CN105186525A - Reactive voltage control partitioning method under wind power integration - Google Patents

Abstract

The invention discloses a reactive voltage control partitioning method under wind power integration. The method comprises the steps as follows: a whole network electrical distance expected matrix considering wind electricity power probability characteristics takes the place of an electrical distance matrix under one power flow section to be used as a partitioning basis; PQ node partitioning is achieved on the basis of AP clustering; the voltage regulation sensitivity of a PV node region considering the wind electricity power probability characteristics is defined by a perturbation method; PV nodes ranked on the basis of the region sensitivity are partitioned; the regional connectivity and controllability are firstly considered in node partitioning; and meanwhile, the condition that the PV nodes are classified into a PQ region which is the most sensitive in control can be ensured to the maximal extent. An index evaluation result shows that the partitioning scheme obtained by the method obtains a good effect; and an auxiliary reference can be provided for voltage control under wind power integration.

Description

Power Network Partitioning method under wind power integration

Technical field

The present invention relates to Power Network Partitioning method under wind power integration.

Background technology

Tertiary voltage control has become the universally recognized a kind of voltage control mode of electric power system, and practical application effect is good.And its optimized integration rational subregion that is system node.Therefore effective partition method is voltage-controlled important topic.

Conventional voltage subregion can be summarized as following five classes in method: clustering algorithm; Graph theory; Intelligent heuristics algorithm; A combined method; Take structure characteristic analysis as the additive method of representative.Existing partition method can be applicable to traditional electrical network preferably, but be applied to large-scale wind power access cause the sub area division of flow state change at random to face the challenge.

In grid nodes subregion conventional method based on the method for cluster because the advantage such as directly perceived, quick is widely used.The new forms of energy access electrical networks such as wind-powered electricity generation make the frequent change at random of electrical distance between the node based on sensitivity, bring difficulty to the application of traditional partition method.Existing document points out that sub area division requirement is as far as possible stable, to select and control strategy changes to reduce Pilot bus under different subregion.Therefore how to obtain that relatively stable to meet again the rationalization partition that voltage control requires in wind-powered electricity generation fluctuation situation be difficult point.It is PV node that existing document proposes wind-powered electricity generation node processing, is obtained and stablizes wind power output, and then take traditional fuzzy cluster subregion by meritorious expectation.It is simple that the method has process, the advantage that operand is little.But wind-powered electricity generation is as the unstable energy, asynchronous or double-fed blower fan all need absorb idle from system side and then set up magnetic field at present, therefore wind-powered electricity generation is treated to the PV node with voltage regulation capability and also not exclusively tallies with the actual situation; Simultaneously by the disposable impact asked for the meritorious fluctuation expecting that elimination wind power output fluctuation is difficult to embody electrical distance and cause subregion.In addition most literature all proposes the zoning requirements of the low coupling of subregion high cohesion, but fresh rare document carries out division result assessment with quantizating index.Document has the two stage partition method of reactive voltage of multiple target quantitative evaluation characteristic. Proceedings of the CSEE 2009,29 (16), propose five quantitative criterias first and make landmark breakthrough in subregion assessment, but its index definition based on meritorious phase angular sensitivity intuitively cannot reflect the regulating and controlling voltage ability of subregion; This index depends on the desirable number of partitions and region desired node number of specifying simultaneously, has certain subjectivity.Therefore the fluctuation problem brought of wind power integration and objective effective subregion evaluation index are the where the shoe pinches of subregion after electrical network access wind-powered electricity generation.

Summary of the invention

For solving the deficiency that prior art exists, the invention discloses Power Network Partitioning method under wind power integration, the present invention chooses AP cluster as core partition method.For the fluctuation problem that wind power integration brings, replace using the electrical distance matrix under single trend section as Regionalization basis with electrical distance expected matrix between the node considering wind power probability characteristics.Consider that PQ node is different from PV node response process, first based on AP clustering algorithm to PQ partition of nodes.Again based on the region voltage regulation and control sensitivity of each PV node during perturbation method definition consideration wind power probability characteristics to each PQ subregion, under the prerequisite ensureing the connectivity of region and controllability, realize PV node classification to suitable PQ subregion based on preferential sensitivity principle take into account region optimal voltage control simultaneously, complete the whole network subregion.Finally division result is assessed from coupling in the interval decoupling of cluster, district and subregion voltage control capability definition subregion quality evaluation index for carrying out objective effective subregion assessment.

For achieving the above object, concrete scheme of the present invention is as follows:

Power Network Partitioning method under wind power integration, comprises the following steps:

Step one: to consider that the whole network electrical distance expected matrix of wind power probability characteristics replaces electrical distance matrix under a certain trend section as Regionalization basis, realize PQ partition of nodes based on AP cluster;

Step 2: by the region voltage regulation and control sensitivity of each PV node during definition consideration wind-powered electricity generation probability characteristics to each PQ subregion, obtain PV partition of nodes data encasement;

Step 3: based on the PV partition of nodes of sensitivity sequence, first node division considers the connectivity of region and controllability, ensures PV node to sort out to control the sensitiveest PQ region to it simultaneously;

Step 4: from subarea management and the PV node voltage control ability two aspect definition subregion quality evaluation index of PQ node, comprise close coupling interval weak coupling index and PV node voltage in district and regulate and control sensitive index, quantification of targets zoning requirements, assesses partition scheme.

