CN104881723A - Source-included power distribution network fault power failure scheme optimization method considering influence of equivalent load points - Google Patents

Abstract

The invention belongs to the technical field of power grid distribution, and particularly relates to a source-included power distribution network fault power failure scheme optimization method considering the influence of equivalent load points. The method includes the steps of making a power distribution network a power distribution network load graph equivalently and determining the number and grade of each load point; acquiring the power of each load point and obtaining power-lossing load functions of each load point during the power failure time period; obtaining user importance degree indexes by an objective weighting method based on user importance influence indexes; obtaining load point importance degree index functions through addition calculation; stimulating a distributed power supply and an upper-grade power supply by means of a Monte Carlo method to obtain the curves of power failure capacity and power failure quantity needed by a system; making statistics of user point power failure loss conditions to obtain power failure loss functions and obtaining the equivalent weight functions of each load point; establishing target functions of minimum equivalent power-lossing capacity and minimum equivalent load loss amount of the system; and making the optimization solution of the target functions to obtain an optimal power failure scheme.

Description

That considers the impact of equivalent load point contains source distribution network failure power failure scheme optimization method

Technical field

The invention belongs to electrical network distribution technique field, particularly relate to a kind of consider that equivalent load point affects containing source distribution network failure power failure scheme optimization method.

Background technology

Along with going deep into of the growing of urban power distribution network and urban network restructuring engineering, the distribution network structure of China becomes increasingly complex, and management area is also increasing.In order to improve the reliability of whole distribution network, for distribution network provides secure support, the access of distributed power source not only adds certain for subsequent use for distribution network, and by the utilization as the clean energy resource such as wind energy, sun power, also progressively achieves the Optimum utilization of the energy.That mentions in the present invention contains source power distribution network, and what refer to is exactly the distribution network accessed containing distributed power source, and distributed power source can be wind power generating set, solar electrical energy generation unit, garbage power unit, miniature gas turbine, fuel cell etc.

The basic goal that distribution network runs is that the generating capacity of power distribution network and generated energy will meet workload demand.Indication of the present invention is containing source distribution network, and its power supply capacity should equal the capacity sum to whole transformer capacities of this supplying power allocation, the station capacity of powering directly to this power distribution network and all distributed power sources of accessing this power distribution network.Break down because upper level power supply or transformer may have, situation that abnormal running, short time ration the power supply or stop transport because of maintenance, and the distributed power source accessed its exert oneself randomness and intermittent feature of having and changing with resource situation difference, therefore, when there is above reason, power distribution network generating capacity likely can not meet the demand of load at that time; Now can select to have a power failure to sub-load or operation of rationing the power supply, ensure the even running of distribution system.Therefore select optimum power failure scheme most important.

The optimum power failure scheme of power distribution network is distribution network automated important component part.Within the set time period, under the prerequisite for certain capacity, electricity must be stopped, how to select to stop for user or load point then very important.Different power failure schemes, stopping for different user, is different for user's impact separately, except loss of outage economically, also may relate to user have a power failure produce social influence, capacity impact, electricity impact is even on the impact of environment.This reflects that user of different nature is different for the degree of dependence of electric system, therefore needs these information summaries to consider.

If in the selection of power failure scheme, need in conjunction with all load point relevant informations, the definition so for load point is then very important.All load point are divided into equivalent load point and user's point by the present invention, and user's point and power directly flow to the node of certain particular user; Equivalent load point is the power dividing point in distribution network, flows to from this power the different user group that this point connects respectively.Load level, contract capacity, power load situation and the environmental protection contribution situation that can consider user is put for each user, obtain user's important coefficient, thus by calculating the load point important coefficient that can obtain each and comprise equivalent load point and user's point.Such sorting technique, makes, when doing the selection of power failure scheme, not only can refine to the control of each load point being cut-off to situation, and can consider the important coefficient of each load point, obtains optimum power failure scheme.

Summary of the invention

At genset section failure, when electricity generation system is exerted oneself and can not be met workload demand, seek optimum load power failure scheme in order to whole containing source distribution system, the present invention proposes a kind of consider that equivalent load point affect contain source distribution network failure power failure scheme optimization method, comprising:

Step 1, according to the distribution network structural drawing in distribution region of delimiting, form distribution network load tree graph, determine the distribution situation of each load point, numbering and rank; Tree is made up of load point load, and load point is divided into equivalent load point and user's point, and equivalent load point refers to the branch node that each active power is shunted, and user's point refers to the outflow point of distribution network least significant end power; When power flows through each load point according to the direction of regulation, the rank accepting the load point of power comparatively provides the rank of the load point of power to want elevated by one step;

Step 2, gather each load point power, obtain the dead electricity function of load of each load point in power off time section; According to customer charge grade, user's contract capacity, user's peak load value and user's environmental protection of enterprise hierarchical level that user puts, adopt objective weighted model to compose weight, and linear weighted function is carried out to indices, obtain user's importance degree index; All equivalent load points under each equivalent load point and user's point are searched for, obtains the importance degree index value of all load point of its subordinate, obtain load point importance degree target function;

Step 3, according to the demand curve of each load point obtained, Monte Carlo method is adopted to carry out state simulation to distributed power source and upper level power supply, and simultaneously according to load condition, calculate the curve of the power failure capacity of the system requirements under various running status, the power failure electricity of system requirements;

Step 4, add up all users point occur different duration power-off condition under loss of outage situation, obtain the loss of outage function of whole load point, the resident choosing loss of outage value in the unit interval minimum compares as benchmark, obtains the equivalent weighting function of each load point under difference power failure duration;

Step 5, set up minimum with the equivalent dead electricity amount of system and equivalent load loss amount is minimum for objective function, constraint condition comprises: the power failure electricity that system actual power failure capacity is greater than the power failure capacity of system requirements, system actual power failure electricity is greater than system requirements, system actual send out that power is not less than workload demand, each generator output meets Power generation limits, each distributed power source is exerted oneself and met Power generation limits; Equivalence dead electricity amount equals the product of dead electricity function of load in timing statistics and load point importance degree target function; Equivalent load loss amount equals the product of dead electricity function of load in timing statistics and equivalent weighting function;

Step 6, the improved adaptive GA-IAGA based on successively revising fitness function weight coefficient is adopted to carry out solving of multiple-objection optimization to objective function; By carrying out binary coding to raw data, reflect that the switch of the different time sections in distribution network structure cut-offs situation, calculate fitness function afterwards, iteration all revises its weight coefficient to the objective function item in fitness function each time, makes it to trend towards the convergence of required result more fast; After repeatedly solving, obtain a string 1,0 coded sequence, be final optimum power failure scheme combination and separate, this coded sequence also represents the combination of the situation of cut-offfing of each period of each load point in timing statistics.

