CN103746376B - A kind of power distribution network dynamic power restoration method based on load curve - Google Patents

Abstract

Based on a power distribution network dynamic power restoration method for load curve, belong to technical field of power distribution system automation.The present invention utilizes computer, pass through program, first input basic parameter, calculate meritorious, the load or burden without work of each node each hour again and determine grid structure when normally running, determine the grid structure after the supply path of fault section and mains side and Fault Isolation, then candidate's grid structure set of each hour minimum mistake load during using traditional genetic algorithm determination fault restoration, the optimum dynamic power recovery scheme during finally determining fault restoration.The present invention makes full use of load curve, formulate Dynamic-Recovery scheme by the hour, grid structure is changed with load variations during fault restoration, thus there is the power supply interrupted district that can effectively reduce because fault causes, shorten customer outage hours, reduce the features such as Custom interruption cost.The present invention can be widely used in power distribution network service restoration.

Description

A kind of power distribution network dynamic power restoration method based on load curve

Technical field

The invention belongs to technical field of power distribution system automation, be specifically related to a kind of power distribution network dynamic power restoration method based on load curve.

Background technology

The production of electric energy and people and live closely related, reliable continued power is to one of basic demand of power system operation.The power distribution network being connected to power transmission network and user is the important step of distributing electric energy, is the important component part of electric power system.The power distribution network had a large capacity and a wide range is difficult to the generation avoiding fault in actual motion, fault outage will cause economic loss to user, have influence on the quality of life of people, therefore power distribution network often adopts closed loop design and open loop operation, effectively can recover the power supply of power failure load when fault occurs.Power distribution network service restoration refers to, the power distribution network broken down is after determining abort situation and carrying out isolated fault, and the load being recovered power supply interrupted district by the interconnection switch between the block switch on reasonable operation distribution feeder and feeder line is powered.The power supply interrupted district because fault causes can be reduced by power distribution network service restoration and shorten the interruption duration of user, thus reduce the loss of outage of user and increase the electricity sales amount of power supply department.Therefore, study effective power distribution network service restoration method, the reliability tool of the fail safe and power supply that improve power distribution network operation is of great significance.

The service restoration method of existing power distribution network, as in " exploitation of the object-oriented service restoration expert system of consideration load variations " (Development of an Object-Oriented Service Restoration Expert System With Load Variations) literary composition in " electric power system " proceedings (IEEE Transactions on Power Systems) of the 23rd volume the 1st phase IEEE in 2008, disclosed method considers the expert system service restoration method of load variations.The concrete steps of described method are: progressively analyze the situation that fault causes user to have a power failure and recover, namely first, calculate the available current surplus of the feeder line main for subsequent use be connected with the interconnection switch of each power failure section, according to abort situation, the merger of interconnective power failure section is become some groups, when the load current of each power failure section group exceedes the available current-carrying surplus of the feeder line main for subsequent use be connected by interconnection switch in set period, this interconnection switch can not be used for the service restoration of this power failure section group, otherwise this interconnection switch is used for the service restoration of this power failure section group.Secondly, when also having power failure section not restore electricity, breakout place section in this power supply interrupted district and the power failure section between this section and fault section are formed main power failure section group, remaining interconnective power failure section forms power failure section group, the power failure members that main power failure section group is connected with the feeder line main for subsequent use with maximum available ampacity also, then recover as independent power failure section group by other power failure section groups.Again, when also having power failure section not restore electricity, from the interconnection switch that the ampacity of main feeder line for subsequent use can be made maximum, at every turn by a sector transfer of this main feeder line for subsequent use to this interconnection switch of connection and on the feeder line secondary for subsequent use run, until the load that as often as possible can recover power failure section is powered; When not obtaining suitable recovery scheme, another secondary feeder line for subsequent use is selected to repeat this step.Then, when having searched for all feeder lines secondary for subsequent use and having also had power failure section not restore electricity, more logical overladen continuous transfer realizes the service restoration of maximum power failure sections.Finally, guarantee that service restoration scheme meets the requirement of voltage constraint and the constraint of feeder line ampacity by Load flow calculation.In this service restoration method, consider that the principle of the termination service restoration process that load variations adopts is that the current-carrying surplus accepting the feeder line of power failure section load is less than the load current of power failure section in set period.The major defect of the method is: in fact the set period 1. in the method refers to fault correction time, therefore can only consider the change of load by fault correction time, load rule over time during can not considering fault restoration by the hour; The principle of the termination service restoration process 2. adopted can cause the service restoration scheme of the feeder line overload abandoning making acceptance power failure section load in certain period in fault correction time, thus causes the increase of sub-load interruption duration; 3. can only obtain a grid structure during fault restoration, can not adjust according to load variations the interruption duration that grid structure reduces user; 4. after obtaining final service restoration scheme, just judge whether scheme meets voltage constraint by Load flow calculation, when the program is discontented with afc voltage constraint, rational service restoration scheme can not be formulated.Therefore with the method can not make full use of load over time rule formulate effective service restoration scheme, the power supply interrupted district caused because of fault and the interruption duration shortening user can not be reduced, thus the loss of outage of user can not be reduced.

