector is chosen the unit, comprising:

Proper vector is chosen subelement, be used for choosing each transformer station's meritorious data of typical case's month workaday sampled point as the cluster feature vector, and the daily load curve data of regular working day is as proper vector;

The normalized subelement is used for each sampling number certificate of constitutive characteristic vector is carried out normalized;

Concrete, the note peak load is P _Max, be P at h load constantly _h(h=1,2 ..., 2208), get P _MaxBe normalization factor, the normalized subelement to each sampling number of constitutive characteristic vector according to carrying out normalized is: x _h=P _h/ P _Max, x wherein _hFor after the normalization in h value constantly.

18. as claimed in claim 17 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that described load model is set up the integrated load model expression formula of setting up the unit and is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M+P_D{+P}_C\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M+Q_D+Q_C\end{array}\right.$

Wherein, P _SLMExpression integrated load model active power, P _ZIPExpression static load model active power, P _MExpression dynamic load model active power, P _DExpression power distribution network load model active power, P _CExpression reactive-load compensation load model active power part, P _C=0;

Q _SLMExpression integrated load model active power, Q _ZIPExpression static load model reactive power, Q _MExpression dynamic load model reactive power, Q _DExpression power distribution network load model reactive power, Q _CExpression reactive-load compensation load model active power part.

19. it is as claimed in claim 18 based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that, do not consider under the situation of power distribution network load model and reactive-load compensation load model that the expression formula that described load model is set up the integrated load model of setting up the unit is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=P_{\mathrm{ZIP}}+P_M\\ Q_{\mathrm{SLM}}=Q_{\mathrm{ZIP}}+Q_M\end{array}\right..$

20. as claim 18 or 19 described based on the touch upon load modeling system of modern interior-point theory of integrated information, it is characterized in that the expression formula that described unified model is optimized the optimal synthesis load model of setting up the unit is:

$\left\{\begin{array}{c}{P}_{\mathrm{SLM}}=0.019{\left(U\right)}^2-0.03\left(U\right)+2.138+u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661\left(u_d{{-\mathrm{\&omega;e}}_d}^{\&prime;}\right)+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ +u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\\ Q_{\mathrm{SLM}}=0.0293{\left(U\right)}^2-0.049\left(U\right)+2.903+u_q\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})+\mathrm{\&omega;}0.129(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})]\\ -u_d\frac{1}{1.1366+{\left(\mathrm{\&omega;}0.129\right)}^2}[1.0661(u_q-{{\mathrm{\&omega;e}}_q}^{\&prime;})-\mathrm{\&omega;}0.129(u_d-{{\mathrm{\&omega;e}}_d}^{\&prime;})]\end{array}\right.;$

Wherein, u _d, u _qRepresent the d axle of induction motor voltage, the coordinate components of q axle respectively; ω is actual angular frequency, and U is voltage magnitude; e _d', e _q' represent the d axle of induction motor transient potential, the coordinate components of q axle respectively.

Images