In described step 2, the computation process of load point importance degree target function comprises:

Step 201, collect each user affect information, and determine importance influence index numerical value;

Customer charge grade S1 refers to that, according to power consumer grade separation, S1 value corresponds to respectively: three class user=1, two class user=2, class user=3, premium user=4;

User's contract capacity S2 refers to the power consumption equipment capacity numerical value allowing attaching for low-voltage customer; Centering high voltage customer refers to the transformer that is directly connected on receiving voltage circuit and directly joins high-tension motor capacity sum;

User's peak load value S3 refers to the peak load value of user in timing statistics section, by the load data information of user in raw data, obtains the peak load value of each user in timing statistics;

User's environmental protection of enterprise hierarchical level S4 refers to the comprehensive evaluation result of country for environmental protection of enterprise, and is divided into five kinds of colors to represent environmental protection of enterprise behavior grade, is respectively the yellow red black of Green Blue; User's environmental protection of enterprise hierarchical level S4 value is green=5, blue=4, yellow=3, red=2, black=1;

Step 202, employing objective weighted model carry out tax weight to customer impact information index, carry out the importance influence index that COMPREHENSIVE CALCULATING obtains user's point afterwards by weigthed sums approach;

If represent the user's point in striked distribution network with i, j represents the importance influence index of each user point, wherein i=1,2 ..., n, n are the number of user's point in distribution network; J=1 in the present invention, 2,3,4; First j customer impact information index under putting i user, forms raw data customer impact information matrix A,

$A={\left[x_{\mathrm{ij}}\right]}_{n\×m}=\left[\begin{array}{cccc}{x}_{11}& x_{22}& ...& x_{1m}\\ x_{21}& x_{22}& ...& x_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ x_{n1}& x_{n2}& ...& x_{\mathrm{nm}}\end{array}\right],$

Respectively under corresponding jth item index, the importance influence index of i-th user's point;

Step 203, employing average variance method ask for the importance influence index weight of every user point, comprising:

A. first need to be normalized raw data, carry out nondimensionalization by index; Adopt linear scaling method, respectively the data of each row are processed;

Pass through formula the customer impact information index matrix obtained after nondimensionalization the following is:

$A^{\′}={\left[{x_{\mathrm{ij}}}^{\′}\right]}_{n\×m}=\left[\begin{array}{cccc}{x_{11}}^{\′}& {x_{22}}^{\′}& ...& {x_{1m}}^{\′}\\ {x_{21}}^{\′}& {x_{22}}^{\′}& ...& {x_{2m}}^{\′}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ {x_{n1}}^{\′}& {x_{n2}}^{\′}& ...& {x_{\mathrm{nm}}}^{\′}\end{array}\right]$

B. the variance of each index is calculated,

${\mathrm{\σ}}_j=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{({x_{\mathrm{ij}}}^{\′}-\stackrel{\‾}{{x_j}^{\′}})}^2}{n}},$

Wherein, n counts for user,

C. the weighted value of parameter j

ω _j＝σ _j/∑σ _j,j＝1,2,...,m

D. linear weighted function determines the importance degree index of each user point

$u_i={\mathrm{\ω}}_1{x}_{i1}+{\mathrm{\ω}}_2{x}_{i2}+...+{\mathrm{\ω}}_m{x}_{\mathrm{im}}=\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ω}}_j{x}_{\mathrm{ij}},$ Wherein i=1,2 ... n, represent user's point, m represents that user puts affects information index, m=4;

Step 204, for all equivalent load points under each equivalent load point and user point search for, obtain the importance degree index value of all load point of its subordinate, load point importance degree target function u (R) asks for, and carry out summation operation, thus obtain the importance degree index of each rank equivalent load point, by calculating the importance degree index that can obtain whole load point, and the different importance degree value u of corresponding different load point obtain load point importance degree target function u (R).

The power failure capacity of the system requirements in described step 3 and the power failure electricity of system requirements are asked for process and are comprised:

Step 301, simulation initialisation, read in conventional power unit and distributed power source unit data, comprise capacity, failure rate, random series of exerting oneself model, and sequential load data;

Step 302, generate distributed unit output timing curve and each normal power supplies state sequence transfer time;

Step 303, random sampling is carried out to the state of normal power supplies and distributed unit, determine the running status of all genset;

Step 304, state based on load, all genset, according to the constraint of electricity generation system power balance, judge whether to need to carry out power failure operation, if needed, and the power failure capacity of computing system demand p _lrepresent the load power that power distribution network connects, P _girepresent conventional power unit power, P _dGirepresent the distributed power source power of the assembling unit, N _grepresent conventional power unit quantity, N _dGrepresent the conventional power unit distributed number formula power supply power of the assembling unit; And record the power failure electricity of the system requirements of this period, if do not needed, return step 303;

Step 305: simulate the operating states of the units in the whole time period, judge whether to simulate the period, if completed, add up the power failure capability value of the system requirements in each moment, and power failure capability value is determined that the period quadratures, the power failure electricity of system requirements can be obtained, if do not completed, return step 303.

In described step 5, objective function specifically comprises:

$\min f_1\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}u\left(R\right)P_c(\mathrm{\λ},R)T$

$\min f_2\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}D(T_R,R)P_c(\mathrm{\λ},R)$

F ₁x () is system equivalence dead electricity amount, f ₁x () is system equivalent load loss amount, R represents that load point is numbered; R (λ) is the load point set that can have a power failure in the corresponding period; P _c(λ, R) represents that load point R is the dead electricity function of load of λ in the period that has a power failure, and u (R) represents load point importance degree target function, and λ represents the time hop count in timing statistics, λ=1,2 ... n, T are the duration of every period, D (T _r, R) represent and consider that user has a power failure the equivalent weighting function of consequence impact, namely load point is R, accumulative power failure duration is T _rtime functional value, T _rrepresent the accumulation power failure duration of load point R.

Beneficial effect of the present invention:

1) the present invention is by being distribution network load tree graph by distribution network figure equivalency transform, embodies the relation in network structure between load point more intuitively.By to the redefining of load point all in network, classify and the division of rank, all load point are divided into equivalent load point and user's point, and an equivalent load point minute rank is divided into k level, embody the relation between load point, what power failure scheme is refine to each load point in distribution network structure cut-offs situation simultaneously.By load point classify and grading method for distinguishing, the influence factor of user side can also being taken in calculating, by asking for load point importance degree index, obtaining the power failure prioritization scheme of final consideration load point importance.