Further, the definition of electrical distance between PQ node:

Trend Jacobian matrix is utilized to define the internodal voltage sensibility of PQ as follows:

${\mathrm{\β}}_{ij}=\frac{\∂U_i}{\∂U_j}=\frac{\∂U_i}{\∂Q_j}\·\frac{\∂Q_j}{\∂U_j}=\frac{{\mathrm{\α}}_{ij}}{{\mathrm{\α}}_{jj}}---\left(1\right)$

In formula: β _ijfor voltage sensibility between node i and j; be N*N square formation, N is PQ nodes, $\left[\begin{array}{cc}{J}_{p\mathrm{\θ}}& J_{pv}\\ J_{q\mathrm{\θ}}& J_{qv}\end{array}\right]$ For trend Jacobian matrix, α _ijand α _jjthe capable j of i being respectively α arranges and the capable j column element of j;

AP clustering algorithm allows to adopt asymmetric electrical distance matrix as input, and between definition PQ node, electrical distance matrix is as follows:

In formula: N is the whole network PQ node number; D _ijrepresent that the electrical distance between arbitrary node i to node j is-lg| β _ij|.

Further, electrical distance expected matrix is set up under wind power integration:

Adopt discrete probability distribution to characterize wind-powered electricity generation probability characteristics, meritorious for wind-powered electricity generation historical sample of exerting oneself is added up, supposes that wind-powered electricity generation rated output is P _e, will exert oneself interval [0,100%P _e] discrete turn to f interval, gain merit sample of exerting oneself of statistics wind-powered electricity generation drops on the frequency in each interval, calculates the probability in each interval, gets power interval intermediate value successively and to exert oneself scene as each interval typical case, can obtain wind-powered electricity generation probability distribution;

All to think under this probability stable exerts oneself for each scene of exerting oneself of discretization gained, and wind-powered electricity generation permeability one timing, when wind-powered electricity generation scene is with the meritorious P that exerts oneself _k(k=1,2 ..., f) access electrical network, access point is treated to PQ node, and ask for mode by traditional electrical network electrical distance and obtain between PQ node that electrical distance matrix D (k) is such as formula shown in (3), corresponding probability is p _k;

D (k) in formula _ij(i ∈ [1, N], j ∈ [1, N]) expression wind power output is P _ktime, the electrical distance between node i and node j is D (k) _ij;

Between the PQ node under each scene, electrical distance asks for expectation, obtains to stablize electrical distance battle array ED and replace electrical distance matrix under a certain trend section as the Regionalization basis under wind power integration;

Wherein, ED is electrical distance expected matrix between the whole network PQ node, ED _ij(i ∈ [1, N], j ∈ [1, N]) represent to consider between node i and node j wind-powered electricity generation inject under electrical distance expect.

Further, PQ partition of nodes is realized based on AP cluster, using between PQ node, consider that the electrical distance expected matrix of wind-powered electricity generation probability characteristics is as input, the spacing of AP clustering algorithm defining node is more little more similar, therefore each for ED matrix element is got negative value and can obtain similarity matrix S, can automatically draw optimum cluster result with S battle array input AP clustering algorithm.

Further, between node, voltage sensibility affects by running status and network parameter, and PV node is to the regulating and controlling voltage sensitivity relation of PQ node:

F(i)·ΔV _PV(i)＝ΔV _PQ(i)(5)

In formula: Δ V _pV(i) and Δ V _pQbe illustrated respectively in PV node and PQ node voltage under running status i to change; F (i) is the sensitivity matrix under running status i;

Under running status i, as follows based on the regulating and controlling voltage sensitivity matrix of perturbation method definition M PV node to N number of PQ node:

In formula, F (i) arbitrary element wherein Δ V _pV(i) _ywith Δ V _pQ(i) _xbe illustrated respectively in the voltage variety of PV node y voltage Perturbation and corresponding PQ node x under running status i.

The regulating and controlling voltage sensitivity under wind power integration of same PV node will present fluctuation, and under expecting to characterize consideration wind power probability characteristics with regulating and controlling voltage sensitivity, PV node is to the ability of regulation and control of each PQ node.

Further, wind-powered electricity generation statistical probability distribution rule definition:

In formula: F arbitrary element under representing wind power integration, PV node x is to the regulating and controlling voltage sensitivity of PQ node y.

The initial condition sorted out due to PV node is L PQ subregion, and define the region voltage regulation and control sensitivity of PV node to PQ subregion is the average that this PV node is expected node voltage regulation and control sensitivity all in this PQ subregion for this reason, definition:

In formula: G arbitrary element represent that PV node x is to PQ subregion Ω _yregion voltage regulation and control sensitivity; R is Ω _yinterior arbitrary PQ node number; n _yfor Ω _yin contained PQ nodes.

Further, PV partition of nodes can realize on its region voltage to each PQ subregion regulation and control sensitivity basis.First node division considers the connectivity of region and controllability, ensures PV node to sort out as far as possible to control the sensitiveest PQ region to it simultaneously, and detailed process is as follows:

(1) by the region voltage regulation and control sensitivity sequence of all PV nodes to first PQ subregion, when ensureing connective, the sensitiveest PV node merger Ru Gai district is selected; Reactive source node selection is in like manner carried out in remaining region, the PV node that the forefoot area should got rid of when at every turn choosing PV node had been selected, ensures that each PQ subregion all has a reactive source node behind first subzone, ensures subregion controllability;

(2) by the PV node sequencing of all the other non-merger, by single PV node to the region voltage of all PQ subregions regulation and control sensitivity sequence, when ensureing connective by this PV node division to the highest PQ subregion of sensitivity.Complete all PV node division successively.

Voltage partition lacks quantification of targets assessment subregion effect, therefore the quality of division result is quantized by the Silhoutte index defined based on electrical distance, to assess division result quality.

