CN105896547A - Large power network graded voltage control method under wind power access - Google Patents

Abstract

The invention discloses a large power network graded voltage control method under wind power access. The control method comprises the steps of (1) measuring the power flow calculation data at the input current moment of the system in real time; (2) carrying out online selection on data leading node based on clustering by fast search and find of density peaks; (3) carrying out three-level voltage control optimization calculation according to the current state, and determining an adjusting target value of the voltage of the leading node according to the optimization result and the selected leading node; and (4) sending down the leading node target value to a two-level voltage control system, implementing two-level voltage control, ensuring the voltage of the leading node at the target value level, and maintaining the overall voltage level of the system; judging whether the system runs to a moment corresponding to a next three-level voltage control cycle or not; if so, returning to the step (1); and or otherwise, returning to the step (4). By adoption of the large power network graded voltage control method, it is ensured that the control machine set, the control target and the constraint conditions do not need to be replaced frequently in the process when the target issued by the three-level voltage control is maintained by the two-level voltage control, so that the feasibility of the practical engineering application can be ensured.

Description

A kind of bulk power grid hierarchical voltage control method under wind power integration

Technical field

The present invention relates to a kind of bulk power grid hierarchical voltage control method under wind power integration.

Background technology

It is the important ring in bulk power grid tertiary voltage control implementation process that Pilot bus selects, as secondary voltage control link Controlling target, its selection is particularly significant to the control effect of network-wide optimization.

Traditional Pilot bus system of selection is the most ripe, mainly includes idle the sensitiveest with electrical distance center or region voltage The node of degree maximum is as Pilot bus, based on sensitivity with the minimum target of load bus voltage deviation by intelligence such as genetic algorithms Pilot bus optimizing can be carried out by algorithm, on the basis of subregion, consider that node controllability factor selects based on the idle degree of coupling of node voltage Pilot bus.Above method once obtained good result in the actual application of power system, but the regenerative resources such as wind-powered electricity generation photovoltaic are big Scale accesses, and system load flow starts to present stronger random walk, and its fluctuation is the most violent and frequent, causes Pilot bus to be incited somebody to action The dynamic migration with power flow changing, brings challenges to the application of tradition system of selection.

To this end, existing document proposes multiple method, carry out optimizing and select Pilot bus, but existing method is the big rule of wind-powered electricity generation After mould accesses, the Pilot bus select permeability in the case of system running state random fluctuation provides good resolving ideas, but by In being limited by algorithm operation quantity, research at present is substantially all and rests on off-line optimization method to obtain optimum in statistical significance In the aspect selected, the running status control effect that the sacrificial section probability that is so bound to is relatively small.

Selecting to improve Pilot bus further, preferable solution is can be according to system running state, online tune in time Whole Pilot bus selects, it is ensured that the optimal control results under every kind of running status.Another kind of solution is to adapt to China's electrical network fortune " soft sectoring " thinking of row working conditions change, selects subregion, Pilot bus and controls unit optimum selection all in control centre by software in fact Existing, hardware fixed partition division relatively in the past and control model so that bulk power grid hierarchical voltage control has well adapted to China's electricity Network operation working conditions change engineering practice faster.Along with communication, the development of wide area measurement technology, system-wide operation and control In message-switching technique processed the most accessible, the most dynamically adjust the control measure that relate generally to of Pilot bus and the reality of control strategy Time coupling only need to improve in the software design of control centre, be feasible from application of engineering project, there is not enforcement The bottleneck of sport technique segment, it is critical only that the theoretical method of on-line selection Pilot bus.

If but still continued to use the exhaustive optimizing in extensive interblock space of conventional Combinatorial Optimization thinking, its amount of calculation would be to calculate in real time The Main Bottleneck selected.In order to on-line selection Pilot bus follows the tracks of system state change, fundamentally solve under the new situation Undulatory property problem, it is necessary to look for another way, uses new approaches to reduce the amount of calculation of optimal choice, for engineer applied based theoretical.

The node with close characteristic quantity can be turned to a class by cluster, and gained cluster centre is permissible as the center of this category node Representing such global feature amount level, this characteristic representing region voltage levvl with Pilot bus has concordance.Based on this, examine Consider and obtain, from cluster angle set off in search, the new method that online Pilot bus selects.Traditional clustering method such as K-means, cohesion Clusters etc. also use iterative search during finding cluster centre, are not suitable in line computation.

And, existing bulk power grid tertiary voltage control is general and takes no account of the condition of wind power integration, and the selection of Pilot bus is at wind Bulk power grid hierarchical voltage control under the access of TV university scale cannot have adaptability.

Summary of the invention

The present invention is to solve the problems referred to above, it is proposed that a kind of bulk power grid hierarchical voltage control method under wind power integration, this method When implementing hierarchical voltage control under the conditions of wind power integration, first can enter according to current state when the tertiary voltage control cycle initiates The Pilot bus selection that row is online, it is then determined that control unit, adjusts secondary voltage control strategy and constraints, then carries out complete Network optimization issues desired value to secondary voltage control device, completes the most complete hierarchical voltage control.Then weigh when next cycle opens Multiple said process completes regulation and control, it is ensured that control the running status change that the fluctuation of target energy real-time tracking wind-powered electricity generation causes.It is simultaneously introduced nothing Need the quick clustering algorithm of iterative process-based on fast search with find density peaks cluster (Clustering by fast search and Find of density peaks), clustered by the PQ node selecting suitable cluster feature vector to treat regulation and control so that poly- Class center is Pilot bus.With cluster centre for control target carry out Control of Voltage eliminate its fluctuation after, can make such own Node voltage fluctuation minimum.