Summary of the invention

The object of the invention is the deficiency of the service restoration method for existing power distribution network, a kind of power distribution network dynamic power restoration method based on load curve is proposed, have in service restoration process, the change of payload is considered by the hour based on load curve, minimum for target to lose load, branch current and node voltage security constraint are that the grid structure of during the preferred distribution network failure of condition is repaired each hour forms the set of candidate's rack, and then can change with load variations during obtaining fault restoration, make the features such as the dynamic power recovery scheme of switch motion least number of times, thus provide scientific basis for formulating rational service restoration scheme, effectively can reduce the power supply interrupted district because fault causes, shorten the interruption duration of user and reduce the loss of outage of user.

The technical scheme realizing the object of the invention is: a kind of power distribution network dynamic power restoration method based on load curve, utilize computer, pass through program, first input basic parameter, calculate the active reactive load of each hour according to load curve again and determine the grid structure after grid structure when normally running, the supply path determining fault section and upstream and Fault Isolation, then candidate's grid structure set of each hour minimum mistake load during using traditional genetic algorithm determination fault restoration, the optimum dynamic power recovery scheme during finally determining fault restoration.The concrete steps of described method are as follows:

(1) basic parameter is inputted

First, the basic parameter needed for input calculates, comprises the basic parameter of power distribution network and the parameter of traditional genetic algorithm.The basic parameter of power distribution network comprises circuit sum (M), circuit number, circuit first and last end node numbering, line resistance (R), line reactance (X), line energizing flow amount, line switching type (not installing switch, installing block switch, installing interconnection switch), circuit place feeder line numbering; Node total number (N), power supply node sum (K _s), power supply node numbering, node burden with power (P _l), node load or burden without work (Q _l), the meritorious daily load rate curve of node load type, various load type; The node voltage upper limit (U _max), node voltage lower limit (U _min); Faulty line numbering, time of failure (t ₁), fault end time (t ₂); Reference voltage (U _b), reference power (S _b).The parameter of traditional genetic algorithm comprises population scale (P _p), maximum evolutionary generation (K _max), generation gap (P _g), crossing-over rate (p _c), aberration rate (p _m).

(2) calculate meritorious, the load or burden without work of each node each hour and determine grid structure when normally running

After (1) step completes, first according to the burden with power (P of each node of (1) step input _l), load or burden without work (Q _l) and the meritorious daily load curve of various load type, calculate the burden with power of each hour of each node and load or burden without work according to the principle that power factor is constant.Again according to the position of the interconnection switch installed in power distribution network, determine grid structure when normally running.

(3) grid structure after the supply path of fault section and mains side and Fault Isolation is determined

After (2) step completes, in grid structure when the normal operation that (2) step is determined, calculate for ease of optimizing and reduce computing time, first failure definition section is centered by fault point and the line set being border with the switch nearest from fault point; The supply path of failure definition section mains side is the line set between power supply node to fault section.Use conventional Depth Priority Algorithm again, whether install switch determination fault section according to faulty line head end; The supply path of fault section mains side is determined from the power supply node of faulty line place feeder line.The boundary switch isolated fault of the grid structure fault section when power distribution network finally determined according to (2) step normally runs, determines the grid structure after Fault Isolation.