2) objective function of the present invention is that system equivalence dead electricity amount is minimum and equivalent load loss amount is minimum, and the impact mainly from user side on power failure scheme takes in obtained power failure scheme.The load point equivalence dead electricity amount wherein proposed, for load point actual dead electricity amount is multiplied long-pending with load point importance degree index, this value considers load point importance degree, for power failure Scheme Choice makes guidance.

3) mathematical model acquiring method of the present invention adopts and solves based on the genetic algorithm successively revising weight coefficient, namely iteration all revises its weight coefficient to the objective function item in fitness function each time, makes it to trend towards the convergence of required result more fast.Weight for fitness function carries out successive iteration adjustment, make the refinement that in evolutionary process, some useful information continue Target Preference, obtained gradually towards the search pressure of ideal point by successive adjustment weight, be more conducive to population evolution and to the close speed of objective function.

4) in power system model, add the impact of distributed power source access herein, consider the stochastic model of distributed power source, for the power failure scheme optimization containing source power distribution network provides guidance.

Accompanying drawing explanation

Fig. 1 is electricity generation system equal-value map;

Fig. 2 is distribution network block diagram example;

Fig. 3 is distribution network load tree isoboles;

Fig. 4 is based on genetic algorithm process flow diagram;

Fig. 5 is overview flow chart of the present invention;

Wherein, G-traditional type genset, DG-distributed power source unit, the load that PL-power distribution network connects; 1 ~ 8 is the numbering of equivalent load point, and 9 ~ 17 is the numbering of user's point.

Embodiment

Below in conjunction with accompanying drawing, embodiment is elaborated.

At genset section failure, when electricity generation system is exerted oneself and can not be met workload demand, select load point equivalence dead electricity amount and the comprehensively minimum the best power failure scheme of equivalent load loss amount.According to above mathematical model, concrete operation step process flow diagram is shown in Fig. 5, and concrete operation step is summed up and PART is described as follows.

Electric system is simplified to the electricity generation system situation as Fig. 1 by the present invention when calculating power balance.Wherein, G represents to the transformer of this supplying power allocation and the generating plant of traditional type that powers directly to this power distribution network, as thermal power generation, hydropower etc.; DG represents the distributed power source of this power distribution network of access, as wind-power electricity generation, photovoltaic generation, garbage power, miniature gas turbine generating etc.; PL represents the load that this power distribution network connects.Namely do not consider the impact that the structure of distribution network causes, herein by means of only power input and the load flow artificial situation of electricity generation system, carry out discussing system and whether meet power balance and electric quantity balancing constraint.

The division of step 1, formation distribution network load tree graph and load point

In the present invention, as the distribution network figure that Fig. 2 represents, distribution network load tree graph can be equivalent to, as shown in Figure 3.Distribution network load sets the network connection relation and propagation that embody between load point, and tree is made up of load point load, produces the shunting load point that namely formation one is new whenever there being power.The concept of load tree embodies the subordinate relation be similar between different load point between trunk and branch.

All load point of whole distribution network, can be divided into equivalent load point and user's point, and the load distribution situation of distribution network, can be reduced to and utilize equivalent load point and user to put to represent.

Equivalent load point represents, every in distribution network have the point of power dividing to be all referred to as an equivalent load point, as being numbered 1 in Fig. 2 and 3,2,3,4,5,6,7,8 are equivalent load point, represent in figure 3 with circle, and its meaning embodied in load tree is the meaning of trunk.

User puts and represents that the power at the least significant end of distribution system flows out point, certain unique user namely directly accessed or the transformer of sole user.As in Fig. 2 and 3, be numbered 9,10,11,12,13,14,15,16,17 are user's point, represent in figure 3 with arrow.

Shown in the figure set according to load, all equivalent loads point minute rank is represented, the equivalent load point of a common k rank can be obtained like this according to load tree from top to bottom, load point rank is up to the 1st grade, as 1 point in Fig. 3, minimum rank is k level, if load point in Fig. 38 is the 6th grade.Wherein, the k-1 level equivalent load point of each branch all contains all k level equivalent load points and user's point of subordinate.Like this, by building distribution network load tree, utilize user to put and equivalent load point, just the network structure relation of whole power distribution network can be showed, each load point on off state is aided with power load distributing just can express concrete power failure scheme.

Asking for of step 2, load point importance degree target function

The load point importance degree considered in the present invention analyzes from user's side angle degree.Different user is different for the degree of dependence of electric system, reflects the difference of different user for the service requirement of power distribution network; Meanwhile, because the power failure Scheme Choice of power distribution network is directly related with user side, therefore when doing power distribution network power failure scheme, need the influence factor of user side to be considered.

Because different power failure schemes may affect making a difference property of different user, the present invention, by the COMPREHENSIVE CALCULATING to user side indication information, obtains user's importance degree index u; By this index, can weigh out after considering all kinds of index of user, the importance degree of different user, be used as the index of a comprehensive consideration Demand-side impact, and during it is calculated in order to the scheme optimization that has a power failure.For the calculating thinking of index u, first by determining the user profile index analysis of consideration customer impact in the present invention, what obtain each user affects information index, afterwards by adopting objective weighted model to compose weight to different customer impact information indexes, and linear weighted function is carried out to indices, each affects information index in interior composite target can to obtain reflection user, and this index is user's importance degree index u.

Consider user side influence index, the relevant information of the user side that Water demand will be considered when setting power failure scheme.The present invention adopts peak load and these four indexs of user's environmental protection of enterprise hierarchical level of user in customer charge grade, user's contract capacity, timing statistics.

Due in the present invention, load point is divided into user's point and equivalent load point, user is put and the computing method of importance degree index of equivalent load point also different.Concrete solution procedure is as follows.

Step 201: the determination of customer impact information index

First need to collect raw data and process, that namely determines each user affects information index numerical value, needs the amount determined to comprise: the load level of user, the contract capacity of user, user's peak load value and user's environmental protection of enterprise hierarchical level.

S1 customer charge grade refers to according to power consumer grade separation, the rank index adopted, and premium user is that the highest grade, and three class users are that grade is minimum.Customer charge grade S1 value corresponds to respectively: three class user=1, two class user=2, class user=3, premium user=4.