To achieve these goals, the present invention adopts the following technical scheme that

A kind of bulk power grid hierarchical voltage control method under wind power integration, comprises the following steps:

(1) the Load flow calculation data of real-time measurement system input current time；

(2) based on fast search and density peaks clustering algorithm, to data Pilot bus on-line selection；

(3) tertiary voltage control optimization calculating is carried out according to current state, according to optimum results and the Pilot bus of selection, really Determine Pilot bus voltage-regulation desired value；

(4) Pilot bus desired value is issued to Secondary Voltage Control System, enforcement secondary voltage control, it is ensured that Pilot bus electricity It is pressed in desired value level, maintains system global voltage level；

(5) judge whether system was run to moment corresponding to next tertiary voltage control cycle, if it is return step (1) step (4), is otherwise returned.

In described step (2), Pilot bus based on fast search and density peaks cluster selects, and specifically includes following steps:

(2-1) utilize Sensitivity Method the most quickly to calculate each PQ node and eliminate the voltage of remaining PQ node after disturbance voltage deviation Deviation, and as the characteristic vector of corresponding PQ node, obtain cluster feature matrix；

(2-2) utilize clustering algorithm according to the local density of each node of cluster feature matrix calculus and its short distance with high density point From, determine candidate cluster center, and determine Pilot bus number according to the idle sensitivity correlation matrix of system voltage under current state；

(2-3) Pilot bus in the current tertiary voltage control cycle is determined by depression of order optimizing.

In described step (2-1), on the PQ node that sum is n, individually it is sequentially added into the reactive disturbance of fixed size, n time Eliminate under disturbance all PQ nodes after the voltage deviation of node i voltage deviation vector average as node i cluster feature to Amount.

In described step (2-1), the cluster feature vector of n node i of each PQ node is arranged in order, forms n Wei Te Levy matrix, this matrix is carried out symmetrization process, obtain cluster feature matrix.

In described step (2-1), using the reactive disturbance of load bus as system disturbance amount, build the whole network PQ node voltage and become Change amount, PQ node idle work variable quantity and the relation of PV node voltage variable quantity, introduce incidence matrix, eliminates destination node electricity Pressure deviation.

In described step (2-2), concrete steps include:

A () calculates local density and itself and the beeline of high density point of n node, choose local density and beeline is equal Cluster centre point higher than average；

B () obtains system Pilot bus number by the correlation matrix feature value analyzing grid nodes reactive voltage sensitivity matrix Method determines electrical network Pilot bus number.

In described step (b), the scale of the correlation matrix of grid nodes reactive voltage sensitivity matrix controls nodes with system Mesh is correlated with.

In described step (2-3), when utilizing Calculation of Sensitivity using y node as Pilot bus, eliminate its voltage under disturbance inclined The voltage deviation absolute value sum of after the recovery the whole network PQ node, the voltage deviation eliminating this y node under disturbance can make the whole network save Point voltage deviation is minimum.

In described step (2-3), after the whole network PQ node cluster based on voltage deviation completes, clustered by price reduction optimizing gained Center is directly as the Pilot bus of system under given state.

In described step (3), the cycle of Pilot bus on-line selection need to consider system running state random fluctuation with three grades Time scale with secondary voltage control.

The invention have the benefit that

(1) when the present invention implements hierarchical voltage control under the conditions of wind power integration, can when the tertiary voltage control cycle initiates, First carry out online Pilot bus according to current state to select, it is then determined that control unit, adjust secondary voltage control strategy peace treaty Bundle condition, then carry out network-wide optimization and issue desired value to secondary voltage control device, complete the most complete hierarchical voltage control.When Next cycle opens then repetition said process and completes regulation and control, it is ensured that control the running status that the fluctuation of target energy real-time tracking wind-powered electricity generation causes Change.

(2) present invention by carrying out the on-line selection of Pilot bus, energy based on fast search and discovery density peaks clustering algorithm Preferably reflect the voltage levvl of system, obtain the bulk power grid tapping voltage under more preferable regulating effect can make wind-powered electricity generation access on a large scale Control that there is higher adaptability, it is possible under reply large-scale wind power access conditions, voltage-controlling effect improves problem.

(3) by introduce without iterative process based on fast search with find density peaks clustering algorithm, use sensitivity meter The voltage deviation of remaining node after elimination node voltage deviation under calculation disturbance, and cluster in this, as cluster feature amount, make to gather Class center can represent system voltage levvl and directly as system Pilot bus.

(4) quick clustering algorithm operation time is extremely short, and the calculating being simultaneously based on sensitivity can avoid repeatedly Load flow calculation, from And make carried algorithm overcome the limitation of the extensive exhaustive combination optimizing of tradition, not only reduce amount of calculation, and select output control to imitate Fruit is more preferably；

(5) simulation result shows, based on quick clustering gained cluster centre as Pilot bus, not only calculates fast and reliable, Can represent system voltage levvl well, and the used time extremely short meets application on site requirement, voltage-controlling effect is better than conventional offline The control effect of the Pilot bus selected；

(6) owing to being the division type clustering algorithm without study based on fast search and discovery density peaks cluster, without iteration Can directly obtain Optimal cluster centers, therefore cluster speed is exceedingly fast so that Pilot bus selects to complete in very short time, full Foot application on site requirement.