(4) candidate's grid structure set of each hour minimum mistake load during using traditional genetic algorithm determination fault restoration

After (3) step completes, in the grid structure after the Fault Isolation that (3) step is determined, for obtaining the grid structure of the power distribution network each hour minimum mistake load during fault restoration after Fault Isolation, to determine candidate's grid structure set of dynamic power recovery scheme, during using the fault restoration obtained based on load curve, each hour load solves the optimal models of corresponding power distribution network service restoration.The target function of this model is minimum mistake load, and constraints is the constraints such as branch current, node voltage and Radiation Network shelf structure.Concrete formula is as follows:

$\min P_{t,c}\left(X_{t,r}\right)=\underset{n\∈{\mathrm{\Ω}}_N}{\mathrm{\Σ}}{P}_{t,n}-\underset{k\∈{\mathrm{\Ω}}_K}{\mathrm{\Σ}}{P}_{t,k}---\left(1\right)$

s.t.I _t,b≤I _b.max(2)

U _min≤U _t,k≤U _max(3)

g _t,i∈G (4)

In formula: t is the period, and the time of one day is divided into 24 hours, it it within every 1 hour, is 1 period; P _t,cfor the mistake load of period t; X _t,rfor the line switching state vector of the service restoration power distribution network of period t, X _t,r=[W _{t, r1}..., W _{t, rm}... W _{t, rM}]; M is circuit sum; W _{t, rm}for X _t,rthe on off state of m article of circuit, be taken as 1 when the switch that circuit does not install switch or installing is in closure state, be taken as 0 when the switch installed is in off-state; Ω _nand P _t,nbe respectively the node set of the power distribution network before service restoration and the node n burden with power at period t; Ω _kand P _t,kbe respectively the node set of service restoration power distribution network and the node k burden with power at period t; N and K is respectively the node total number of power distribution network before and after service restoration; I _t,b, U _t,kand g _t,ibe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k and grid structure; I _b.maxfor the ampacity of circuit b; U _maxand U _minbe respectively the bound of node voltage; G is the set of Radiation Network shelf structure.

During solving fault restoration by traditional genetic algorithm, the concrete steps of the grid structure of the minimum mistake load that each hour load is corresponding are as follows:

1) initial population is formed

After (3) step completes, according to the population scale (P of (1) step input _p), with { 0,1} binary code is encoded to the individuality that the random number produced is population scale, formed initial population.In initial population, the code length of each individuality is circuit sum, the circuit head end on off state of each coding stands power distribution network.When being encoded to 0, represent that this line switching is in off-state; When being encoded to 1, represent that this line switching is in closure state.

2) target function value individual in population is calculated

The (4)-1) after step completes, for evaluating the (4)-1) quality of individual performance in the initial population that formed of step, the target function value of individuality each in this population is calculated.Individual target function value is less illustrates that individual performance is more excellent.As follows to the concrete steps of individual calculating target function value each in described population:

1. individual UVR exposure is revised

The (4)-1) after step completes, to the (4)-1) step produce initial population in individuality carry out code modification, namely first the line switching state of the circuit and fault section mains side supply path of not installing switch in individual UVR exposure is set to 1; Again the on off state of fault section border circuit in individual UVR exposure is set to 0, realizes Fault Isolation.

2. individuality is decoded

The (4)-2)-1. walked after, according to the (4)-2)-1. walk the individual UVR exposure after revising to decode, namely according to the track data of revised individual UVR exposure and input, determine that individual UVR exposure breaker in middle state is the track data of 1 correspondence, realize individual decoding.

3. the Radiation distribution network that individual different electrical power node connects is formed

The (4)-2)-2. walked after, for obtaining the Radiation Network shelf structure that each power supply node connects, to the (4)-2)-2. walk the Depth Priority Algorithm of decoded individual line data application routine, first from different power supply node with interconnection switch place circuit for border carries out searching for the prompt radiation shape grid structure forming each power supply node and connect; Two end nodes that the circuit of interconnection switch is housed are searched for again respectively successively in prompt radiation shape grid structure.When interconnection two end node is all searched out-of-date, disconnect this interconnection switch and on off state corresponding for this interconnector be set to 0 in this individual UVR exposure to meet the requirement of Radiation Network shelf structure constraints formula (4), otherwise the decoded individual line data of the prompt radiation shape grid structure that search interconnection headend node or endpoint node connect and not searched mistake, form individual final Radiation Network shelf structure.