S2 user's contract capability value, refers to the power consumption equipment capacity numerical value allowing attaching for low-voltage customer; Centering high voltage customer refers to the transformer that is directly connected on receiving voltage circuit and directly joins high-tension motor capacity sum.

S3 user's peak load value, by the load data information of user in raw data, obtains the peak load value of each user in timing statistics;

S4 user's environmental protection of enterprise hierarchical level refers to the comprehensive evaluation result of country for environmental protection of enterprise, and is divided into five kinds of colors to represent environmental protection of enterprise behavior grade, is respectively the yellow red black of Green Blue.Environmental protection of enterprise hierarchical level index S 4 value is that green=5, blue=4, yellow=3, red=2, black=1 are to represent environmental protection of enterprise behavior grade from high in the end.

Step 202: asking for of the importance degree index that user puts

In the present invention, adopt objective weighted model to carry out tax weight to customer impact information index, carry out by weigthed sums approach the importance degree index that COMPREHENSIVE CALCULATING obtains user's point afterwards.

Objective weighted model is a kind of method that size of the quantity of information provided according to indices observed reading carrys out agriculture products weight.Because each use has four customer impact information index numerical value per family, for all user's points in a distribution region, customer impact information index matrix can be listed,

$A={\left[x_{\mathrm{ij}}\right]}_{n\×m}=\left[\begin{array}{cccc}{x}_{11}& x_{22}& ...& x_{1m}\\ x_{21}& x_{22}& ...& x_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ x_{n1}& x_{n2}& ...& x_{\mathrm{nm}}\end{array}\right],$ N represents that user counts, and m represents customer impact information index, m=4 in the present invention.To comprehensively ask for user's importance degree index, need index to carry out asking for weight.Because each index is to the influence degree of synthesis result, need the numerical value by comparing each user under this index, the index that numerical value differs greatly, the impact in last COMPREHENSIVE CALCULATING is larger.The objective weighted model adopted in the present invention is average variance method, and the standard deviation of certain index is larger, illustrates in same index, and each scheme value gap is larger, and in comprehensively asking for, role is larger, and its weight is also larger; On the contrary, the standard deviation of certain index is less, and in comprehensively asking for, role is less, and its weight is also less.The method step is comparatively simply clear, has objectivity; Objective Weight is carried out based on " variance drive " principle, namely the provided information gap opposite sex is larger, also larger in the impact of result, such tax power process more intuitively embodies this meaning, also better reflects the influence degree of each customer impact information index in specific distribution region from the result composing power.Adopt method of weighted mean afterwards, synthetic user each affect information index, be weighted average after obtain user and put importance degree index.Specifically ask for process as follows.

If represent the user's point in striked distribution network with i, j represent each user point importance influence index (wherein i=1,2 ..., n, n are the number of user's point in distribution network; J=1 in the present invention, 2,3,4).First j customer impact information index under putting i user, forms raw data customer impact information matrix A,

$A={\left[x_{\mathrm{ij}}\right]}_{n\×m}=\left[\begin{array}{cccc}{x}_{11}& x_{22}& ...& x_{1m}\\ x_{21}& x_{22}& ...& x_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ x_{n1}& x_{n2}& ...& x_{\mathrm{nm}}\end{array}\right],$ Respectively under corresponding jth item index, the index value of i-th user's point.

Process below for adopting average variance method to ask for every customer impact information index weight:

A. first need to be normalized raw data, carry out nondimensionalization the present invention by index and adopt linear scaling method, respectively the data of each row (namely under each index) are processed.

Pass through formula the customer impact information index matrix obtained after nondimensionalization the following is,

$A^{\′}={\left[{x_{\mathrm{ij}}}^{\′}\right]}_{n\×m}=\left[\begin{array}{cccc}{x_{11}}^{\′}& {x_{22}}^{\′}& ...& {x_{1m}}^{\′}\\ {x_{21}}^{\′}& {x_{22}}^{\′}& ...& {x_{2m}}^{\′}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ {x_{n1}}^{\′}& {x_{n2}}^{\′}& ...& {x_{\mathrm{nm}}}^{\′}\end{array}\right]$

B. the variance of each index is calculated,

${\mathrm{\σ}}_j=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{({x_{\mathrm{ij}}}^{\′}-\stackrel{\‾}{{x_j}^{\′}})}^2}{n}},$

Wherein, n counts for user,

C. the weighted value of parameter j

ω _j＝σ _j/∑σ _j,j＝1,2,...,m

D. linear weighted function determines the importance degree index of each user point

wherein i=1,2 ... n, represent user's point, m represents that user puts affects information index.

Step 203: load point importance degree target function u (R) asks for

Above-mentioned steps describes the acquiring method that user puts importance degree index.For the acquiring method of the importance degree index of each equivalent load point, need to search for all equivalent load points under this rank equivalent load point and user's point, obtain the importance degree index value of all load point of its subordinate, and carry out summation operation, thus obtain the importance degree index of each rank equivalent load point.As in Fig. 3, u ₈=u ₁₆+ u ₁₇, u ₇=u ₈+ u ₁₅wherein u ₁₅, u ₁₆and u ₁₇for user puts the importance degree index of 15,16 and 17.

By calculating the importance degree index that can obtain whole load point, and the different importance degree value u of corresponding different load point can obtain load point importance degree target function u (R).

Step 3, the power failure capacity asking for system requirements and power failure electricity, need to ask for according to the running status of electricity generation system and Real-time Load situation.Here Monte Carlo method is adopted to stop calculating for capacity to demand.Common power, i.e. higher level's transformer or energy source, employing failure rate is λ _grepair rate is μ _gtwo state models; Distributed power source unit is except consideration malfunction, also adopt the probabilistic model that output power changes with resource situation, as Wind turbine adopts WTG statistical history Wind speed model, the discrete model etc. that photovoltaic unit adopts the average power supply of day part corresponding to statistical history period intensity of illumination to exert oneself.

Calculate the power failure capacity of the system requirements when the period is λ, use formula concrete steps are as follows,

Step 301: simulation initialisation, reads in conventional power unit and distributed power source unit data, comprises capacity, failure rate, series model etc. immediately of exerting oneself, and sequential load data;

Step 302: generate distributed unit output timing curve and each normal power supplies state sequence transfer time;

Step 303: random sampling is carried out to the state of normal power supplies and distributed unit, determines the running status of all genset;

Step 304: based on the state of load, all genset, according to the constraint of electricity generation system power balance, judges whether to need to carry out power failure operation, if needed, and the power failure capacity of computing system demand and record the power failure electricity of the system requirements of this period, if do not needed, return step 303;

Step 305: simulate the operating states of the units in the whole time period, has judged whether to simulate the period, if completed, has added up the power failure capability value of the system requirements in each moment, afterwards to the P tried to achieve _needto determining that the period quadratures, the power failure electricity W of system requirements can be obtained _needif do not completed, return step 303.