Accompanying drawing explanation

Fig. 1 is the Pilot bus on-line selection flow chart schematic diagram of the coupling system control of the present invention；

Fig. 2 is the node voltage deviation average ratio relatively curve synoptic diagram of the present invention；

Fig. 3 is the whole network PQ node voltage deviation and curve synoptic diagram under each disturbance of the present invention；

Voltage-controlling effect comparison curves schematic diagram in the odd-numbered day that Fig. 4 is the present invention；

Fig. 5 is the node voltage skew average ratio relatively curve synoptic diagram of the present invention；

Fig. 6 be the present invention each disturbance under the whole network PQ node voltage deviation and curve synoptic diagram.

Detailed description of the invention:

The invention will be further described with embodiment below in conjunction with the accompanying drawings.

Cause the situation of state random fluctuation for wind power integration, the on-line tuning that Pilot bus selects must take into tertiary voltage control The engineering of molding formula is actual.If it is excessively frequent that Pilot bus adjusts, secondary voltage control will be made constantly to adjust control target, Many difficulties is brought at regulating strategy with on the regulation and control cycle.If Pilot bus select to update excessively slow, again cannot in time with The change of track system running state, the regulating effect that impact is overall.To this end, the cycle of Pilot bus on-line selection needs comprehensively to examine Consider system running state random fluctuation and three grades and the time scale of secondary voltage control.

For realize three grades with secondary voltage control decoupling in time, the whole network idle work optimization cycle be generally 15 minutes to hour level, The secondary voltage control cycle is a minute level.Being 15 minutes in view of the Real-Time Scheduling schedule regeneration cycle, the whole network idle work optimization cycle is past Toward being set to 15 minutes.Add up according to wind power fluctuation pattern, wind power fluctuation range in the short-term time scale of 15 minutes Less and distribution is more concentrated, it can therefore be concluded that cause system running state random fluctuation the least, i.e. at tertiary voltage control In cycle, system running state is the most steady.Therefore select can be as the cycle for online Pilot bus, so that secondary voltage Control to issue at maintenance tertiary voltage control object procedure to be not required to frequently change control unit, control target and constraints, really Protect the feasibility of practical engineering application.

Based on above-mentioned analysis, under the conditions of wind power integration implement hierarchical voltage control time, can when the tertiary voltage control cycle initiates, First carry out online Pilot bus according to current state to select, it is then determined that control unit, adjust secondary voltage control strategy peace treaty Bundle condition, then carry out network-wide optimization and issue desired value to secondary voltage control device, complete the most complete hierarchical voltage control.When Next cycle opens then repetition said process and completes regulation and control, it is ensured that control the running status that the fluctuation of target energy real-time tracking wind-powered electricity generation causes Change.Its detailed process is as shown in Figure 1.

For ensureing the correctness of selection and the rapidity of Pilot bus, present invention introduces the quick clustering algorithm without iterative process- Based on fast search and discovery density peaks cluster (Clustering by fast search and find of density peaks), logical Cross the PQ node selecting suitable cluster feature vector to treat regulation and control to cluster so that cluster centre is Pilot bus.With Cluster centre for control target carry out Control of Voltage eliminate its fluctuation after, can make such all node voltages fluctuation minimum.Due to Clustering with discovery density peaks based on fast search is the division type clustering algorithm without study, can directly obtain optimum without iteration Cluster centre, therefore cluster speed is exceedingly fast so that Pilot bus selects to complete in very short time, meets application on site requirement. Simulation result shows, based on quick clustering gained cluster centre as Pilot bus, not only calculates fast and reliable, moreover it is possible to preferably The voltage levvl of reflection system, obtains the bulk power grid hierarchical voltage control tool under more preferable regulating effect can make wind-powered electricity generation access on a large scale There is higher adaptability.

This clustering algorithm key problem is to determine cluster centre.It realizes based on two basic features below in relation to cluster centre: (1) local density of node is generally lower than the local density of such cluster centre about；(2) its with have more highdensity The distance of node is the biggest.Algorithm is primarily based on cluster feature amount (such as euclidean distance between node pair), and to calculate the local of each data point k close Degree ρ_kWith its distance minima δ with the node higher than its density_k, and according to local density's ρ value of whole nodes and each joint Put the distance minima δ-value with the node higher than its density and make ρ-δ decision diagram to determine cluster centre point.Algorithm regulation is same Time there is bigger ρ_kWith bigger δ_kData point k could be as cluster centre point.After cluster centre determines, take each data Belonging to the high density point that point is nearest, cluster centre completes classification as its cluster centre.

Pilot bus fast selecting method based on PQ node clustering

For break through tradition with the minimum target of voltage deviation by intelligent algorithm optimizing search Pilot bus pattern limitation, first from The method that Pilot bus is chosen in the proposition of cluster visual angle, the Pilot bus obtained with less amount of calculation based on the quick clustering algorithm introduced, Can preferably reflect system voltage level, to meet follow-up application on site demand.

The basic thought of Pilot bus is obtained based on cluster

Cluster itself is to excavate data point to unite the process of characteristic, and cluster feature amount has multiple choices, during based on different purposes cluster Selected characteristic quantity is different, and gained cluster centre also will have different characteristics.Such as the cluster for the purpose of subregion, permissible The electrical distance of grid nodes clusters as characteristic quantity, and gained cluster centre i can represent its place class Ω_iElectrical distance Center or node i are Ω_iIn and other nodes between electrical distance and minimum point.Pilot bus is the representative of region voltage level Property node, the voltage deviation that can make remaining node after occurring to eliminate its voltage deviation during disturbance is minimum.If therefore disappearing with under disturbance After the voltage deviation of arbitrary node j, the voltage deviation size of node k clusters as " distance " between j and k, gained Cluster centre node to other all internodal " distance " sums in class by inclined for the voltage eliminating cluster centre under minimum, i.e. disturbance After the recovery, the voltage deviation of its remaining node of apoplexy due to endogenous wind also will minimize.Now gained cluster centre will meet determining of Pilot bus selection Justice.As long as therefore the characteristic quantity needed for Rational choice cluster, can obtain main by choosing cluster centre based on suitable clustering algorithm Lead node.