Further, based under Silhouttte index and wind power integration between PQ node electrical distance expect that the interval weak coupling index of close coupling is as follows in definition:

${\mathrm{QNJ}}_i=\frac{1}{n_i}\underset{t=1}{\overset{n}{\Σ}}(\frac{b_i\left(t\right)-a_i\left(t\right)}{max\{a_i\left(t\right),b_i\left(t\right)\}}\·C_i)(i=1,2,...,L)---\left(9\right)$

$QNJ=\frac{1}{L}\underset{i=1}{\overset{L}{\Σ}}{\mathrm{QNJ}}_i---\left(10\right)$

In formula: QNJ _irepresent the coupling index of i-th PQ subregion; QNJ represents the whole network PQ subarea management index; a _it () represents all PQ node electrical distance averages in subregion i interior nodes t and district; b _it () represents that subregion i interior nodes t is to PQ node electrical distance averages all outside district; C _ibe used to indicate PQ subregion i whether to there is node and pass through situation, when exist one or more node not with it around any node divide to same district, then there is isolated node or there is reachability problem in subregion, this season make QNJ=-1; Otherwise C _i=1, n _ifor PQ nodes in subregion i; L is the PQ number of partitions; QNJ _iand the value of QNJ is all between [-1,1].

Further, PV node voltage regulates and controls sensitive index:

PV node divides in best PQ region with regulating and controlling voltage sensitivity, ideally, each PV node with peak response control region interior nodes voltage, simultaneously perturbing area exterior node voltage hardly, meaning is set out thus, controls sensitive indices P VC as follows based on perturbation method definition PV node voltage:

${\mathrm{PVC}}_j=\frac{\underset{p\∈{\mathrm{\Ω}}_j}{\Σ}\left|{\mathrm{\ΔU}}_p\right|}{\underset{q\∈\mathrm{\Ω}}{\Σ}\left|{\mathrm{\ΔU}}_q\right|}\·C_{{\mathrm{\Ω}}_j},(j=1,2,...,M)---\left(11\right)$

$PVC=\frac{1}{M}\underset{j=1}{\overset{M}{\Σ}}{\mathrm{PVC}}_j---\left(12\right)$

In formula: PVC _jrepresent the sensitive index of regulating and controlling voltage of PV node j; PVC represents that the whole network PV node voltage regulates and controls sensitive index; Ω _jfor PQ node set in PV node j affiliated area, Ω is that the whole network PQ gathers, scalar region Ω _jinterior PV distributing equilibrium degree parameter, when this district PV nodes be greater than 0 and PV node and its PQ node that is directly connected divide to during same district then otherwise due to wind power integration, under different scene, PV node voltage controls sensitivity difference, | Δ U _p| with | Δ U _q| be respectively the voltage magnitude perturbation absolute value of PQ node p and q, adopt after wind-powered electricity generation injects and expect voltage deviation process fluctuation problem, namely Δ U _pi: wind-powered electricity generation scene i accesses the voltage magnitude Perturbation of lower PQ node p; Δ U _qiwind-powered electricity generation scene i accesses the voltage magnitude Perturbation of lower PQ node q, and Pi is the probability of wind-powered electricity generation scene i.

Beneficial effect of the present invention:

Herein for the sub area division in access wind-powered electricity generation situation, to consider that the electrical distance expected matrix of wind power probability characteristics replaces electrical distance matrix under single trend section and can overcome as Regionalization basis the fluctuation problem that wind power integration brings.The quantitative evaluation being voltage partition from PQ partition of nodes coupling and the objective subregion project evaluation chain index of PV node voltage ability of regulation and control definition two provides effective reference.The partition scheme that index evaluation result display institute extracting method obtains obtains good result, and under can be access wind-powered electricity generation, voltage control provides auxiliary reference.

Accompanying drawing explanation

Fig. 1 NewEngland39 node system;

Fig. 2 NewEngland39 node system division result figure;

Embodiment:

Below in conjunction with accompanying drawing, the present invention is described in detail:

PQ partition of nodes based on AP clustering algorithm:

First, the PQ node electrical distance expected matrix of wind power probability characteristics is introduced

Characterize with electrical distance the method contacting tightness between node under traditional electrical pessimistic concurrency control very ripe, deterministic electrical distance matrix between PQ node can be obtained by Jacobian matrix when running status and network parameter are determined.Under causing consolidated network structure after wind power integration electrical network, electrical distance matrix has uncertainty.For this reason, propose to consider that the whole network electrical distance expected matrix of wind power probability characteristics replaces electrical distance matrix under a certain trend section as Regionalization basis herein.Process with this fluctuation problem acquisition brought after wind-powered electricity generation injects and stablize subregion.

Wherein, the definition of electrical distance between PQ node:

The method that Jacobian matrix based on Load flow calculation gained obtains voltage sensibility and then definition electrical distance has all been applied and has obtained good result in many documents.

The internodal voltage sensibility of existing document utilization trend Jacobian matrix definition PQ is as follows:

${\mathrm{\β}}_{ij}=\frac{\∂U_i}{\∂U_j}=\frac{\∂U_i}{\∂Q_j}\·\frac{\∂Q_j}{\∂U_j}=\frac{{\mathrm{\α}}_{ij}}{{\mathrm{\α}}_{jj}}---\left(1\right)$

In formula: β _ijfor voltage sensibility between node i and j; be N*N square formation, N is PQ nodes. $\left[\begin{array}{cc}{J}_{p\mathrm{\θ}}& J_{pv}\\ J_{q\mathrm{\θ}}& J_{qv}\end{array}\right]$ For trend Jacobian matrix.α _ijand α _jjthe capable j of i being respectively α arranges and the capable j column element of j.

AP clustering algorithm allows to adopt asymmetric electrical distance matrix as input, defines electrical distance matrix between PQ node herein as follows:

In formula: N is the whole network PQ node number; D _ijrepresent the electrical distance between arbitrary node i to node j.

Secondly, electrical distance expected matrix is set up under wind power integration:

The direct basis of subregion is electrical distance matrix, it is non-linear relation between node power and electrical distance, the electrical distance considering wind-powered electricity generation probability characteristics is adopted to expect, the impact of wind-powered electricity generation fluctuation on electrical distance can be described more accurately, directly portray the factor directly perceived affecting subregion, make electrical distance expected matrix gained partition scheme used have stronger adaptability to wind power integration.