Step 4, calculate consider user have a power failure consequence impact equivalent weighting function D (T _r, R) refer to, the equivalence power failure weight of different load point under difference power failure accumulation duration.To load point power failure consequences analysis in the present invention, consider from economic factors angle, namely by calculating the loss of outage size W of load point, afterwards with the loss of outage value of a certain user for benchmark, compare with it with other user and obtain equivalent weighted value, be that load point is R, accumulation power failure duration is T _rtime this load point equivalent weight.

That the present invention adopts load point loss of outage size is load point loss of outage function W (T _r, R), namely represent that load point R is T at accumulation power failure duration _rtime corresponding loss of outage, unit is unit/kW.The calculating of loss of outage adopts the variation relation of the interruption cost section in time for dissimilar user wherein, C _iTrepresent that i-th user is T at interruption duration _rhour interruption cost numerical value C (unit/kWh), asking for of this numerical value, need to carry out loss of outage investigation to different user, the loss of outage value of investigation user under the power-off condition that different duration occurs is estimated (direct economic loss of outage and the indirect loss summation brought of resuming production after having a power failure).Add up the loss of outage value situation of all users of each national economy classification, carry out curve fitting and obtain this classification Custom interruption cost with period change list, as shown in table 1, the loss of outage of equivalent load point, then according to the covering scope of load point, to add up the specific power loss of outage section delta data in time obtaining each load point to Custom interruption cost; T _rrepresent the accumulation power failure duration of load point in timing statistics section, be T _r=n × T _λ, n represents the power off time hop count of the load point R that the program is corresponding, T _λfor every period.W (T of the present invention _r, R) illustrate corresponding to each load point different durations under loss of outage value.As the W of load point 8 equivalent in Fig. 3 ₈(T ₈) value is W ₁₆(T ₈)+W ₁₇(T ₈), represent that equivalent load point 8 is T at power failure accumulation duration ₈time, its value is T for user puts 16 and 17 at power failure accumulation duration ₈unit loss of outage.

Sorted users specific power loss of outage change list (unit/kW in time in certain distribution region of table 1 ^-1)

The loss of outage function W (T of the whole load point of network is being obtained by calculating _r, R) after, choose a minimum resident of loss of outage value in the unit interval as benchmark, setting its D value is 1, does with it ratio, namely by the loss of outage value under the Different periods of other all user equivalent weight D value corresponding under obtaining all the other each load point Different periods.If the user one (resident) in table 1 is as benchmark, then the equivalent weight of user one under each has a power failure the period is all set to 1; So the equivalent weight calculation under each power failure duration of user three is: accumulative power failure duration is 1 hour, accumulation power failure duration is 2 hours, ... finally obtain the equivalent weight of each period of user three.

By calculating the loss of outage value between whole load point, and being compared, the equivalent weighting function D (T of each load point under difference power failure duration can be obtained _r, R).

Step 5, set up minimum with the equivalent dead electricity amount of system and equivalent load loss amount is minimum for objective function, objective function specifically describes as follows,

Objective function is 1.: make the equivalent dead electricity amount of system minimum

In a period of time, the dead electricity amount of load point is the product of dead electricity time and dead electricity load.When considering load point important coefficient, accumulate the dead electricity amount of all power off time sections,

$\min f_1\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}u\left(R\right)P_c(\mathrm{\λ},R)T$

In formula, R represents that load point is numbered; R (λ) is the load point set that can stop in the corresponding period; P _c(λ, R) represents that load point R is the dead electricity function of load of λ in the period that has a power failure, the unit kw that the present invention gets; U (R) represents load point importance degree target function; λ represents the time hop count in timing statistics, and T is the duration of every period, the unit that the present invention gets for hour.

Objective function is 2.: make system equivalent load loss amount minimum

$\min f_2\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}D(T_R,R)P_c(\mathrm{\λ},R)$

In formula, R represents that load point is numbered; D (T _r, R) represent and consider that user has a power failure the equivalent weighting function of consequence impact, namely load point is R, accumulative power failure duration is T _rtime functional value; T _rrepresent the accumulation power failure duration of load point R, the unit that the present invention gets for hour; λ represents the power off time section in timing statistics, being one day, carrying out segmentation by one day according to 24 hours as added up duration in the present invention.

Constraint condition comprises:

The actual power failure capacity of system is greater than the power failure capacity of system requirements

$P_{\mathrm{need\λ}}\≤\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}}$

Actual power failure electricity is greater than the power failure electricity of system requirements

$W_{\mathrm{need\λ}}\≤\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}}$

System actual send out power and be not less than workload demand

$\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}+\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}}\≥\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}$

Each generator output meets bound constraint

P _Gimin≤P _Gi≤P _Gimax，i∈N _G

Each distributed generator is exerted oneself and is met bound constraint

0≤P _DGi≤P _DGimax，i∈N _DG

Wherein, P _gifor the meritorious of generator i is exerted oneself; P _dGifor the meritorious of distributed power source is exerted oneself; P _{cr λ}and P _{need λ}for the load summate variable of load point r when the period is λ and the power failure load of system requirements, P _drfor the burden with power demand of load point, the unit that the present invention gets is kw; W _{cr λ}and W _{need λ}for the actual power failure electricity of load point r when the period is λ and system requirements power failure electricity, the unit that the present invention gets is kwh; P _giminand P _gimaxfor the Power generation limits of generator i; N _g, N _dGand R _dfor the set of all load point comprised in all generators, distributed power source and distribution region.

The determination of objective function prerequisite situation and scope of statistics

The situation of this objective function discussion is, discuss when electricity generation system be in because of unit break down, abnormal running, short time ration the power supply or under the state that can not normally run because maintenance is stopped transport, need by removal of load, reduce the demand of load side, ensure system stable operation.By making a choice to the situation of cut-offfing of different load point, obtain stopping for load point and equivalent dead electricity amount, loss of outage, in numerous power failure scheme, selecting with these two targets is minimum power failure scheme, is optimum power failure scheme.

Before setting up objective function, first set the time range (as asking for a year, power failure scheme in January or one day) of power failure scheme, in the present invention, the duration of setting statistics is for the time being one day, and according to hour being divided into 24 periods.