4. the trend of the Radiation distribution network that individual different electrical power node connects is calculated

The (4)-2)-3. walked after, first by the (4)-2)-3. walk the line set of the Radiation network configuration of trying to achieve, according to the power distribution network basic parameter of input and the active reactive load of each node during fault restoration of trying to achieve, determine each line impedance of the Radiation network that this period different electrical power node connects and each node active reactive load; Use again traditional before push back power flow algorithm and calculate the node voltage of Radiation distribution network and branch current that each power supply node connects.In calculating, when the trend of the Radiation power distribution network that certain power supply node connects does not restrain, terminate whole Load flow calculation.

5. individual target function value is calculated

The (4)-2)-4. walked after, when the (4)-2)-4. calculate the Radiation distribution network trend that different electrical power node connects in step and have when not restraining situation, the target function value of individuality is composed a larger value, namely 10 ⁹.When the trend of the Radiation power distribution network that all power supply nodes connect all restrains, calculate individual total out-of-limit value, concrete formula is:

$F_{t,\mathrm{pen}}=\underset{i-1}{\overset{K_S}{\mathrm{\Σ}}}(\underset{k\∈{\mathrm{\α}}_{t,i}}{\mathrm{\Σ}}\frac{U_{\min }-U_{t,k}}{U_B}+\underset{k\∈{\mathrm{\β}}_{t,i}}{\mathrm{\Σ}}\frac{U_{t,k}-U_{\max }}{U_B}+\underset{b\∈{\mathrm{\γ}}_{t,i}}{\mathrm{\Σ}}\frac{I_{t,b}}{I_{b.\max }})---\left(5\right)$

In formula: F _{t, pen}for the out-of-limit value of t period service restoration power distribution network; K _sfor power supply node sum; α _t,i, β _t,iand γ _t,ibe respectively the node voltage more node set of lower limit, the node voltage more node set of the upper limit and the set of fingers of line current overload of the Radiation power distribution network that t period power supply node i connects; U _maxand U _minbe respectively node voltage bound; U _bfor reference voltage; I _b.maxfor the ampacity of circuit b; I _t,b, U _t,kbe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k.

When the out-of-limit value of individuality is not 0, individual goal functional value is composed a higher value, namely 10 ⁹; When the out-of-limit value of individuality is 0, calculate individual target function value according to formula (1).

3) Evolution of Population calculates

The (4)-2)-5. walked after, use traditional genetic algorithm to be optimized calculating.First, defining current population is parent population, according to the (4)-2) target function value of individuality of trying to achieve in step calculates fitness value individual in parent population; Secondly, selection operation, interlace operation and mutation operation are carried out to parent population and obtains progeny population; Again, the (4)-2 is adopted) the whole individual goal functional values of step to described progeny population calculate; Then, described progeny population is inserted into parent population and obtains new population; Finally judge that Evolution of Population calculates whether to terminate, when Evolution of Population algebraically does not reach maximum evolutionary generation, evolutionary generation is added 1, and new population is repeated the calculating of this step as current population, until Evolution of Population algebraically reaches in (1) step the maximum evolutionary generation (K inputted _max) till, just obtain coding and the target function value of each individuality in population; Otherwise terminate Evolution of Population to calculate.

4) the rack set of minimum mistake load is determined

The (4)-3) after step completes, from the (4)-3) select the minimum mistake load individuality different with coding the Evolution of Population result of calculation that obtains of step and form the set of candidate's rack.

(5) the optimum dynamic power recovery scheme during determining fault restoration

After (4) step completes, in the rack set of the minimum mistake load determined in (4) step, first define the set that dynamic power recovery scheme is made up of the service restoration scheme of each hour during fault restoration.Again according to the hourage that network selection frame number as a child each during fault restoration and fault continue, enumerate the whole dynamic power recovery scheme of combination.Then the switch motion number of times of whole dynamic power recovery scheme is calculated.Minimum first assembled scheme of last selector switch action frequency is as optimum dynamic power recovery scheme.