Adopt discrete probability distribution to characterize wind-powered electricity generation probability characteristics, meritorious for wind-powered electricity generation historical sample of exerting oneself is added up.Suppose that wind-powered electricity generation rated output is P _e, will exert oneself interval [0,100%P _e] discrete turn to f interval.Gain merit sample of exerting oneself of statistics wind-powered electricity generation drops on the frequency in each interval, calculates the probability in each interval.Get power interval intermediate value successively to exert oneself scene as each interval typical case, wind-powered electricity generation probability distribution can be obtained.

All to think under this probability stable exerts oneself for each scene of exerting oneself of discretization gained.Wind-powered electricity generation permeability one timing, when wind-powered electricity generation scene is with the meritorious P that exerts oneself _k(k=1,2 ..., f) access electrical network, access point is treated to PQ node.Ask for mode by traditional electrical network electrical distance and obtain between PQ node that electrical distance matrix D (k) is such as formula shown in (3), corresponding probability is p _k.

D (k) in formula _ij(i ∈ [1, N], j ∈ [1, N]) expression wind power output is P _ktime, the electrical distance between node i and node j is D (k) _ij.

Between the PQ node under each scene, electrical distance asks for expectation, obtains to stablize electrical distance battle array ED and replace electrical distance matrix under a certain trend section as the Regionalization basis under wind power integration.

Wherein, ED is electrical distance expected matrix between the whole network PQ node.ED _ij(i ∈ [1, N], j ∈ [1, N]) represent to consider between node i and node j wind-powered electricity generation inject under electrical distance expect.

PQ partition of nodes is realized based on AP cluster:

AP cluster is a kind of unmanned Supervised Clustering Methods be newly published in 2007 on science.Algorithm only with similarity matrix between domain node for input.Using between PQ node, consider that the electrical distance expected matrix of wind-powered electricity generation probability characteristics is as input, the spacing of AP clustering algorithm defining node is more little more similar, therefore each for ED matrix element is got negative value and can obtain similarity matrix S.Optimum cluster result can be automatically drawn with S battle array input AP clustering algorithm.Algorithm details are see document: FreyBJ, DueckD.Clusteringbypassingmessagesbetweendatapoints [J] .science, 2007,315 (5814): 972-976..

Next step, by the region voltage regulation and control sensitivity of each PV node during definition consideration wind-powered electricity generation probability characteristics to each PQ subregion, obtains PV partition of nodes data encasement.

Based on the PV partition of nodes of region voltage regulation and control sensitivity sequence under wind power integration:

Need after completing PQ partition of nodes to carry out subregion classification to PV node, each subregion reactive source is evenly distributed and ensures that each reactive source controls the optimal voltage of each PQ subregion as far as possible when ensureing connective.

Voltage sensibility between the PV node region regulating and controlling voltage sensitivity nodes of perturbation method definition consideration wind-powered electricity generation probability characteristics is utilized to affect by running status and network parameter, white according to document (YUN will, Liu Yutian, Liang Jun, Deng. consider the Pilot bus system of selection [J] of wind power fluctuation probability characteristics. Automation of Electric Systems, 2014,38 (9): 20 ~ 25.) there is the regulating and controlling voltage sensitivity relation of following PV node to PQ node:

F(i)·ΔV _PV(i)＝ΔV _PQ(i)(5)

In formula: Δ V _pV(i) and Δ V _pQbe illustrated respectively in PV node and PQ node voltage under running status i to change; F (i) is the sensitivity matrix under running status i.

Under running status i, as follows based on the regulating and controlling voltage sensitivity matrix of perturbation method definition M PV node to N number of PQ node:

In formula, F (i) arbitrary element wherein Δ V _pV(i) _ywith Δ V _pQ(i) _xbe illustrated respectively in the voltage variety of PV node y voltage Perturbation and corresponding PQ node x under running status i.

The regulating and controlling voltage sensitivity under wind power integration of same PV node will present fluctuation.Under expecting to characterize consideration wind power probability characteristics with regulating and controlling voltage sensitivity, PV node is to the ability of regulation and control of each PQ node.

Based on the definition of wind-powered electricity generation statistical probability distribution rule:

In formula: F arbitrary element under representing wind power integration, PV node x is to the regulating and controlling voltage sensitivity of PQ node y.

The initial condition sorted out due to PV node is L PQ subregion, and define the region voltage regulation and control sensitivity of PV node to PQ subregion is the average that this PV node is expected node voltage regulation and control sensitivity all in this PQ subregion for this reason.Definition:

In formula: G arbitrary element represent that PV node x is to PQ subregion Ω _yregion voltage regulation and control sensitivity; R is Ω _yinterior arbitrary PQ node number; n _yfor Ω _yin contained PQ nodes.

PV partition of nodes based on sensitivity sequence:

PV partition of nodes can realize on its region voltage to each PQ subregion regulation and control sensitivity basis.First node division considers the connectivity of region and controllability, ensures PV node to sort out as far as possible to control the sensitiveest PQ region to it simultaneously.Detailed process is as follows:

(1) by the region voltage regulation and control sensitivity sequence of all PV nodes to first PQ subregion, when ensureing connective, the sensitiveest PV node merger Ru Gai district is selected; Reactive source node selection is in like manner carried out in remaining region, the PV node that the forefoot area should got rid of when at every turn choosing PV node had been selected.Ensure behind first subzone that each PQ subregion all has a reactive source node, ensure subregion controllability.

(2) by the PV node sequencing of all the other non-merger.By single PV node to the region voltage of all PQ subregions regulation and control sensitivity sequence, when ensureing connective by this PV node division to the highest PQ subregion of sensitivity.Complete all PV node division successively.