To this end, determine that euclidean distance between node pair defines by Pilot bus definition, and then choose to eliminate the voltage deviation cluster as core special Levy vector, be re-introduced into can quick clustering based on fast search with find density peaks clustering algorithm, for online the most quickly select main Lead node based theoretical.

Based on Calculation of Sensitivity quick obtaining cluster feature vector

Based on fast search and the characteristic vector finding density peaks cluster there is following two feature, one be Arbitrary Digit strong point r all There is its characteristic vector γ_r=[γ_r1,…γ_rj,…γ_rn] (n is cluster numbers strong point sum), arbitrary element γ_rjRepresent the spy of some r and some i Levy the degree of association；Two be cluster operating basis be the symmetry square matrix being made up of each data point characteristic vector.In order to based on being proposed cluster calculation Method directly obtains Pilot bus, needs to obtain the eigenmatrix of node to be clustered under given system running state as operating basis.

For obtaining the node diagnostic vector being applicable to carry out Pilot bus selection based on fast search and discovery density peaks cluster, fixed After eliminating the voltage deviation of node i under justice disturbance, the vector formed using the voltage deviation of remaining load bus of the whole network is as node i Cluster feature vector.Owing to needing the reactive disturbance measurer arranged to have multiple choices, the present invention to define any PQ of formation in calculating The cluster feature vector detailed process of node i is as follows: be individually sequentially added into the idle of fixed size on the PQ node that sum is n Disturbance Δ Q, after eliminating the voltage deviation of node i under n disturbance, the average of the voltage deviation vector of whole PQ nodes is as node The cluster feature vector γ of i_i。γ_iConcrete acquisition process as follows:

Individually add the reactive disturbance of Δ Q at g node, after eliminating the voltage deviation of node i, the whole network node voltage deviation constitute to Amount Δ VⁱG () is as follows:

${\mathrm{\ΔV}}^i\left(g\right)=\left[\begin{array}{c}{\mathrm{\ΔV}}_1^i\left(g\right)\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{i-1}^i\left(g\right)\\ 0\\ {\mathrm{\ΔV}}_{i+1}^i\left(g\right)\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_n^i\left(g\right)\end{array}\right]---\left(1\right)$

In formula: Δ VⁱWhen () is for being only fixed reactive disturbance Δ Q on PQ node g g, whole after eliminating the voltage deviation of node i The vector that the voltage deviation amount of load bus is constituted.ExtremelyAfter being respectively the voltage deviation of elimination node i, node The respective voltage deviation of 1 to node n.

Thus, when can obtain individually adding reactive disturbance on 1 to n node successively, the whole network after the voltage deviation of node i is eliminated Node bias in the power grid vector Δ Vⁱ(1) to Δ Vⁱ(n).The new vector that the average of n vector respective element is constituted is i.e. the spy of node i Levy vector γ_i。

${\mathrm{\γ}}_i=\left[\begin{array}{c}\frac{1}{n}\underset{g=1}{\overset{n}{\Σ}}{\mathrm{\ΔV}}_1^i\left(g\right)\\ .\\ .\\ .\\ \frac{1}{n}\underset{g=1}{\overset{n}{\Σ}}{\mathrm{\ΔV}}_{i-1}^i\left(g\right)\\ 0\\ \frac{1}{n}\underset{g=1}{\overset{n}{\Σ}}{\mathrm{\ΔV}}_{i+1}^i\left(g\right)\\ .\\ .\\ .\\ \frac{1}{n}\underset{g=1}{\overset{n}{\Σ}}{\mathrm{\ΔV}}_n^i\left(g\right)\end{array}\right]---\left(2\right)$

Obtain the cluster feature vector of each PQ node the most successively, and form following n dimensional feature matrix D:

D=[γ₁ … γ_i … γ_n] (3)

In formula: γ_i(i=1 ... n) be the characteristic vector of node i.

Now by D matrix by the i.e. available operating basis based on fast search with discovery density peaks cluster of formula (4) symmetrization Matrix S.

$S_{ij}=S_{ji}=\frac{1}{2}(D_{ij}+D_{ji}),(i=1...n,j=1...n)---\left(4\right)$

Voltage deviation amount Δ V in formula (1)ⁱG () is the key constituting cluster feature vector.Under traditional approach, can be by under disturbance Before and after eliminating destination node voltage deviation, twice system load flow calculates acquisition, but Load flow calculation amount greatly cannot adapt to quickly apply need Asking, for avoiding repeatedly Load flow calculation, the present invention, as a example by the regulation measure of electromotor, introduces and obtains voltage deviation based on sensitivity Method.Concrete calculating process is as follows:

Using the reactive disturbance of load bus as system disturbance amount.The whole network PQ node voltage variable quantity, PQ node idle work variable quantity And there is following relation in PV node voltage variable quantity:

$\left[\begin{array}{c}{\mathrm{\ΔV}}_{PQ1}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PQi}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PQn}\end{array}\right]=B*\left[\begin{array}{c}{\mathrm{\ΔQ}}_{PQ1}\\ .\\ .\\ .\\ {\mathrm{\ΔQ}}_{PQi}\\ .\\ .\\ .\\ {\mathrm{\ΔQ}}_{PQn}\end{array}\right]+M*\left[\begin{array}{c}{\mathrm{\ΔV}}_{PV1}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PVj}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PVm}\end{array}\right]---\left(5\right)$

In formula: Δ V_PQWith Δ Q_PQIt is the column vector (n is network PQ nodes) of n*1, its element corresponding each PQ joint respectively The voltage of point and idle work variable quantity.ΔV_PVFor the column vector (m is PV nodes in network) of m*1, its element is followed successively by respectively PV node voltage variable quantity.B is that the sensitivity matrix of n*n characterizes the spirit that PQ node voltage is changed by the idle change of PQ node Sensitivity.M is that the sensitivity matrix of n*m characterizes the PV node voltage change sensitivity to the change of PQ node voltage.

