CN105528645A - Frangibility prediction method for large power grid - Google Patents

Abstract

The invention relates to a frangibility prediction method for the large power grid. The method is characterized by comprising the following steps that 1) power-grid rack program data and a system operation constrained condition are obtained; 2) a complex network mathematical model of the power grid is established according to the obtained constrained condition; 3) aimed at the complex network mathematical model, the purely graph-theory betweenness, the line capacity based trend distribution factor electric betweenness and the transmission capacity based trend distribution factor electric betweenness of each node/side are calculated; 4) a static fragile link of the grid is determined according to the three types of betweennesses obtained in the step 3); 5) a dynamic fragile link of the grid is determined by taking the three types of betweennesses in the step 3) as the basis; and 6) the static fragile link is integrated with the dynamic fragile link to obtain a fragile link of the whole large power grid. According to the invention, the large power grid is analyzed conveniently and effectively by utilizing complex network theories, so that a design staff can carry repeated examination and timely adjustment in the design process of the rack structure, and the process is simple.

Description

A kind of Forecasting Methodology of bulk power grid fragility

Technical field

The present invention relates to power grid security analysis technical field, particularly relate to a kind of Forecasting Methodology of bulk power grid fragility.

Background technology

The reliable and stable operation of electric system and national product, life and even national security closely bound up.Along with carrying out energetically of China's power grid construction, the development of generation of electricity by new energy, and the Practical Project such as D.C. high voltage transmission is increasing, carries out effective, scientific and reasonable prediction seem particularly important to these programmes.In the preconsolidation stress process of power grid construction, be an indispensable link to the fail-safe analysis of target grid.At present, the experts and scholars of association area put among the research of bulk power grid reliability and fragility in electric system, achieve significant achievement, have multiple method and study the reliability of bulk power grid and fragility.Patent documentation " power system vulnerability appraisal procedure " (application publication number is CN104156769A) provides so a kind of Forecasting Methodology: the complex network structures setting up electric system, fragile structure level of factor is set up by the electric betweenness in complex network structures, by flow data during static energy flow function method acquisition network implementation, the fragile factor of computing mode the fragile factor of integrated structure calculates comprehensive fragility prediction index, the weak section of last output power system.Above-mentioned Forecasting Methodology effectively can realize the prediction to power system vulnerability, but fragile structure Summing Factor state is fragile needs correspondence respectively to calculate because of the period of the day from 11 p.m. to 1 a.m obtaining for it, and process is more loaded down with trivial details.

Summary of the invention

The object of this invention is to provide the Forecasting Methodology of bulk power grid fragility, in order to solve the problem to bulk power grid fragility Forecasting Methodology very complicated in prior art.

For achieving the above object, the solution of the present invention comprises:

A Forecasting Methodology for bulk power grid fragility, is characterized in that, in turn includes the following steps:

Step 1: the constraint condition obtaining Net Frame of Electric Network layout data and system cloud gray model, comprising: space truss project data, system cloud gray model constraint condition;

Step 2: according to the complex network mathematical model of the constraint condition structure electrical network obtained;

Step 3: for above-mentioned complex network mathematical model, uses the betweenness algorithm based on shortest path to calculate pure graph theory betweenness, the electric betweenness of trend distribution factor based on circuit capacity, the electric betweenness of trend distribution factor based on transmission capacity on each node and limit;

Step 4: according to the three kinds of betweenness obtained in step 3, picks out several all forward nodes of three kinds of index rankings and limit, using the static state fragile link of bus, circuit and the transformer corresponding to these nodes and limit as electrical network;

Step 5: based on kind of the betweenness index of three in step 3, use attack simulating strategy, remove the bus in network or element, and calculate the state index removing at every turn and terminate rear electrical network, until state index is lower than certain number percent, these buses removed or element form the dynamic fragile link of network;

Step 6: the fragile link of static state draw respectively step 4 and step 5 and dynamic fragile link are integrated: common factor is got in fragile for static state link set and dynamic fragile link set, and the fragile link obtaining whole bulk power grid obtains the fragile link of whole bulk power grid.

Further, the space truss project data in described step 1 are: node data, transmission line data, transformer data, load data; System constraints is: each genset is exerted oneself the peak power output of higher limit and circuit.