PQ node is the most node of electrical network, first carries out PQ partition of nodes be conducive to giving partitioning algorithm enough information to draw the rational number of partitions when number of partitions the unknown.AP clustering algorithm can ensure that PQ partition of nodes is connective simultaneously.First carry out PQ subregion in PV node merge process and select reactive source node, ensureing to make each subregion all containing at least one reactive source node, to meet voltage control requirement under connective prerequisite.

Subregion quality evaluation index:

Lot of documents provides the requirement of voltage partition, but fresh rare document proposes the quantizating index weighing subregion effect.First voltage partition will ensure that subregion is connective; Require subregion inner close coupling by stages Approximate Decoupling simultaneously; PV node should ensure district's interior nodes voltage control the strongest simultaneously minimum on the impact of district's exterior node when carrying out regulating and controlling voltage.For this reason herein from subarea management and the PV node voltage control ability two aspect definition subregion quality evaluation index of PQ node.Quantification of targets zoning requirements, can carry out objective evaluation to partition scheme.

The interval weak coupling index of close coupling in PQ node area:

PQ node is the most nodes in electrical network, thus can assess the coupling of final subregion based on the coupling power of PQ partition of nodes.Cluster Silhouttte index effectively can reflect compactness and interval separability in cluster district, therefore based under Silhouttte index and wind power integration between PQ node electrical distance expect that the interval weak coupling index of close coupling is as follows in definition:

${\mathrm{QNJ}}_i=\frac{1}{n_i}\underset{t=1}{\overset{n_i}{\Σ}}(\frac{b_i\left(t\right)-a_i\left(t\right)}{max\{a_i\left(t\right),b_i\left(t\right)\}}\·C_i),(i=1,2,...,L)---\left(9\right)$

$QNJ=\frac{1}{L}\underset{i=1}{\overset{L}{\Σ}}{\mathrm{QNJ}}_i---\left(10\right)$

In formula: QNJ _irepresent the coupling index of i-th PQ subregion; QNJ represents the whole network PQ subarea management index; a _it () represents all PQ node electrical distance averages in subregion i interior nodes t and district; b _it () represents that subregion i interior nodes t is to PQ node electrical distance averages all outside district; C _ibe used to indicate PQ subregion i whether to there is node and pass through situation, when exist one or more node not with it around any node divide to same district, then there is isolated node or there is reachability problem in subregion, this season make QNJ=-1; Otherwise C _i=1.N _ifor PQ nodes in subregion i; L is the PQ number of partitions;

QNJ _iand the value of QNJ is all between [-1,1].Index is with QNJ large small quantization the whole network coupling, and in the district of the better i.e. the whole network partition scheme of its value larger expression subarea management, the stronger interval coupling simultaneously of coupling is more weak.When the QNJ index of different schemes is close, each subregion QNJ under same scheme _iwhen fluctuating little, illustrate that each subarea management level is similar to, overall plan is more reasonable.

PV node voltage regulates and controls sensitive index:

PV node divides in best PQ region with regulating and controlling voltage sensitivity.Ideally, each PV node with peak response control region interior nodes voltage, simultaneously perturbing area exterior node voltage hardly.Meaning is set out thus, controls sensitive indices P VC as follows based on perturbation method definition PV node voltage:

${\mathrm{PVC}}_j=\frac{\underset{p\∈{\mathrm{\Ω}}_j}{\Σ}\left|{\mathrm{\ΔU}}_p\right|}{\underset{q\∈\mathrm{\Ω}}{\Σ}\left|{\mathrm{\ΔU}}_q\right|}\·C_{{\mathrm{\Ω}}_j},(j=1,2,...,M)---\left(11\right)$

$PVC=\frac{1}{M}\underset{j=1}{\overset{M}{\Σ}}{\mathrm{PVC}}_j---\left(12\right)$

In formula: PVC _jrepresent the sensitive index of regulating and controlling voltage of PV node j; PVC represents that the whole network PV node voltage regulates and controls sensitive index; Ω _jfor PQ node set in PV node j affiliated area, Ω is that the whole network PQ gathers. scalar region Ω _jinterior PV distributing equilibrium degree parameter, when this district PV nodes be greater than 0 and PV node and its PQ node that is directly connected divide to during same district then otherwise due to wind power integration, under different scene, PV node voltage controls sensitivity difference.| Δ U _p| with | Δ U _q| be respectively the voltage magnitude perturbation absolute value of PQ node p and q.Adopt after wind-powered electricity generation injects and expect voltage deviation process fluctuation problem, namely $\left|{\mathrm{\ΔU}}_p\right|=\underset{i=1}{\overset{f}{\Σ}}\left|{\mathrm{\ΔU}}_{pi}\right|\·p_i,\left|{\mathrm{\ΔU}}_q\right|=\underset{i=1}{\overset{f}{\Σ}}\left|{\mathrm{\ΔU}}_{qi}\right|\·p_i.$

Index proposes based on perturbation method.When a jth PV node voltage perturbs, in this PV affiliated area, all with the ratio of the whole network PQ node voltage increment absolute value sum, PQ node voltage increment absolute value sum can reflect that this PV is to one's respective area voltage control capability and perturbing area external voltage ability.M PV node control degree average can reflect the whole network all PV nodes integrated voltage control ability.PVC _jwith PVC value all between [-1,1], close to 1, desired value more represents that subregion voltage control sensitivity is higher.When different schemes PVC index is close, under same scheme, each PV node voltage regulates and controls sensitive indices P VC _jless expression each reactive source node ability of regulation and control that fluctuates is close, and overall plan regulating and controlling voltage is more excellent.