Formula (5) illustrates that PQ node voltage is affected by two kinds of factors, i.e. the voltage of the idle and PV node of PQ node.Two kinds The comprehensive response of factor is i.e. the voltage variety of PQ node.If therefore sensitivity matrix B Yu M is all it is known that given PQ saves Point reactive disturbance amount and PV node voltage regulated quantity then can directly calculate the voltage variety of each load bus and need not calculate tide Stream.Reactive voltage sensitivity matrix B can directly be obtained by Jacobian matrix.Sensitive between PV node voltage and PQ node voltage Degree relation is shown below, and when therefore giving system running state, B, Metzler matrix all can be tried to achieve.

$\left[\begin{array}{c}\mathrm{\Δ}\mathrm{\θ}\\ \frac{{\mathrm{\ΔV}}_{PQ}}{V_{PQ}}\end{array}\right]={\left[\begin{array}{cc}\frac{\∂\mathrm{\Δ}P}{\∂\mathrm{\θ}}& \frac{\∂\mathrm{\Δ}P}{\∂V_{PQ}}{V}_{PQ}\\ \frac{\∂\mathrm{\Δ}Q}{\∂\mathrm{\θ}}& \frac{\∂\mathrm{\Δ}Q}{\∂V_{PQ}}{V}_{PQ}\end{array}\right]}^{-1}\left[\begin{array}{c}\frac{\∂\mathrm{\Δ}P}{\∂V_{PV}}{V}_{PV}\\ \frac{\∂\mathrm{\Δ}Q}{\∂V_{PV}}{V}_{PV}\end{array}\right]\frac{{\mathrm{\ΔV}}_{PV}}{V_{PV}}---\left(6\right)$

In formula: subscript PQ and PV represent PQ node and PV node respectively.

For eliminating destination node voltage deviation, introduce incidence matrix C so that

$C*\left[\begin{array}{c}{\mathrm{\ΔV}}_{PQ1}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PQi}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PQn}\end{array}\right]={\mathrm{\ΔV}}_{PQt}---\left(7\right)$

In formula: Δ V_{P Qt}It it is the voltage variety of target load node t.C is that n ties up row vector, wherein

$C_{1j}=\{\begin{array}{c}1,(j=t)\\ 0,(j\≠t)\end{array}---\left(8\right)$

By formula (5) both sides with premultiplication C so that

${\mathrm{\ΔV}}_{PQt}=C*B*\left[\begin{array}{c}{\mathrm{\ΔQ}}_{PQ1}\\ .\\ .\\ .\\ {\mathrm{\ΔQ}}_{PQi}\\ .\\ .\\ .\\ {\mathrm{\ΔQ}}_{PQn}\end{array}\right]+C*M*\left[\begin{array}{c}{\mathrm{\ΔV}}_{PV1}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PVj}\\ .\\ .\\ .\\ {\mathrm{\ΔV}}_{PVm}\end{array}\right]---\left(9\right)$

Making formula (9) is 0, at network disturbance Δ Q_PQTime known, each needed for counter can solving the voltage deviation eliminating destination node t The voltage variety Δ V of PV node_PV.By Δ V_PQWith Δ V_PVBring formula (5) into and can solve the voltage eliminating destination node t under disturbance The vector that after deviation, the voltage deviation amount of remaining load bus is formed.

Thus, whole bearing can quickly be calculated on the premise of avoiding repeatedly Load flow calculation based on Calculation of Sensitivity under given system mode The cluster feature matrix of lotus node, sets up operating basis for following cluster computing.

Pilot bus based on cluster with depression of order optimizing selects

After obtaining cluster feature matrix S, can carry out clustering the cluster centre under acquisition system current state.

The local density ρ of each PQ node is calculated successively by eigenmatrix S_i(i=1 ... n) with itself and the beeline of high density point δ_i(i=1 ... n).Based on fast search with find density peaks clustering algorithm point out cluster centre should have simultaneously bigger ρ value and δ-value, therefore ρ value and δ-value according to whole nodes carry out candidate cluster central point decision-making, and detailed process is:

First calculate the σ value of n node with formula (10), and select with this in cluster with bigger ρ value and δ-value simultaneously Heart point.