After the present invention adopts technique scheme, mainly contain following effect:

1. the present invention is when considering load variations, for during fault restoration each hour payload formulate recover rack, compared with the method for existing consideration load variations, consider more careful, during taking into account fault restoration more accurately, load rule over time, make use of load curve more fully.

2. dynamic power recovery scheme of the present invention, consider for payload hourly during fault restoration, therefore during fault restoration, there is multiple different rack, can adjust grid structure according to the needs of load variations, scheme has time and space dynamic in good time.The present invention compared with the conventional method, can reduce customer outage hours, recovers more electricity.Meanwhile, the present invention also reduces the loss that brings because having a power failure suddenly of user by prior notice customer outage hours.

3. in the process of the present invention's each hour minimum mistake load candidate rack set during determining fault restoration, consider the security constraint of branch current and node voltage, the dynamic power recovery scheme therefore determined can be avoided occurring that final recovery scheme becomes the situation of inefficient solution owing to not meeting security constraint.

The present invention can be widely used in the service restoration of power distribution network, effectively can shorten the interruption duration of user, reduces the power failure area of power distribution network, improves the power supply reliability of power distribution network, for the formulation of power distribution network service restoration scheme provides effective scientific basis.

Accompanying drawing explanation

Fig. 1 is program flow chart of the present invention;

Fig. 2 is the schematic diagram of 70 node power distribution net wiring of embodiment.

In figure, for node serial number, 1 ~ 72 is circuit number, and #1 ~ #4 is feeder line numbering, for block switch or interconnection switch, for interconnection, for distribution line.

Embodiment

The present invention is further illustrated below in conjunction with embodiment.

Embodiment

As shown in Figure 1, 2, a kind of concrete steps of the power distribution network dynamic power restoration method based on load curve are as follows:

(1) basic parameter is inputted

First, the basic parameter needed for input calculates, comprises the basic parameter of power distribution network and the parameter of traditional genetic algorithm.The basic parameter of power distribution network comprises circuit sum (M=72), circuit number (1,2,, 72), circuit first and last end node numbering, line resistance (R), line reactance (X), line energizing flow amount, line switching type (do not install switch, installing block switch, installing interconnection switch), circuit place feeder line numbering; Node total number (N=70), power supply node sum (K _s=2), power supply node numbers (1,70), node burden with power (P _l), node load or burden without work (Q _l), the meritorious daily load rate curve of node load type, various load type; The node voltage upper limit (U _max=1.1p.u.), node voltage lower limit (U _min=0.9p.u.); Faulty line numbering (55), time of failure (t ₁=18), fault end time (t ₂=22); Reference voltage (U _b=11kV), reference power (S _b=100MVA).The parameter of traditional genetic algorithm comprises population scale (P _p=100), maximum evolutionary generation (K _max=300), generation gap (P _g=0.95), crossing-over rate (p _c=0.9), aberration rate (p _m=0.01).

(2) calculate meritorious, the load or burden without work of each node each hour and determine grid structure when normally running

After (1) step completes, first according to the burden with power (P of each node of (1) step input _l), load or burden without work (Q _l) and the meritorious daily load curve of various load type, calculate the burden with power of each hour of each node and load or burden without work according to the principle that power factor is constant.Again according to the position of the interconnection switch installed in power distribution network, determine grid structure when normally running.

For the power distribution network of Fig. 2, the grid structure calculating each node active reactive load of each hour and determine when normally running is described.According to the active reactive load of each node and the meritorious daily load curve of various load type, according to the principle that power factor is constant obtain at period t=18(18:00 to 19:00) time part of nodes load as shown in table 1.

Table 1 is at period t=18(18:00 to 19:00) time part of nodes load

As seen from Figure 2, the circuit installing interconnection switch has { 69,70,71,72}.The circuit of interconnection switch will not be installed as grid structure during normal operation in interconnection switch position according to installing.