Further, in described step 2, the structure of electrical network complex network mathematical model comprises the following steps:

S01: using all buses as node processing, circuit and transformer process as limit, the topological model of structure electrical network;

S02: build the constraint condition that power grid topology model runs;

S03: calculate the shortest path between often pair of generator-load bus in power grid topology model;

S04: according to the reactance of circuit in power grid topology model, forms the impedance matrix of system, and calculates the trend transmission distribution factor between often pair of node thus;

S05: the condition calculated according to above-mentioned steps, builds electrical network complex network mathematical model.

Further, the topological model in described step S01 comprises following essential information: generating node, transmission node, load bus, and the quiet meritorious of node is exerted oneself, the branch road reactance value be made up of circuit and transformer.

Further, in described step S03 shortest path calculating adopt be Dijkstra's algorithm.

Further, the certain value that in described step 4, ranking is forward is 20, after filtering out the node and limit that three kinds of index rankings are all positioned at front 20, by these three kinds of indexs according to a certain betweenness descending sort, sets up the static fragile link list of electrical network.

Further, in described step 5, the Forecasting Methodology of bulk power grid fragility is: respectively with random attack strategies, and progressively remove bus in network or element by the calculated attack strategy based on each betweenness index, applied load fill rat and the largest connected subset size of network are as prediction index, until load supply rate and largest connected subset size are lower than the number percent of specifying, the bus be now removed or element constitute the dynamic weak link of system.

Further, attack simulating strategy in described step 5 comprises: node is attacked at random, node betweenness is attacked, the circuit capacity PTDF betweenness of node is attacked, the transmission capacity PTDF betweenness of node is attacked, circuit is attacked at random, and the circuit capacity PTDF betweenness of circuit is attacked, and circuit transmission capacity PTDF betweenness is attacked.

Further, in the engineering of described attack simulating strategy, node is under attack, all circuits removing this node and be connected with this node from electrical network; Branch road is under attack, then remove all both-end elements of this branch road.

Further, the state index in described step 5 is load supply rate and network maximum UNICOM subset size.

Beneficial effect of the present invention is: use Complex Networks Theory to carry out analysis to bulk power grid convenient and effective, in the computation process of betweenness, without the need to iterating, also without the need to multiple sampling, be convenient to designer's repeated examinations in the design process of grid structure, and adjust in time, process is simple.

Simultaneously, by the heterogeneity of said method by prediction nodes characteristic, thus find out the key node in network exactly, remind designer to protect these joint strengthenings, or by rewiring, make the importance degree of node more be tending towards equalization, be tending towards reasonable.

Accompanying drawing explanation

Fig. 1 is the betweenness calculation flow chart of bulk power grid fragility Forecasting Methodology;

Fig. 2 is the network attack calculation flow chart of bulk power grid fragility Forecasting Methodology;

Fig. 3 is the electric hookup of RTS-48 system in embodiment.

Embodiment

Below in conjunction with accompanying drawing, the present invention will be further described in detail.

The invention provides a kind of bulk power grid fragility Forecasting Methodology based on Complex Networks Theory, comprise the following steps:

Step 1: the constraint condition obtaining Net Frame of Electric Network layout data and system cloud gray model, comprising: space truss project data, system cloud gray model constraint condition;

Step 2: according to the complex network mathematical model of the constraint condition structure electrical network obtained;

Step 3: for above-mentioned complex network mathematical model, uses the betweenness algorithm based on shortest path to calculate pure graph theory betweenness, the electric betweenness of trend distribution factor based on circuit capacity, the electric betweenness of trend distribution factor based on transmission capacity on each node and limit;

Step 4: according to the three kinds of betweenness obtained in step 3, picks out several all forward nodes of three kinds of index rankings and limit, using the static state fragile link of bus, circuit and the transformer corresponding to these nodes and limit as electrical network;

Step 5: based on kind of the betweenness index of three in step 3, use attack simulating strategy, remove the bus in network or element, and calculate the state index removing at every turn and terminate rear electrical network, until state index is lower than certain number percent, these buses removed or element form the dynamic fragile link of network;

Step 6: the fragile link of static state draw respectively step 4 and step 5 and dynamic fragile link are integrated: common factor is got in fragile for static state link set and dynamic fragile link set, and the fragile link obtaining whole bulk power grid obtains the fragile link of whole bulk power grid.