σ_i=ρ_iδ_i(i=1 ..., n) (10)

Generally, the σ value of most data points is the most close and value is the least, σ value very greatly and different with many numerical difference bigger just It it is cluster centre point.Therefore algorithm is by after the σ value descending of each node, directly selects the number of significantly greater than most node σ value Strong point is as cluster centre.Therefore cluster centre and nothing can be immediately arrived at based on fast search with discovery density peaks clustering algorithm itself Cluster numbers need to be preset.If the inconspicuous σ value causing cluster centre point of characteristic is not apparent from exceeding most node but data point itself is united During σ value, the precision of algorithms selection cluster centre point will reduce.In order to reduce σ value difference different unconspicuous situation, Pilot bus is selected Selecting the error caused, the present invention combines Pilot bus feature and is improved the process of algorithm decision-making cluster centre.First will tool There is the node of bigger σ value all as candidate cluster center, then determine system Pilot bus number by traditional method, utilize depression of order Optimizing is found out in the heart in candidate cluster and is adapted to the Pilot bus that the precision of current system conditions is higher.Detailed process is as follows:

1) determination at candidate cluster center

After being calculated each node σ value by formula (10), formula (11) calculate the σ average of whole node

$\stackrel{\‾}{\mathrm{\σ}}=\frac{1}{n}\underset{i=1}{\overset{n}{\Σ}}{\mathrm{\σ}}_i---\left(11\right)$

By the σ value descending of node each in domain, take and be wherein not less thanAll x elements, x node of its correspondence Or there is bigger ρ value or there is bigger δ-value or its ρ value is the biggest with δ-value.Therefore this x node will be constituted in candidate cluster Heart point.

2) Pilot bus number is determined

The method obtaining system Pilot bus number by the correlation matrix feature value analyzing grid nodes reactive voltage sensitivity matrix Can determine that electrical network Pilot bus number y.When in candidate cluster, calculation x is not more than y, directly using all candidate cluster centers as master Lead node, otherwise after obtaining x candidate cluster center, Pilot bus depression of order optimizing screening will be carried out with Pilot bus feature.

It is relevant that the scale of the correlation matrix of grid nodes reactive voltage sensitivity matrix controls interstitial content to system.Due to generating Machine node is the minority node in system, therefore the amount of calculation calculating Pilot bus number is the least, is adapted to application on site requirement.

3) Pilot bus is selected based on the depression of order optimizing of candidate cluster center

The use of clustering algorithm makes calculation x in candidate cluster the least, i.e. x=y+g, wherein g < < n.Therefore n data point is sought Look for y put optimum organization number byDrop toEven if needing to use local optimal searching, computing still can complete to meet soon in short-term Speed computing demand.When utilizing sensitivity can calculate y node as Pilot bus, complete after eliminating its voltage deviation under disturbance The voltage deviation absolute value sum of net PQ node.In x candidate cluster, optimizing in the heart can select y node based on this, disturb The dynamic lower voltage deviation eliminating this y node can make the whole network node voltage deviation minimum.

After thus the whole network PQ node cluster based on voltage deviation completes, can be directly as by price reduction optimizing gained cluster centre The Pilot bus of system under given state.

Emulation experiment:

The measured data of the Shandong Power wind field used, and it is chosen at 2,7,16 three nodes access wind fields equally.System loading And reactive disturbance situation is using New England39 node system initial load situation as paddy load condition, by whole loads by 1.3 With 1.7 proportionality coefficient amplify respectively obtain waist load condition and peak load condition；System disturbance is set to each load bus and adds The reactive disturbance of any multiple in himself load or burden without work size 0 to 5%.Wind energy turbine set rated capacity is the 12% of peak load total amount. (1) optimizing obtains the traditional method representative of Pilot bus；(2) wind electricity volatility is processed with mathematic expectaion.Both are off-line side Formula, inventive algorithm will be controlled effectiveness comparison with both the above typical method.Owing in system, electromotor nodes is 10 Individual, the operation time that actual Pilot bus number determines is extremely short, does not affect the integral operation amount that online Pilot bus selects.Emulation In Pilot bus number under various state be set be 4.

For visual representation Pilot bus voltage-controlling effect, each PQ node after eliminating the voltage deviation of Pilot bus under calculation perturbation Voltage deviation accounts for before its disturbance that the ratio of voltage is as node voltage deviation percent, later referred to as voltage deviation, and unit is %. The most all load bus global voltage deviation is the least, shows that corresponding Pilot bus voltage-controlling effect is the best.

Pilot bus effectiveness comparison under system typical operation

Arranging system mode for arranging load condition is paddy load, and system mode is the initial shape of New England39 node system State.

First carry out the whole network Load flow calculation, obtain B, M sensitivity matrix in formula (5).Then load bus be sequentially added in The system disturbance of itself load or burden without work 5 percent times, utilizes Sensitivity Method to be eliminated the voltage of each load bus the most successively The voltage deviation amount of remaining load bus during disturbance, and then obtain the cluster feature vector of each PQ node.To obtain No. 1 node Proper phasor as a example by carry out as described below.First the system disturbance of himself load or burden without work 5 percent it is incorporated at No. 1 node, After eliminating No. 1 node voltage deviation, the whole network load bus voltage deviation constitutes phasor Δ V¹(1).The most individually exist After No. 2 nodes add disturbance, eliminate No. 1 node voltage deviation and obtain now the whole network node bias in the power grid vector Δ V¹(2).Depend on subsequently Secondary after 3 to No. 29 nodes add disturbance, eliminate No. 1 node voltage deviation and can obtain voltage deviation vector successively.Final 1 The cluster feature vector γ of number node₁As follows:

${\mathrm{\&gamma;}}_1=\frac{1}{29}\underset{g=1}{\overset{29}{\&Sigma;}}{\mathrm{\&Delta;V}}^1\left(g\right)$

In like manner can obtain the characteristic vector of 2 to No. 29 nodes.Permissible by each node clustering characteristic vector convolution (3) formula (4) Build and cluster with the symmetrical feature matrix S finding density peaks cluster based on fast search.Obtain cluster centre point that is master Leading node is 2,5,16,23.Under typicalness, method (1) Pilot bus selects result to be 18,20,22,26；Method (2) selects Pilot bus is 9,11,17,28.