What objective function known conditions was asked for further illustrates

In objective function 1, P _c(λ, R) function representation be the dead electricity load of load point in a certain period.Do section process to statistics duration, if statistics duration is divided into N section, then the workload demand added up in duration can be expressed as N number of level curve and to be connected the piecewise function obtained.P _cwhat (λ, R) represented is that load power value is with the function selecting period and the change of power failure load point.

In objective function 1, the meaning that R (λ) regional extent represents is, the λ period may be used for the scope of the load point calculating equivalent dead electricity amount.According to genetic algorithm of the present invention, as shown in Figure 4, the power failure scheme combining random of initial acquisition, the situation that this power failure branch road higher level and subordinate load point all have a power failure may be comprised, when if produce, in same load path, the superior and the subordinate's load point is in power failure operation state simultaneously, the load point that then priority level is higher takes power failure operation, and more low-level load point is according to not power-off operational processes., avoid because higher level's equivalent load point carries out double counting to comprising of subordinate load point when calculating power failure load value to ensure like this.And in fact in this load path, the real work situation of all load point of subordinate of this equivalent load point R should be whole disconnection, just encoded in order to the calculating of the scheme that has a power failure and be placed in the position of connection.Such as, in load tree graph 3, if equivalent load point 6 is power failure operation, all subordinate load points of its correspondence all carry out encoding setting according to not having a power failure at (comprising equivalent load point and user's point), namely (7,8,14,15,16,17) all as connection process.

In objective function 1, u (R) function is the function of reflection load point importance index, whole objective function f ₁what x () represented is consider the important sex equivalent dead electricity amount of load point.Namely by being multiplied by load point importance function coefficient, actual for original load point dead electricity amount is modified to load point equivalence dead electricity amount, this value reflects the mistake charge condition considering load point importance.

Solving of step 6, majorized function

Adopt and carry out solving this multiple objective function based on the improved adaptive GA-IAGA successively revising weight coefficient.First weight assignment to fitness function and makeover process are described.

1. the weight makeover process of fitness function

On the basis of traditional genetic algorithm, the formation of fitness function is revised, after each calculated fitness function, weight calculation is carried out to this function, the optimization aim formed for every generation is individual, all re-starts weight calculation, makes the refinement that in evolutionary process, some useful information continue Target Preference, obtained by successive adjustment weight and trend towards the evolution of optimal anchor direction gradually, be more conducive to population evolution and to the close speed of objective function.Be the makeover process of fitness function below.

In the process of genetic evolution, per in generation, all produces the new progeny population P of some individualities composition.In all individualities of current population, optimization object function f ₁maximal value and minimum value be respectively:

$Z_{f_1}^{\max }=\max \left\{f_1(\mathrm{\λ},R)\right\}$

$Z_{f_1}^{\min }=\min \left\{f_1(\mathrm{\λ},R)\right\}$

Similar, objective function f can be obtained ₂maximal value and minimum value with working as former generation, to the weighted value of each objective function as shown in the formula asking for:

wherein j=1, corresponding two objective functions of 2 difference.

For the λ that the given individual i of population P is corresponding _iand R _i, have:

$Z_{f_1}=f_1({\mathrm{\λ}}_i,R_i),Z_{f_2}=f_2({\mathrm{\λ}}_i,R_i)$

Then can obtain considering that the fitness objective function item of weight is:

$Z(\mathrm{\λ},R)=\underset{j=1}{\overset{2}{\mathrm{\Σ}}}{c}_j(Z_{f_j}-Z_{f_j}^{\min })$

Namely ${f_1}^{\′}=c_1(Z_{f_1}-Z_{f_1}^{\min }),f_2^{\′}=c_2(Z_{f_2}-Z_{f_2}^{\min }),$ Wherein f ₁' and f ₂' be revised functional value.

Revised genetic algorithm fitness function is:

$\mathrm{Fit}(\mathrm{\λ},R)=\underset{j=1}{\overset{2}{\mathrm{\Σ}}}{c}_j(Z_{f_j}-Z_{f_j}^{\min })+W,$ Wherein, W is penalty term.

2. the concrete implementation step that the present invention is based on the improved adaptive GA-IAGA successively revising weight coefficient is as follows:

Step 601: input all load point (comprising all equivalent loads point and the user's point) load value corresponding to Different periods.

Step 602: carry out initial code to raw data, carries out initialization process to population P, adopts binary coding form herein.With the genic value of each individuality in equally distributed generating random number initial population.

The phenotype n=of input data [1,2 ..., N] represent all load point (comprising all equivalent loads point and user's point) numbering, wherein N is load point number.The genotype of data adopts 0 and 1 to represent, 0 represents that this load point does not take power-off measure, and 1 represents that this load point takes power-off measure.Like this, at each fixing period t _λin, 0,1 matrix that one group of line number is 1, columns is N can be formed, as [1,0,0 ..., 0,1], all load point that what this section of coded sequence represented is within this fixing period cut-off situation; Altogether λ time period in total timing statistics, each individuality can be expressed as that line number is 1, columns is the matrix sequence of N × λ, is N × λ position gene order of this individuality.Suppose that the generation population number formed is m.

Step 603: according to multiple objective function and constraint condition model, build genetic algorithm fitness function model, fitness calculating is carried out to population P, and fitness function is revised.

Known target function is

$\min f_1\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}u\left(R\right)P_c(\mathrm{\λ},R)T_{\mathrm{\λ}}$

$\min f_2\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}D(T_R,R)P_c(\mathrm{\λ},R)$

Constraint condition is

$P_{\mathrm{need\λ}}\≤\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}}$

$W_{\mathrm{need\λ}}\≤\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}}$

$\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}+\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}}\≥\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}$

The fitness function model that can obtain genetic algorithm is

Fit(λ,R)＝f ₁+f ₂+cf ₃+df ₄+ef ₅

When constraint condition meets, make f ₃=0, f ₄=0, f ₅=0, fitness function can be written as

Fit(λ,R)＝f ₁+f ₂

When constraint condition does not meet, make respectively

$f_3=P_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}},$

$f_4=W_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}},$

$f_5=\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}-\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}-\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}},$

Then the fitness function of genetic algorithm can be write as,

$\mathrm{Fit}(\mathrm{\λ},R)=f_1+f_2+c(P_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}})+d(W_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}})+e(\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}-\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}-\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}})$