Below in conjunction with the analytic process of IEEE-48 example system, when applying the present invention and carrying model, need first to obtain related data from Electric Power Network Planning department.Wherein, the electric hookup of RTS-48 system as shown in Figure 3.

Corresponding to step 1, the constraint condition of Net Frame of Electric Network layout data and system cloud gray model is obtained from Electric Power Network Planning department.Wherein, the computation model input data obtained from Electric Power Network Planning department comprise following data:

The annexation of grid plan modification structure median generatrix and circuit;

The generator of every bar bus, power load distributing situation;

The impedance parameter of circuit, and maximum transmission power value.

Corresponding to step 2, according to the complex network mathematical model of the constraint condition structure electrical network obtained.

As shown in Figure 1, first by the above-mentioned computation model input data input network parameter collected, and then pure graph theory structure (topological structure) and the admittance matrix of network is set up according to the network parameter read.

Concrete, pure graph theory structure comprises the nodal information formed according to grid structure, and the link information (limit) between node.According to the raw data of electric hookup and network, write data file.

Using the bus of each in electrical network all as node processing.Symbol according to the clean meritorious injecting power of each node can determine node type.If injecting power is just, then it is generating node; Injecting power is negative, be then load bus; Injecting power is zero, be then transmission node.Once determine the type of node, net electric generation or the load of each node just can be known.

The connection between node can be known, i.e. the distribution on limit in network according to the connection of rack.According to the maximum delivery power of circuit and transformer, determined the maximum delivery power on every bar limit by the mode of summation.

After establishing topology of networks, according to the component parameters in space truss project, set up the bus admittance matrix Y of whole network.Get the susceptance part of admittance matrix.

B＝Im(Y)(1)

To susceptance matrix inversion, reactance matrix can be obtained

X＝B ^-1(2)

Then, according to the network topology of above-mentioned formation, use the Dijkstra's algorithm for solving signal source shortest path try to achieve respectively all generating-load buses in network between shortest path.

As other embodiments, solve signal source shortest path and can also adopt other algorithms, such as Floyd algorithm, dynamic programming algorithm and intelligent optimization algorithm etc.

Then, the betweenness index of computing node and limit (branch road), comprise pure graph theory betweenness, node and limit based on the electric betweenness of PTDF of circuit capacity, node and the limit electric betweenness of PTDF based on transmission capacity, this corresponds to step 3.Wherein PTDF (PowerflowTransmissionDistributionFactor) refers to trend transmission distribution factor.

Concrete, first calculate pure graph theory betweenness, at this, employing be betweenness algorithm based on shortest path, along each shortest path, add up the quantity of the shortest path that each node and limit are passed through in whole network, be their pure graph theory betweenness.

Then, computing node and limit are based on the electric betweenness of PTDF of circuit capacity.For any node v in network, make its electric betweenness

$T\left(v\right)=\frac{1}{2}\underset{g\∈G}{\Σ}\underset{d\∈D}{\Σ}{G}_{gd}\underset{l\∈v}{\Σ}\left|h_{l-gd}\right|,v\≠g\≠d\∈B---\left(3\right)$

Wherein, B is the set of all buses, and G, D are respectively generator node and load bus set.H _l-gdthe injection active power of node g, d when changing+1 and-1 respectively, the overpowering variable quantity in circuit l upper reaches, i.e. PTDF.In above formula (3), C _gdit is the power upper limit considering all circuit l in whole network time, generator load is to the peak power can transmitted between (g, d).Its computing formula is

$C_{gd}=\underset{l\∈L}{\min }\left\{\frac{P_l^{max}}{\left|h_{l-gd}\right|}\right\}---\left(4\right)$

Wherein, L is the set of all circuits.

Make the bus at branch road l two ends number and be respectively m and n, then, when+1 ,-1 change occurs the injecting power of node i-j respectively, the power variation that branch road l passes through, namely PTDF is

$h_{l-gd}=\frac{X_{mi}-X_{mj}-X_{ni}+X_{nj}}{x_{mn}}---\left(5\right)$

Wherein, X _abthat reactance matrix X a is capable, the element of b row.X _mnthe reactance of branch road l, if susceptance matrix is known, then

$x_{mn}=-\frac{1}{B_{mn}}---\left(6\right)$ Wherein, B _mnthat susceptance matrix B m is capable, the element of the n-th row.