For observing Pilot bus voltage-controlling effect, now it is incorporated in himself load or burden without work size 0 to 5% to each load bus The arbitrarily reactive disturbance of multiple.Above the system disturbance of 5 percent is different from random disturbance herein, and the former is for obtaining feature Phasor and the disturbance that is manually set, the latter is the random disturbance set up for checking Pilot bus effect.For the most selected main Lead node control effect, randomly choose 20 groups of different system disturbances, observe and eliminate Pilot bus voltage under 20 groups of disturbances The whole network load bus voltage deviation situation after deviation.In subsequent simulation, system disturbance set-up mode is constant.

Under system typical operation, under 20 groups of disturbances, eliminate 20 times after Pilot bus voltage deviation with same load bus The average of different voltage deviations is observed as its voltage deviation, and 29 load bus voltage deviations of line are all worth three kinds successively Variation curve under different Pilot bus selection modes, as shown in Figure 2.In figure, " variation average 1 " curve is by this The variation situation of the Pilot bus of bright method choice, " variation average 2 " curve shows by the leading joint of method (1) gained The Control of Voltage situation of point, " variation average 3 " curve is the voltage-controlling effect of the Pilot bus that method (2) selects.

Observing the voltage deviation situation of 29 PQ nodes, on curve 1, the voltage deviation of 20 nodes is less than the joint on curve 2 Point deviation, on curve 1,15 node voltage deviations are less than the node deviation on curve 3.Article three, the voltage deviation of curve and point It is not 0.7427%, 1.6479%, 0.9862%.I.e. curve 1 shows entirety and more preferably controls effect.

, eliminate after Pilot bus voltage deviation under each disturbance, the whole network load bus voltage deviation and also can reflect leading saving meanwhile Point voltage controls effect.Observe in 20 groups of disturbances successively, often eliminate the whole network PQ node after Pilot bus voltage deviation under group disturbance Voltage deviation and, result is as shown in Figure 3.

Curve shows under the disturbance of many groups, and Pilot bus as selected by the inventive method can make the whole network PQ node voltage deviation and the lowest In two kinds of traditional methods, after i.e. eliminating the voltage deviation of Pilot bus selected by the inventive method, voltage deviation average is respectively less than remaining Two kinds of methods.Thus demonstrate that from another side typical operation modes, Pilot bus selected by the inventive method has overall more preferable Voltage-controlling effect.

On-line selection Pilot bus effectiveness comparison under wind power integration

Access in the case of wind-powered electricity generation, Pilot bus can with system mode dynamic migration.Emulation use Shandong wind field with 5 points Clock is spaced the force data of gaining merit in real time surveyed.The quick Pilot bus that on-line tracing trend migrates selects to be equivalent in handoff scenario Pilot bus under rear quick obtaining current state, then transfer to obtain the Pilot bus under state after T time is spaced.As long as thus The trend section providing recursion in time can obtain a series of Pilot bus result.

Emulation presents Pilot bus on-line selection and corresponding on-Line Voltage Control effect in the odd-numbered day, takes regulation and control cycle T=15min.Respectively Wind field exerts oneself and is divided into 0, less than 50%, more than 50% 3 kind of interval, taking interval power average is that this interval represents power, therefore A kind of lower three wind fields of load condition have been exerted oneself 27 kinds of combine scenes.For Pilot bus situation, example under the various state of complete observation The wind power output scene of one day preset by middle need, arranges and changes one every a regulation and control cycle and exert oneself scene, by presetting the load period Load condition is set.The fluctuation scene of exerting oneself of lower three wind fields of various load conditions all has 27 kinds of combinations, under the most each control cycle Select different wind power output combine scenes so that in the set odd-numbered day, to contain three kinds of load conditions respective as far as possible for whole scenes 27 kinds of combinations of exerting oneself, the wind-powered electricity generation state generated in example in the odd-numbered day with this.In odd-numbered day, the wind-powered electricity generation scene under each regulation and control cycle uses people The actual wind power output state being provided without for set scene in the odd-numbered day, this is owing to wind electricity volatility is relatively strong, the reality in the odd-numbered day Wind power output can not contain state of all exerting oneself.By being manually set scene, the most possible combination of exerting oneself all is covered by list In scene in a few days, can complete observation Pilot bus control effect under various wind-powered electricity generations inject.For ease of with traditional method result Objective comparison, under peak waist paddy state, wind-powered electricity generation injecting power at times is carried out by method (2) statistical result.

After eliminating Pilot bus voltage deviation under disturbance, the whole network PQ node voltage absolute value of the bias sum is imitated as Control of Voltage Really quantizating index.Increase by arbitrary proportion in 0 to 5% owing to disturbance is set to the whole network load or burden without work, therefore ask for random 20 successively Eliminate under group disturbance after Pilot bus voltage deviation the whole network node voltage deviation and, and using the average of 20 groups of voltage deviation sums as this Pilot bus voltage-controlling effect measurement index.Simulation result shows, the Pilot bus result of method (1) is 18,20,22,26. Method (2) selects Pilot bus to be 9,11,17,28.The accumulative control Contrast on effect of three kinds of methods is as shown in Figure 4.