Subregion assessment, respectively from subarea management and the PV node voltage ability of regulation and control of PQ node, takes into account and considers the factors such as whether connective the and PV Node distribution in partition of nodes even.Owing to adopting AP clustering algorithm can automatically determine optimum partition number, therefore think that gained cluster has met desirable number of partitions index herein.Document (Wang Kaijun, Zhang Junying, Li Dan, etc. self adaptation affine propagation clustering [J]. automation journal, 2007,33 (12): 1242 ~ 1246.) and when pointing out N number of some cluster, the rational optimum clustering number upper limit is simulation result shows that the AP cluster number of partitions meets this requirement, therefore also reflects that AP cluster is applied to the validity of grid nodes subregion.Define two indexs herein and substantially contain document (Chen Xia; Sun Haishun; Sui Xianchao; Deng. a kind of region couples degree index and the application study in voltage power-less zonal control [J] thereof. protecting electrical power system and control; 2011,39 (7): 83 ~ 88.) all factors of index evaluation in.Outstanding assessment subregion voltage control capability, and objective quantification assessment can not carried out to division result containing artificial subjective factor.

Sample calculation analysis: with the node system of NewEngland39 shown in Fig. 1 for analogue system, random selecting No. 12 nodes carry out wind power integration.Added wind-powered electricity generation exerts oneself sampled data for 1 year for sample with Ji NORTEL net wind field, and the sampling interval is 5min, and specified gaining merit is exerted oneself as 200MW.It is PQ node that wind-powered electricity generation injects node processing, and No. 31 balance node do not participate in subregion and are directly divided to the PQ node place subregion be directly connected.

PQ partition of nodes: first NORTEL net wind field 1 year meritorious historical sample point of exerting oneself of wind-powered electricity generation in Ji is added up, obtain wind-powered electricity generation probability distribution.Sample presents less and 0 sample of exerting oneself of meritorious larger respective frequencies of exerting oneself and occurs than feature more frequently.Therefore adopt 0 to exert oneself and add up the interval division mode that when exerting oneself large, interval is slightly large separately.During owing to exerting oneself large, probability is less thus very little on the impact of electrical distance expected matrix, therefore is merged by the power interval after rated output 40%, and forming four discretization power interval is 0%P _e, (0%P _e, 20%P _e], (20%P _e, 40%Pe], (40%P _e, 100%P _e].Add up gain merit sample of exerting oneself of annual wind-powered electricity generation and drop on the frequency in each interval, and calculate the probability in each interval.Get four typical cases scene of exerting oneself and be respectively each interval intermediate value.Statistical probability is as shown in table 1.When calculating wind-powered electricity generation permeability is 50%, each scene wind power output.Each generator output is kept to account for the power division of total load ratio when wind-powered electricity generation injects successively according to scene constant.

Table 1 output of wind electric field statistical probability

To calculate under each scene electrical distance matrix between PQ node respectively, and ask for electrical distance between the whole network PQ node according to probability and expect.In this, as the input of AP cluster, show that the whole network PQ node clustering result is

{1,2,3,25},{4,5,6,7,8,9,10,11,12,13,14},{15,16,17,18,21,22,23,24,27},{19,20},{26,28,29}。There is not reachability problem in division result display PQ node clustering; When document [26] is pointed out N number of some cluster, the rational optimum cluster upper limit is cluster and N=29 is carried out, gained cluster numbers for the whole network 29 PQ nodes aP cluster meets this requirement well.Therefore carry out PQ partition of nodes based on AP cluster and can automatically draw rationalization partition number, and there is not reachability problem.

PV partition of nodes: after PQ partition of nodes completes, then based on the region voltage regulation and control sensitivity of perturbation method definition PV node to each PQ subregion, to have sorted PV partition of nodes based on sensitivity.Calculate the region voltage regulation and control sensitivity of each PV node to each PQ subregion under considering wind power probability characteristics as shown in table 2.PV partition of nodes domain is 9 reactive source nodes except balance node.First traveling through successively each PQ subregion, making each PQ subregion select the sensitiveest PV node successively when ensureing connective.As shown in table 2, subregion 1 chooses No. 30 nodes, and subregion 2 chooses No. 32 the sensitiveest nodes in remaining reactive source node, and remaining subregion chooses 35,33, No. 38 nodes successively.Now each subregion all divides 1 PV node and can ensure region voltage controllability.After its sensitivity to each PQ subregion is sorted by remaining 4 PV nodes successively, this PV node is divided in the maximum PQ region of sensitivity.Complete the whole network subregion thus, division result is as shown in table 3.After subregion completes, each subregion is all containing at least 1 PV node, voltage controllability and subregion connectedness all satisfied.

The sensitivity of table 2PV node region regulating and controlling voltage

Table 3NewEngland39 system whole-network division result

System partitioning schematic diagram as shown in Figure 2.As seen from Figure 2.There is not reachability problem in division result, and each PV node also divides to its direct-connected PQ node region.Gained division result and document [24] result are similar to, and only part of nodes subregion is different, will compare both division result below by quantification of targets.

Table 4PVC quantitative evaluation results contrast

Subregion based on quantizating index is assessed:

After the whole network subregion completes, be quantitative evaluation division result, calculate PVC and the QNJ index of partition scheme herein respectively as shown in table 4 and table 5.Division result and document [24] division result will contrast herein, calculate document [24] index respectively as shown in table 4 Yu table 6.Stiffness of coupling and interval decoupling zero degree in the district that QNJ index indicates subregion, its value is between [-1,1], and index to show more greatly in cluster district that coupling is stronger and is intervally coupled more weak, and namely clustering result quality is good.Result display herein partition scheme the whole network QNJ index is higher, and each subregion index more evenly fluctuates not quite, does not occur negative value, therefore the interval weak coupling index of subregion close coupling in district herein obtains good result.PVC index expression PV node is to region voltage control sensitivity.Subregion overall PV node control is highly sensitive herein in result display, and each PV control effects more evenly fluctuates not quite, and comparatively document [24] partition scheme obtains better effect.

Table 5 is QNJ quantitative evaluation result herein

Table 6 document [24] QNJ quantitative evaluation result

Document [24]: Qiao Liang, Lu Jiping, Huang Hui, etc. containing the learning algorithms partition method [J] of wind field. electric power network technique, 2010,34 (10): 163 ~ 168.