Use and variously can calculate the electric betweenness of PTDF of each node based on circuit capacity above.The PTDF betweenness of circuit (or limit) is defined as

T(l)＝max{T ^p(l),|T ⁿ(l)|}(6)

Wherein, T ^p(l) and | T ⁿ(l) | represent the positive betweenness through transmission line l and negative betweenness respectively:

$T^p\left(l\right)=\underset{g\∈G}{\Σ}\underset{d\∈D}{\Σ}{C}_{gd}{h}_{l-gd},\∀h_{l-gd}>0---\left(7\right)$

$T^n=\underset{g\&Element;G}{\&Sigma;}\underset{d\&Element;D}{\&Sigma;}{C}_{gd}{h}_{l-gd},\&ForAll;h_{l-gd}<0---\left(8\right)$

Finally, computing node and limit are based on the electric betweenness of PTDF of transmission capacity.The electric betweenness of circuit (limit)

$B\left(l\right)=\left|\underset{g\&Element;G}{\&Sigma;}\underset{d\&Element;D}{\&Sigma;}{w}_{gd}{h}_{l-gd}\right|---\left(9\right)$

Wherein, h _l-gdbe generating node i and load bus j when injecting unit active power 1 and-1 respectively, circuit l produces active power, and direction is from m to n, and this parameter is exactly trend transmission distribution factor PTDF; w _gdbe from generating node i to the weight of load bus j electric energy transmitting size, get the value of a less side in rated generation capacity and rated load demand.Namely

w _gd＝min{S _g,S _d}(10)

Except the electric betweenness of the transmission node of generator except load bus can be pushed away by be directly the connected electric betweenness of circuit of cumulative node therewith

$B\left(v\right)=\frac{1}{2}\underset{l\&Element;v}{\&Sigma;}B\left(l\right)---\left(11\right)$ For generating and load bus, when calculating electric betweenness, the injecting power of these nodes should be counted.For generating node, its electric betweenness is

$B\left(v\right)=\frac{1}{2}\&lsqb;\underset{l\&Element;v}{\&Sigma;}B\left(l\right)+\underset{d\&Element;D}{\&Sigma;}{w}_{vd}\&rsqb;,v\&Element;G---\left(12\right)$

Here by w _gd=min{S _g, S _dsue for peace about generating node g and all load bus d.

Similar with it, for load bus, its electric betweenness is

$B\left(v\right)=\frac{1}{2}\&lsqb;\underset{l\&Element;v}{\&Sigma;}B\left(l\right)+\underset{d\&Element;G}{\&Sigma;}{w}_{vd}\&rsqb;,v\&Element;D---\left(13\right)$

As other embodiments, the betweenness algorithm based on shortest path is adopted to the calculating of the betweenness on each node and limit, is not limited to the algorithm based on shortest path that the present embodiment adopts, can also other algorithms be adopted.

Finally, corresponding to step 4, respectively the above-mentioned betweenness index calculated is sorted, filter out possible fragile node, and then generating network index result.Concrete, walk according to first three the three kinds of betweenness calculated, filter out node and limit that three kinds of index rankings are all positioned at front 20, and according to a certain betweenness descending sort, can check and analysis to facilitate.For RTS-48 node system, shown in the node obtained after screening and branch road are listed as follows:

In table 1RTS-48 system, three kinds of betweenness are all positioned at the bus of front 20

In table 2RTS-48 system, three kinds of betweenness are all positioned at the branch road of front 20

The fragile node shown in table 3 and branch road list can be drawn by above-mentioned table 1, table 2.

The fragile node of table 3RTS-48 example system and branch road list

Bus is numbered	216	215	123	116		217	113
								Branch number	123-120	123-217	215-113	217-216		215-216	216-214
Bus is numbered	203	111	110	214		109
								Branch number	214-211	116-114	114-111	113-111

As other embodiments, additive method can also be adopted to realize predicting the static fragile link of electrical network, be not limited to the embodiment that the present embodiment provides.

Corresponding to step 5, the dynamic fragile link of electrical network is predicted.Concrete, as shown in Figure 2, first read network parameter, then specify attack strategies and remove element ratio, strategy below concrete use respectively carries out attack simulating to network:

Node is attacked at random: random erasure node from electrical network at every turn.Get the mean value of Multi simulation running result as net result.