In Fig. 4, voltage deviation and 1,2,3 represents the Pilot bus that the present invention, method (1), method (2) select respectively Control effect.Simulation result shows, within the majority regulation and control cycle, selected by the inventive method, Pilot bus control effect is superior to tradition Method, only under part regulation and control cycle correspondence scene, the control effect of three kinds of methods is close.These control effect phase as shown in Figure 4 Near scene is in the peak load period, and is an at least scene injected more than 50% nominal output in three wind fields.Specifically Corresponding diagram 4 regulates and controls the default scene in cycle accordingly in 09:30,15:15,15:30,16:15,18:15,18:45 moment.Cause Under this long-play, the inventive method demonstrates preferable voltage-controlling effect.

Above-mentioned is that all status numbers of limit from the statistical significance observe control effect, now randomly draws a kind of actual motion state and carries out Careful observation.In extracted running status, three blower fans are exerted oneself and are distinguished the 2.2% of rated power, 27.3%, 70.3%, and run Under peak load condition.It is 2,5,16,23 that emulation obtains Pilot bus.Pilot bus voltage deviation is eliminated under 20 groups of disturbances Rear each node voltage deviation average is as shown in Figure 5.In Fig. 5, curve 1,2,3 is respectively the present invention, method (1), method (2) Control of Voltage situation, curve 1 has 16 node voltages less than curve 2 and curve 3.Article three, the voltage deviation of curve and difference It is 1.1098%, 1.5444%, 1.3870%.Therefore the inventive method shows more preferable global voltage control effect.

Observe under each group of disturbance whole PQ node voltage deviations and curve as shown in Figure 6 simultaneously.The inventive method as shown in Figure 6 Lower voltage deviation and respectively less than two kinds traditional methods, therefore emulation display the inventive method relatively traditional method has overall preferably electricity Pressure control ability.

Under the acquisition system currently regulation and control cycle after system mode, from carrying out cluster feature matrix data preparation to using cluster to obtain time Selecting cluster centre point, then depression of order optimizing obtains the Pilot bus of adaptive system current state, required simulation time amounts to 0.3851 second. The calculating time that whole Pilot bus is chosen, therefore online actual application had feasibility below 1 second.

Although the detailed description of the invention of the present invention is described by the above-mentioned accompanying drawing that combines, but not limit to scope System, one of ordinary skill in the art should be understood that on the basis of technical scheme, and those skilled in the art need not pay Go out various amendments or deformation that creative work can make still within protection scope of the present invention.

Claims (10)

1. the bulk power grid hierarchical voltage control method under wind power integration, is characterized in that: comprise the following steps:

(1) the Load flow calculation data of real-time measurement system input current time；

(2) based on fast search and density peaks clustering algorithm, to data Pilot bus on-line selection；

(3) tertiary voltage control optimization calculating is carried out according to current state, according to optimum results and the Pilot bus of selection, really Determine Pilot bus voltage-regulation desired value；

(4) Pilot bus desired value is issued to Secondary Voltage Control System, enforcement secondary voltage control, it is ensured that Pilot bus electricity It is pressed in desired value level, maintains system global voltage level；

(5) judge whether system was run to moment corresponding to next tertiary voltage control cycle, if it is return step (1) step (4), is otherwise returned.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 1, is characterized in that: described step Suddenly in (2), Pilot bus based on fast search and density peaks cluster selects, and specifically includes following steps:

(2-1) utilize Sensitivity Method the most quickly to calculate each PQ node and eliminate the voltage of remaining PQ node after disturbance voltage deviation Deviation, and as the characteristic vector of corresponding PQ node, obtain cluster feature matrix；

(2-2) utilize clustering algorithm according to the local density of each node of cluster feature matrix calculus and its short distance with high density point From, determine candidate cluster center, and determine Pilot bus number according to the idle sensitivity correlation matrix of system voltage under current state；

(2-3) Pilot bus in the current tertiary voltage control cycle is determined by depression of order optimizing.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly in (2-1), on the PQ node that sum is n, individually it is sequentially added into the reactive disturbance of fixed size, eliminates under n disturbance After the voltage deviation of node i, the average of the voltage deviation vector of whole PQ nodes is as the cluster feature vector of node i.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly in (2-1), the cluster feature vector of n node i of each PQ node is arranged in order, forms n dimensional feature matrix, will This matrix carries out symmetrization process, obtains cluster feature matrix.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly in (2-1), using the reactive disturbance of load bus as system disturbance amount, the whole network PQ node voltage variable quantity, PQ are built Node idle work variable quantity and the relation of PV node voltage variable quantity, introduce incidence matrix, eliminates destination node voltage deviation.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly, in (2-2), concrete steps include:

A () calculates local density and itself and the beeline of high density point of n node, choose local density and beeline is equal Cluster centre point higher than average；

B () obtains system Pilot bus number by the correlation matrix feature value analyzing grid nodes reactive voltage sensitivity matrix Method determines electrical network Pilot bus number.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 6, is characterized in that: described step Suddenly, in (b), it is relevant that the scale of the correlation matrix of grid nodes reactive voltage sensitivity matrix controls interstitial content to system.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly, in (2-3), when utilizing Calculation of Sensitivity using y node as Pilot bus, the whole network after its voltage deviation under disturbance, is eliminated The voltage deviation absolute value sum of PQ node, the voltage deviation eliminating this y node under disturbance can make the whole network node voltage deviation Minimum.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described step Suddenly, in (2-3), after the whole network PQ node cluster based on voltage deviation completes, directly made by price reduction optimizing gained cluster centre For the Pilot bus of system under given state.

Bulk power grid hierarchical voltage control method under a kind of wind power integration the most as claimed in claim 2, is characterized in that: described In step (3), the cycle of Pilot bus on-line selection needs to consider system running state random fluctuation and three grades and two grades of electricity The time scale of voltage-controlled system.