Wind power integration causes network operation state to have stochastic volatility, consider that wind-powered electricity generation injects electrical distance expected matrix between lower the whole network node and replaces electrical distance matrix under single trend section as Regionalization basis, obtain and can adapt to the stable the whole network subregion of various wind power output.Consider that PQ node is different from PV node response process, first based on AP clustering algorithm, subregion is carried out to PQ node; Then with PV node for domain, under the prerequisite ensureing the connectivity of region and controllability, realize PV node based on preferential sensitivity principle sort out and take into account region optimal voltage to suitable PQ subregion simultaneously and control, finally complete the whole network subregion.Finally from coupling and the requirement of subregion voltage control capability in the interval decoupling of cluster, district, definition subregion quality evaluation index, objective evaluation subregion quality.Simulation result shows feasibility and the validity of institute's extracting method herein.

By reference to the accompanying drawings the specific embodiment of the present invention is described although above-mentioned; but not limiting the scope of the invention; one of ordinary skill in the art should be understood that; on the basis of technical scheme of the present invention, those skilled in the art do not need to pay various amendment or distortion that creative work can make still within protection scope of the present invention.

Claims (10)

1. Power Network Partitioning method under wind power integration, is characterized in that, comprise the following steps:

Step one: to consider that the whole network electrical distance expected matrix of wind power probability characteristics replaces electrical distance matrix under a certain trend section as Regionalization basis, realize PQ partition of nodes based on AP cluster;

Step 2: by the region voltage regulation and control sensitivity of each PV node during definition consideration wind-powered electricity generation probability characteristics to each PQ subregion, obtain PV partition of nodes data encasement;

Step 3: based on the PV partition of nodes of sensitivity sequence, first node division considers the connectivity of region and controllability, ensures PV node to sort out to control the sensitiveest PQ region to it simultaneously;

Step 4: from subarea management and the PV node voltage control ability two aspect definition subregion quality evaluation index of PQ node, comprise close coupling interval weak coupling index and PV node voltage in district and regulate and control sensitive index, quantification of targets zoning requirements, assesses partition scheme.

2. Power Network Partitioning method under wind power integration as claimed in claim 1, is characterized in that, the definition of electrical distance between PQ node:

Trend Jacobian matrix is utilized to define the internodal voltage sensibility of PQ as follows:

${\mathrm{\β}}_{ij}=\frac{\∂U_i}{\∂U_j}=\frac{\∂U_i}{\∂Q_j}\·\frac{\∂Q_j}{\∂U_j}=\frac{{\mathrm{\α}}_{ij}}{{\mathrm{\α}}_{jj}}---\left(1\right)$

In formula: β _ijfor voltage sensibility between node i and j; be N*N square formation, N is PQ nodes; $\left[\begin{array}{cc}{J}_{p\mathrm{\θ}}& J_{pv}\\ J_{q\mathrm{\θ}}& J_{qv}\end{array}\right]$ For trend Jacobian matrix, α _ijand α _jjthe capable j of i being respectively α arranges and the capable j column element of j;

AP clustering algorithm allows to adopt asymmetric electrical distance matrix as input, and between definition PQ node, electrical distance matrix is as follows:

In formula: N is the whole network PQ node number; D _ijrepresent that the electrical distance between arbitrary node i to node j is-lg| β _ij|.

3. Power Network Partitioning method under wind power integration as claimed in claim 2, is characterized in that, set up electrical distance expected matrix under wind power integration:

Adopt discrete probability distribution to characterize wind-powered electricity generation probability characteristics, meritorious for wind-powered electricity generation historical sample of exerting oneself is added up, supposes that wind-powered electricity generation rated output is P _e, will exert oneself interval [0,100%P _e] discrete turn to f interval, gain merit sample of exerting oneself of statistics wind-powered electricity generation drops on the frequency in each interval, calculates the probability in each interval, gets power interval intermediate value successively and to exert oneself scene as each interval typical case, can obtain wind-powered electricity generation probability distribution;

All to think under this probability stable exerts oneself for each scene of exerting oneself of discretization gained, and wind-powered electricity generation permeability one timing, when wind-powered electricity generation scene is with the meritorious P that exerts oneself _k(k=1,2 ..., f) access electrical network, access point is treated to PQ node, and ask for mode by traditional electrical network electrical distance and obtain between PQ node that electrical distance matrix D (k) is such as formula shown in (3), corresponding probability is p _k;

D (k) in formula _ij(i ∈ [1, N], j ∈ [1, N]) expression wind power output is P _ktime, the electrical distance between node i and node j is D (k) _ij;

Between the PQ node under each scene, electrical distance asks for expectation, obtains to stablize electrical distance battle array ED and replace electrical distance matrix under a certain trend section as the Regionalization basis under wind power integration;

Wherein, ED is electrical distance expected matrix between the whole network PQ node, ED _ij(i ∈ [1, N], j ∈ [1, N]) represent to consider between node i and node j wind-powered electricity generation inject under electrical distance expect.

4. Power Network Partitioning method under wind power integration as claimed in claim 1, it is characterized in that, PQ partition of nodes is realized based on AP cluster, using between PQ node, consider that the electrical distance expected matrix of wind-powered electricity generation probability characteristics is as input, the spacing of AP clustering algorithm defining node is more little more similar, therefore each for ED matrix element is got negative value and can obtain similarity matrix S, can automatically draw optimum cluster result with S battle array input AP clustering algorithm.