Node betweenness is attacked: each node that a deletion betweenness is maximum from electrical network, recalculates index subsequently, then the node that the next betweenness of random erasure is maximum.Get the mean value of Multi simulation running result as net result.

The circuit capacity PTDF betweenness of node is attacked: each node that a deletion circuit capacity PTDF betweenness is maximum from electrical network, recalculates index subsequently, then deletes the maximum node of next circuit capacity PTDF betweenness.

The transmission capacity PTDF betweenness of node is attacked: each node that a deletion transmission capacity PTDF betweenness is maximum from electrical network, recalculates index subsequently, then deletes the maximum node of next transmission capacity PTDF betweenness.

Circuit is attacked at random: random erasure branch road from electrical network at every turn.Get the mean value of Multi simulation running result as net result.

Circuit betweenness attacks each branch road that a deletion betweenness is maximum from electrical network, recalculates index subsequently, then the node that the next betweenness of random erasure is maximum.Get the mean value of Multi simulation running result as net result.

The circuit capacity PTDF betweenness of circuit is attacked: each branch road that a deletion circuit capacity PTDF betweenness is maximum from electrical network, recalculates index subsequently, then deletes the maximum branch road of next circuit capacity PTDF betweenness.

Circuit transmission capacity PTDF betweenness is attacked: each branch road that a deletion transmission capacity PTDF betweenness is maximum from electrical network, recalculates index subsequently, then deletes the maximum branch road of next transmission capacity PTDF betweenness.

And remove element ratio, be then in attack simulating process, if node is under attack, this node and coupled all circuits will be removed from electrical network; If the branch road between two nodes is under attack, all both-end elements on this branch road will be removed from electrical network.

Bus in network or element is progressively removed respectively according to the evaluation index of above attack strategies and formulation, and after removing end each time, calculate and record the state index of electrical network, until the number removing bus or element reaches the number percent of specifying, here, bus or element stop the condition removing be lose load proportion reach 99.9% or the ratio that removes element reach designated value.Reach 99.9% or after the ratio that removes element reaches designated value, then generating network attacks statistics losing load proportion, show that state index is about the change curve removing bus or element ratio and increase, network attack calculates and terminates.

State index used herein is load supply rate and the largest connected subset size of network.

Specify the load supply amount of electrical network

P _LS＝min{P _G,P _D}(14)

Wherein, P _gand P _dthat the total meritorious of electrical network is exerted oneself and total burden with power respectively.If electrical network is unconnected, then for each connection subset calculated load supply, then the load supply amount of all connection subsets of whole electrical network directly can be added up.

After then under attack, the load supply rate of electrical network can calculate according to the following formula:

$R_{LS}=\frac{P_{LS}}{P_{LS,0}}\&times;100\%---\left(15\right)$

The computing method of largest connected subset size are:

$R_{MCS}=\frac{N}{N_0}\&times;100\%---\left(16\right)$

Wherein, N be under attack, remove part of nodes or branch road after, the number of nodes of the largest connected subset of network; N ₀it is the number of nodes of former network.

As other embodiments, the method that the present embodiment provides is not limited to the prediction of the dynamic fragile link of electrical network, other embodiments can also be adopted to be achieved.

Corresponding to step 6, the fragile link screened in step 4 and step 5 is integrated: common factor is got in fragile for static state link set and dynamic fragile link set, and the fragile link obtaining whole bulk power grid obtains the fragile link of whole bulk power grid.

Above-described embodiment gives the present invention concrete embodiment; the fragile link of the bulk power grid that the bulk power grid fragility Forecasting Methodology that the present invention is based on Complex Networks Theory predicts; electrical reticulation design personnel can according to the high betweenness node screened and bus; the wiring of electrical network is adjusted, considers the protective measure strengthening fragile node.And on this basis, more different electrical reticulation design scheme time under attack, the difference of state index dynamic change trend, thus filter out more rational electrical reticulation design scheme.

Be presented above the concrete embodiment that the present invention relates to three themes, but the present invention is not limited to described embodiment.Under the thinking that the present invention provides; the mode easily expected to those skilled in the art is adopted to convert the technological means in above-described embodiment, replace, revise; and the effect played goal of the invention that is substantially identical with the relevant art means in the present invention, that realize is also substantially identical; the technical scheme of such formation is carried out fine setting to above-described embodiment and is formed, and this technical scheme still falls within the scope of protection of the present invention.