(3) grid structure after the supply path of fault section and mains side and Fault Isolation is determined

After (2) step completes, in grid structure when the normal operation that (2) step is determined, calculate for ease of optimizing and reduce computing time, first failure definition section is centered by fault point and the line set being border with the switch nearest from fault point; The supply path of failure definition section mains side is the line set between power supply node to fault section.Use conventional Depth Priority Algorithm again, whether install switch determination fault section according to faulty line head end; The supply path of fault section mains side is determined from the power supply node of faulty line place feeder line.The boundary switch isolated fault of the grid structure fault section when power distribution network finally determined according to (2) step normally runs, determines the grid structure after Fault Isolation.

For the power distribution network of Fig. 2, the grid structure after the supply path and Fault Isolation determining fault section and upstream is described.When circuit 55 fault, the fault section using conventional Depth Priority Algorithm to search is for { 54,55,56}, fault section boundary switch circuit is { 54,56}; The supply path of fault section upstream is { 52,53,58,59,60,61}.To the switch 54 of grid structure open failure section boundaries circuit during normal operation, the grid structure after isolated fault after 56, can be obtained.

(4) candidate's grid structure set of each hour minimum mistake load during using traditional genetic algorithm determination fault restoration

After (3) step completes, in the grid structure after the Fault Isolation that (3) step is determined, for obtaining the grid structure of the power distribution network each hour minimum mistake load during fault restoration after Fault Isolation, to determine candidate's grid structure set of dynamic power recovery scheme, during using the fault restoration obtained based on load curve, each hour load solves the optimal models of corresponding power distribution network service restoration.The target function of this model is minimum mistake load, and constraints is the constraints such as branch current, node voltage and Radiation Network shelf structure.Concrete formula is as follows:

$\min P_{t,c}\left(X_{t,r}\right)=\underset{n\∈{\mathrm{\Ω}}_N}{\mathrm{\Σ}}{P}_{t,n}-\underset{k\∈{\mathrm{\Ω}}_K}{\mathrm{\Σ}}{P}_{t,k}---\left(1\right)$

s.t.I _t,b≤I _b.max(2)

U _min≤U _t,k≤U _max(3)

g _t,i∈G (4)

In formula: t is the period, and the time of one day is divided into 24 hours, it it within every 1 hour, is 1 period; P _t,cfor the mistake load of period t; X _t,rfor the line switching state vector of the service restoration power distribution network of period t, X _t,r=[W _{t, r1}..., W _{t, rm}... W _{t, rM}]; M is circuit sum; W _{t, rm}for X _t,rthe on off state of m article of circuit, be taken as 1 when the switch that circuit does not install switch or installing is in closure state, be taken as 0 when the switch installed is in off-state; Ω _nand P _t,nbe respectively the node set of the power distribution network before service restoration and the node n burden with power at period t; Ω _kand P _t,kbe respectively the node set of service restoration power distribution network and the node k burden with power at period t; N and K is respectively the node total number of power distribution network before and after service restoration; I _t,b, U _t,kand g _t,ibe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k and grid structure; I _b.maxfor the ampacity of circuit b; U _maxand U _minbe respectively the bound of node voltage; G is the set of Radiation Network shelf structure.

During solving fault restoration by traditional genetic algorithm, the concrete steps of the grid structure of the minimum mistake load that each hour load is corresponding are as follows:

1) initial population is formed

After (3) step completes, according to the population scale (P of (1) step input _p), with { 0,1} binary code is encoded to the individuality that the random number produced is population scale, formed initial population.In initial population, the code length of each individuality is circuit sum, the circuit head end on off state of each coding stands power distribution network.When being encoded to 0, represent that this line switching is in off-state; When being encoded to 1, represent that this line switching is in closure state.

For the power distribution network of Fig. 2, illustrate in the initial population formed with { the individuality of 0,1} binary code coding.Being encoded to of body one by one in the initial population produced at random when period t=18:

[1111110110 1110011101 1001111111 1111010111 1000111101 0001110100 1111111101 11]

The code length of this individuality is 72; Position 7 in coding, 10,14,15,19,22,23,35,37,42-44,49,51-53,57,59, the element of 60 and 69 is 0, represent that corresponding line switching is off state, the switch of all the other circuits is closure state.