Wherein, c, d, e are penalty term coefficient, and value is c respectively ₁, d ₁, e ₁as penalty term coefficient.The determination of these three coefficients, namely the intensity size of penalty term function is determined, by the size value of penalty term function, the fitness function not meeting constraint condition can be carried out the amplification in proper range, thus in the process selected, the scheme not meeting constraint be screened out.If the numerical value intensity of penalty term function is too little, some individuals still likely destroys constraint condition, so do not ensure that the individuality that genetic operation obtains must be the feasible solution meeting constraint condition; If the intensity of penalty term function is too large, likely makes again individual fitness difference little, reduce the competitive power between individuality, thus affect the operational efficiency of genetic algorithm.Therefore need to ensure that objective function item is consistent with penalty term numerical value magnitude by the value of penalty term coefficient, as the value by penalty term coefficient, suitable penalty term functional value can be obtained, help the screening meeting selection scheme.As can be seen from two objective functions, unit demand is kw, loss of outage unit is unit, chronomere is hour, objective function and penalty difference should between 1-100 times, and therefore value penalty term coefficient range is (0,100), and by the debugging in computation process, objective function and penalty term function can be made to adapt.

After each calculated fitness function, by carrying out weight calculation to objective function item, the objective function part of fitness function being formed and revised, forming new fitness function, for the next generation carry out fitness function ask for time.First time calculates and adopts original genetic algorithm fitness function formula, does not do weight coefficient and calculates.Detailed process is see the makeover process of fitness function above.

The fitness function that each revised constraint does not meet in situation is,

$\mathrm{Fit}(\mathrm{\λ},R)={f_1}^{\′}+f_2^{\′}+c(P_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}})+d(W_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}})+e(\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}-\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}-\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}})$

Be

$\mathrm{Fit}(\mathrm{\λ},R)=\underset{j=1}{\overset{2}{\mathrm{\Σ}}}{c}_j(Z_{f_j}-Z_{f_j}^{\min })+c(P_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{P}_{\mathrm{Cr\λ}})+d(W_{\mathrm{need\λ}}-\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}{W}_{\mathrm{Cr\λ}})+e(\underset{r\∈R}{\mathrm{\Σ}}{P}_{\mathrm{Lr}}-\underset{i\∈N_G}{\mathrm{\Σ}}{P}_{\mathrm{Gi}}-\underset{j\∈N_{\mathrm{DG}}}{\mathrm{\Σ}}{P}_{\mathrm{DGi}})$

When constraint condition meets, Fit (λ, R)=f ₁'+f ₂', be

$\mathrm{Fit}(\mathrm{\λ},R)=\underset{j=1}{\overset{2}{\mathrm{\Σ}}}{c}_j(Z_{f_j}-Z_{f_j}^{\min })$

Step 604: selection course, by ideal adaptation degree size, copies out from population by a certain proportion of individuality, forms a new population, to realize the continuity of individuality " excellent genes ".This process is the survival of the fittest process based on fitness function.The selection method of operating adopted herein is random ergodic sampling, and individual selected probability calculates according to the ratio of fitness assignment, in formula, P _ifor the probability that individual i is selected; f _ifor the fitness of certain individual i; M is quantity individual in population.

Step 605: intersection genetic recombination process, is undertaken the part in parent two chromosomes exchanging the new individual filial generation of alternative formation.Coded system is herein binary coding, and the single-point in adopting scale-of-two to intersect intersects.

Step 606: gene mutation process, the gene after the new intersection of formation carries out the process made a variation with minimum probability.Adopt scale-of-two mutation process herein, row stochastic upset is entered to the chromosome variable in binary coding.

Step 607: the population at individual obtaining a new generation, and the iteration returning that step c carries out next time, regulation reaches q=q when iterations _stime, iteration stopping, obtains optimum solution.

This embodiment is only the present invention's preferably embodiment; but protection scope of the present invention is not limited thereto; anyly be familiar with those skilled in the art in the technical scope that the present invention discloses, the change that can expect easily or replacement, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (4)

1. consider that equivalent load point affects containing a source distribution network failure power failure scheme optimization method, it is characterized in that, comprising:

Step 1, according to the distribution network structural drawing in distribution region of delimiting, form distribution network load tree graph, determine the distribution situation of each load point, numbering and rank; Tree is made up of load point load, and load point is divided into equivalent load point and user's point, and equivalent load point refers to the branch node that each active power is shunted, and user's point refers to the outflow point of distribution network least significant end power; When power flows through each load point according to the direction of regulation, the rank accepting the load point of power comparatively provides the rank of the load point of power to want elevated by one step;

Step 2, gather each load point power, obtain the dead electricity function of load of each load point in power off time section; According to customer charge grade, user's contract capacity, user's peak load value and user's environmental protection of enterprise hierarchical level that user puts, adopt objective weighted model to compose weight, and linear weighted function is carried out to indices, obtain user's importance degree index; All equivalent load points under each equivalent load point and user's point are searched for, obtains the importance degree index value of all load point of its subordinate, obtain load point importance degree target function;

Step 3, according to the demand curve of each load point obtained, Monte Carlo method is adopted to carry out state simulation to distributed power source and upper level power supply, and simultaneously according to load condition, calculate the curve of the power failure capacity of the system requirements under various running status, the power failure electricity of system requirements;

Step 4, add up all users point occur different duration power-off condition under loss of outage situation, obtain the loss of outage function of whole load point, the resident choosing loss of outage value in the unit interval minimum compares as benchmark, obtains the equivalent weighting function of each load point under difference power failure duration;

Step 5, set up minimum with the equivalent dead electricity amount of system and equivalent load loss amount is minimum for objective function, constraint condition comprises: the power failure electricity that system actual power failure capacity is greater than the power failure capacity of system requirements, system actual power failure electricity is greater than system requirements, system actual send out that power is not less than workload demand, each generator output meets Power generation limits, each distributed power source is exerted oneself and met Power generation limits; Equivalence dead electricity amount equals the product of dead electricity function of load in timing statistics and load point importance degree target function; Equivalent load loss amount equals the product of dead electricity function of load in timing statistics and equivalent weighting function;

Step 6, the improved adaptive GA-IAGA based on successively revising fitness function weight coefficient is adopted to carry out solving of multiple-objection optimization to objective function; By carrying out binary coding to raw data, reflect that the switch of the different time sections in distribution network structure cut-offs situation, calculate fitness function afterwards, iteration all revises its weight coefficient to the objective function item in fitness function each time, makes it to trend towards the convergence of required result more fast; After repeatedly solving, obtain a string 1,0 coded sequence, be final optimum power failure scheme combination and separate, this coded sequence also represents the combination of the situation of cut-offfing of each period of each load point in timing statistics.