Claims (10)

1. a Forecasting Methodology for bulk power grid fragility, is characterized in that, in turn includes the following steps:

Step 1: the constraint condition obtaining Net Frame of Electric Network layout data and system cloud gray model, comprising: space truss project data, system cloud gray model constraint condition;

Step 2: according to the complex network mathematical model of the constraint condition structure electrical network obtained;

Step 3: for above-mentioned complex network mathematical model, uses the betweenness algorithm based on shortest path to calculate pure graph theory betweenness, the electric betweenness of trend distribution factor based on circuit capacity, the electric betweenness of trend distribution factor based on transmission capacity on each node and limit;

Step 4: according to the three kinds of betweenness obtained in step 3, picks out several all forward nodes of three kinds of index rankings and limit, using the static state fragile link of bus, circuit and the transformer corresponding to these nodes and limit as electrical network;

Step 5: based on kind of the betweenness index of three in step 3, use attack simulating strategy, remove the bus in network or element, and calculate the state index removing at every turn and terminate rear electrical network, until state index is lower than certain number percent, these buses removed or element form the dynamic fragile link of network;

Step 6: the fragile link of static state draw respectively step 4 and step 5 and dynamic fragile link are integrated: common factor is got in fragile for static state link set and dynamic fragile link set, obtains the fragile link of whole bulk power grid.

2. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 1, it is characterized in that, the space truss project data in described step 1 are: node data, transmission line data, transformer data, load data; System constraints is: each genset is exerted oneself the peak power output of higher limit and circuit.

3. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 1, it is characterized in that, in described step 2, the structure of electrical network complex network mathematical model comprises the following steps:

S01: using all buses as node processing, circuit and transformer process as limit, the topological model of structure electrical network;

S02: build the constraint condition that power grid topology model runs;

S03: calculate the shortest path between often pair of generator-load bus in power grid topology model;

S04: according to the reactance of circuit in power grid topology model, forms the impedance matrix of system, and calculates the trend transmission distribution factor between often pair of node thus;

S05: the condition calculated according to above-mentioned steps, builds electrical network complex network mathematical model.

4. a kind of Forecasting Methodology of bulk power grid fragility according to claim 1 or 3, it is characterized in that, topological model in described step S01 comprises following essential information: generating node, transmission node, load bus, and the quiet meritorious of node is exerted oneself, the branch road reactance value be made up of circuit and transformer.

5. a kind of Forecasting Methodology of bulk power grid fragility according to claim 1 or 3, is characterized in that, what in described step S03, the calculating of shortest path adopted is Dijkstra's algorithm.

6. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 1, it is characterized in that, the certain value that in described step 4, ranking is forward is 20, after filtering out the node and limit that three kinds of index rankings are all positioned at front 20, by these three kinds of indexs according to a certain betweenness descending sort, set up the static fragile link list of electrical network.

7. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 1, it is characterized in that, in described step 5, the Forecasting Methodology of bulk power grid fragility is: respectively with random attack strategies, and progressively remove bus in network or element by the calculated attack strategy based on each betweenness index, applied load fill rat and the largest connected subset size of network are as prediction index, until load supply rate and largest connected subset size are lower than the number percent of specifying, the bus be now removed or element constitute the dynamic weak link of system.

8. a kind of Forecasting Methodology of bulk power grid fragility according to claim 1 or 7, it is characterized in that, attack simulating strategy in described step 5 comprises: node is attacked at random, node betweenness is attacked, the circuit capacity PTDF betweenness of node is attacked, and the transmission capacity PTDF betweenness of node is attacked, and circuit is attacked at random, the circuit capacity PTDF betweenness of circuit is attacked, and circuit transmission capacity PTDF betweenness is attacked.

9. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 8, it is characterized in that, in the engineering of described attack simulating strategy, node is under attack, all circuits removing this node and be connected with this node from electrical network; Branch road is under attack, then remove all both-end elements of this branch road.

10. the Forecasting Methodology of a kind of bulk power grid fragility according to claim 1, it is characterized in that, the state index in described step 5 is load supply rate and network maximum UNICOM subset size.