5. Power Network Partitioning method under wind power integration as claimed in claim 1, it is characterized in that, between node, voltage sensibility affects by running status and network parameter, and PV node is to the regulating and controlling voltage sensitivity relation of PQ node:

F(i)·ΔV _PV(i)＝ΔV _PQ(i)(5)

In formula: Δ V _pV(i) and Δ V _pQbe illustrated respectively in PV node and PQ node voltage under running status i to change; F (i) is the sensitivity matrix under running status i;

Under running status i, as follows based on the regulating and controlling voltage sensitivity matrix of perturbation method definition M PV node to N number of PQ node:

In formula, F (i) arbitrary element wherein Δ V _pV(i) _ywith Δ V _pQ(i) _xbe illustrated respectively in the voltage variety of PV node y voltage Perturbation and corresponding PQ node x under running status i;

The regulating and controlling voltage sensitivity under wind power integration of same PV node will present fluctuation, and under expecting to characterize consideration wind power probability characteristics with regulating and controlling voltage sensitivity, PV node is to the ability of regulation and control of each PQ node.

6. Power Network Partitioning method under wind power integration as claimed in claim 1, is characterized in that, the definition of wind-powered electricity generation statistical probability distribution rule:

In formula: F arbitrary element under representing wind power integration, PV node x is to the regulating and controlling voltage sensitivity of PQ node y.

7. Power Network Partitioning method under wind power integration as claimed in claim 1, it is characterized in that, the initial condition sorted out due to PV node is L PQ subregion, define the region voltage regulation and control sensitivity of PV node to PQ subregion is the average that this PV node is expected node voltage regulation and control sensitivity all in this PQ subregion for this reason, definition:

In formula: G arbitrary element represent that PV node x is to PQ subregion Ω _yregion voltage regulation and control sensitivity; R is Ω _yinterior arbitrary PQ node number; n _yfor Ω _yin contained PQ nodes.

8. Power Network Partitioning method under wind power integration as claimed in claim 1, it is characterized in that, PV partition of nodes can realize on its region voltage to each PQ subregion regulation and control sensitivity basis, first node division considers the connectivity of region and controllability, ensure PV node to sort out as far as possible to control the sensitiveest PQ region to it, detailed process is as follows simultaneously:

(1) by the region voltage regulation and control sensitivity sequence of all PV nodes to first PQ subregion, when ensureing connective, the sensitiveest PV node merger Ru Gai district is selected; Reactive source node selection is in like manner carried out in remaining region, the PV node that the forefoot area should got rid of when at every turn choosing PV node had been selected, ensures that each PQ subregion all has a reactive source node behind first subzone, ensures subregion controllability;

(2) by the PV node sequencing of all the other non-merger, by single PV node to the region voltage of all PQ subregions regulation and control sensitivity sequence, when ensureing connective by this PV node division to the highest PQ subregion of sensitivity; Complete all PV node division successively.

9. Power Network Partitioning method under wind power integration as claimed in claim 1, is characterized in that, based under Silhouttte index and wind power integration between PQ node electrical distance expect that in definition, the interval weak coupling index of close coupling is as follows:

${\mathrm{QNJ}}_i=\frac{1}{n_i}\underset{t=1}{\overset{n}{\Σ}}(\frac{b_i\left(t\right)-a_i\left(t\right)}{max\{a_i\left(t\right),b_i\left(t\right)\}}\·C_i)(i=1,2,...,L)---\left(9\right)$

$QNJ=\frac{1}{L}\underset{i=1}{\overset{L}{\Σ}}{\mathrm{QNJ}}_i---\left(10\right)$

In formula: QNJ _irepresent the coupling index of i-th PQ subregion; QNJ represents the whole network PQ subarea management index; a _it () represents all PQ node electrical distance averages in subregion i interior nodes t and district; b _it () represents that subregion i interior nodes t is to PQ node electrical distance averages all outside district; C _ibe used to indicate PQ subregion i whether to there is node and pass through situation, when exist one or more node not with it around any node divide to same district, then there is isolated node or there is reachability problem in subregion, this season make QNJ=-1; Otherwise C _i=1, n _ifor PQ nodes in subregion i; L is the PQ number of partitions; QNJ _iand the value of QNJ is all between [-1,1].

10. Power Network Partitioning method under wind power integration as claimed in claim 1, it is characterized in that, PV node voltage regulates and controls sensitive index:

PV node divides in best PQ region with regulating and controlling voltage sensitivity, ideally, each PV node with peak response control region interior nodes voltage, simultaneously perturbing area exterior node voltage hardly, meaning is set out thus, controls sensitive indices P VC as follows based on perturbation method definition PV node voltage:

${\mathrm{PVC}}_j=\frac{\underset{p\∈{\mathrm{\Ω}}_j}{\Σ}\left|{\mathrm{\ΔU}}_p\right|}{\underset{q\∈\mathrm{\Ω}}{\Σ}\left|{\mathrm{\ΔU}}_q\right|}\·C_{{\mathrm{\Ω}}_j}(j=1,2,...,M)---\left(11\right)$

$PVC=\frac{1}{M}\underset{j=1}{\overset{M}{\Σ}}{\mathrm{PVC}}_j---\left(12\right)$

In formula: PVC _jrepresent the sensitive index of regulating and controlling voltage of PV node j; PVC represents that the whole network PV node voltage regulates and controls sensitive index; Ω _jfor PQ node set in PV node j affiliated area, Ω is that the whole network PQ gathers, scalar region Ω _jinterior PV distributing equilibrium degree parameter, when this district PV nodes be greater than 0 and PV node and its PQ node that is directly connected divide to during same district then otherwise due to wind power integration, under different scene, PV node voltage controls sensitivity difference, | Δ U _p| with | Δ U _q| be respectively the voltage magnitude perturbation absolute value of PQ node p and q, adopt after wind-powered electricity generation injects and expect voltage deviation process fluctuation problem, namely Δ U _pi: wind-powered electricity generation scene i accesses the voltage magnitude Perturbation of lower PQ node p; Δ U _qiwind-powered electricity generation scene i accesses the voltage magnitude Perturbation of lower PQ node q, and Pi is the probability of wind-powered electricity generation scene i.