2. method according to claim 1, it is characterized in that, in described step 2, the computation process of load point importance degree target function comprises:

Step 201, collect each user affect information, and determine importance influence index numerical value;

Customer charge grade S1 refers to that, according to power consumer grade separation, S1 value corresponds to respectively: three class user=1, two class user=2, class user=3, premium user=4;

User's contract capacity S2 refers to the power consumption equipment capacity numerical value allowing attaching for low-voltage customer; Centering high voltage customer refers to the transformer that is directly connected on receiving voltage circuit and directly joins high-tension motor capacity sum;

User's peak load value S3 refers to the peak load value of user in timing statistics section, by the load data information of user in raw data, obtains the peak load value of each user in timing statistics;

User's environmental protection of enterprise hierarchical level S4 refers to the comprehensive evaluation result of country for environmental protection of enterprise, and is divided into five kinds of colors to represent environmental protection of enterprise behavior grade, is respectively the yellow red black of Green Blue; User's environmental protection of enterprise hierarchical level S4 value is green=5, blue=4, yellow=3, red=2, black=1;

Step 202, employing objective weighted model carry out tax weight to customer impact information index, carry out the importance influence index that COMPREHENSIVE CALCULATING obtains user's point afterwards by weigthed sums approach;

If represent the user's point in striked distribution network with i, j represents the importance influence index of each user point, wherein i=1,2 ..., n, n are the number of user's point in distribution network; J=1 in the present invention, 2,3,4; First j customer impact information index under putting i user, forms raw data customer impact information matrix A,

$A=[x_{\mathrm{ij}}{]}_{n\×m}=\left[\begin{array}{cccc}{x}_{11}& x_{22}& ...& x_{1m}\\ x_{21}& x_{22}& ...& x_{2m}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ x_{n1}& x_{n2}& ...& x_{\mathrm{nm}}\end{array}\right],$

Respectively under corresponding jth item index, the importance influence index of i-th user's point;

Step 203, employing average variance method ask for the importance influence index weight of every user point, comprising:

A. first need to be normalized raw data, carry out nondimensionalization by index; Adopt linear scaling method, respectively the data of each row are processed;

Pass through formula the customer impact information index matrix obtained after nondimensionalization the following is:

$A^{\′}=[{x_{\mathrm{ij}}}^{\′}{]}_{n\×m}=\left[\begin{array}{cccc}{x_{11}}^{\′}& {x_{22}}^{\′}& ...& {x_{1m}}^{\′}\\ {x_{21}}^{\′}& {x_{22}}^{\′}& ...& {x_{2m}}^{\′}\\ .& .& & .\\ .& .& & .\\ .& .& & .\\ {x_{n1}}^{\′}& {x_{n2}}^{\′}& ...& {x_{\mathrm{nm}}}^{\′}\end{array}\right]$

B. the variance of each index is calculated,

${\mathrm{\σ}}_j=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{({x_{\mathrm{ij}}}^{\′}-\stackrel{\‾}{{x_j}^{\′}})}^2}{n}},$

Wherein, n counts for user,

C. the weighted value of parameter j

${\mathrm{\ω}}_j={\mathrm{\σ}}_j/\mathrm{\Σ}{\mathrm{\σ}}_j,j=\mathrm{1,2},...,m$

D. linear weighted function determines the importance degree index of each user point

$u_i={\mathrm{\ω}}_1{x}_{i1}+{\mathrm{\ω}}_2{x}_{i2}+\·\·\·+{\mathrm{\ω}}_m{x}_{\mathrm{im}}=\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\ω}}_j{x}_{\mathrm{ij}},$ Wherein i=1,2 ... n, represent user's point, m represents that user puts affects information index, m=4;

Step 204, for all equivalent load points under each equivalent load point and user point search for, obtain the importance degree index value of all load point of its subordinate, load point importance degree target function u (R) asks for, and carry out summation operation, thus obtain the importance degree index of each rank equivalent load point, by calculating the importance degree index that can obtain whole load point, and the different importance degree value u of corresponding different load point obtain load point importance degree target function u (R).

3. method according to claim 1, it is characterized in that, the power failure capacity of the system requirements in described step 3 and the power failure electricity of system requirements are asked for process and are comprised:

Step 301, simulation initialisation, read in conventional power unit and distributed power source unit data, comprise capacity, failure rate, random series of exerting oneself model, and sequential load data;

Step 302, generate distributed unit output timing curve and each normal power supplies state sequence transfer time;

Step 303, random sampling is carried out to the state of normal power supplies and distributed unit, determine the running status of all genset;

Step 304, state based on load, all genset, according to the constraint of electricity generation system power balance, judge whether to need to carry out power failure operation, if needed, and the power failure capacity of computing system demand p _lrepresent the load power that power distribution network connects, P _girepresent conventional power unit power, P _dGirepresent the distributed power source power of the assembling unit, N _grepresent conventional power unit quantity, N _dGrepresent the conventional power unit distributed number formula power supply power of the assembling unit; And record the power failure electricity of the system requirements of this period, if do not needed, return step 303;

Step 305: simulate the operating states of the units in the whole time period, judge whether to simulate the period, if completed, add up the power failure capability value of the system requirements in each moment, and power failure capability value is determined that the period quadratures, the power failure electricity of system requirements can be obtained, if do not completed, return step 303.

4. method according to claim 1, it is characterized in that, in described step 5, objective function specifically comprises:

$\min f_1\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}u\left(R\right)P_c(\mathrm{\λ},R)T$

$\min f_2\left(x\right)=\underset{\mathrm{\λ}=1}{\overset{n}{\mathrm{\Σ}}}\underset{R\∈R\left(\mathrm{\λ}\right)}{\mathrm{\Σ}}D(T_R,R)P_c(\mathrm{\λ},R)$

F ₁x () is system equivalence dead electricity amount, f ₁x () is system equivalent load loss amount, R represents that load point is numbered; R (λ) is the load point set that can have a power failure in the corresponding period; P _c(λ, R) represents that load point R is the dead electricity function of load of λ in the period that has a power failure, and u (R) represents load point importance degree target function, and λ represents the time hop count in timing statistics, λ=1,2 ... n, T are the duration of every period, D (T _r, R) represent and consider that user has a power failure the equivalent weighting function of consequence impact, namely load point is R, accumulative power failure duration is T _rtime functional value, T _rrepresent the accumulation power failure duration of load point R.