2) target function value individual in population is calculated

The (4)-1) after step completes, for evaluating the (4)-1) quality of individual performance in the initial population that formed of step, the target function value of individuality each in this population is calculated.Individual target function value is less illustrates that individual performance is more excellent.As follows to the concrete steps of individual calculating target function value each in described population:

1. individual UVR exposure is revised

The (4)-1) after step completes, to the (4)-1) step produce initial population in individuality carry out code modification, namely first the line switching state of the circuit and fault section mains side supply path of not installing switch in individual UVR exposure is set to 1; Again the on off state of fault section border circuit in individual UVR exposure is set to 0, realizes Fault Isolation.

For Fig. 2 power distribution network, illustrate the (4)-1) correction carried out of the individual UVR exposure that obtains in step.As seen from Figure 2, the line set not installing switch is { 3,5,8,10,14,16,19,21,23,25,27,30,33,35,38,39,41,44,45,46,47,50,55,57,59,61,65,67,68}; By the circuit { 52,53,58 of the circuit and fault section mains side supply path of not installing switch in individual UVR exposure, line switching state corresponding to 59,60,61} is set to 1, the circuit of fault section boundary switch the on off state that 54,56} is corresponding is set to 0, obtains revised individual UVR exposure to be:

[1111110111 1111011111 1011111111 1111110111 1001111101 0110101111 1111111101 11]，

In revised individual UVR exposure, the element of the position of the switch of corresponding line 10,14,19,23,35,44,52,53,57,59 and 60 is modified to 1 by 0, and the element of the position of the switch of corresponding line 54 and 56 is modified to 0 by 1, and the element of all the other line switching positions corresponding is constant.

2. individuality is decoded

The (4)-2)-1. walked after, according to the (4)-2)-1. walk the individual UVR exposure after revising to decode, namely according to the track data of revised individual UVR exposure and input, determine that individual UVR exposure breaker in middle state is the track data of 1 correspondence, realize individual decoding.

For Fig. 2 power distribution network, the decoding to individual UVR exposure is described.To the (4)-1)-the individual UVR exposure that 1. obtains in step, by its on off state be 1 circuit propose to obtain corresponding line set be:

{1，2，3，4，5，6，8，9，10，11，12，13，14，16，17，18，19，20，21，23，24，25，26，27，28，29，30，31，32，33，34，35，36，38，39，40，41，44，45，46，47，48，50，52，53，55，57，58，59，60，61，62，63，64，65，66，67，68，70，71，72}

First and last end node and the switchtype information of each circuit can be obtained according to the on off state line set that is 1.

3. the Radiation distribution network that individual different electrical power node connects is formed

The (4)-2)-2. walked after, for obtaining the Radiation Network shelf structure that each power supply node connects, to the (4)-2)-2. walk the Depth Priority Algorithm of decoded individual line data application routine, first from different power supply node with interconnection switch place circuit for border carries out searching for the prompt radiation shape grid structure forming each power supply node and connect; Two end nodes that the circuit of interconnection switch is housed are searched for again respectively successively in prompt radiation shape grid structure.When interconnection two end node is all searched out-of-date, disconnect this interconnection switch and on off state corresponding for this interconnector be set to 0 in this individual UVR exposure to meet the requirement of Radiation Network shelf structure constraints formula (4), otherwise the decoded individual line data of the prompt radiation shape grid structure that search interconnection headend node or endpoint node connect and not searched mistake, form individual final Radiation Network shelf structure.

For Fig. 2 power distribution network, the Radiation distribution network forming individual different electrical power node and connect is described.To the (4)-1)-the line set that 2. obtains in step, first with conventional Depth Priority Algorithm respectively from power supply node 1 and 70 with interconnection switch place circuit for searching on border, form the prompt radiation shape grid structure that two power supply nodes connect, its line set is as follows:

The line set of the prompt radiation shape grid structure that power supply node 1 connects is { 1,2,3,4,5,6,9,10,11,12,13,14,17,18,19,20,21,24,25,26,27,28,29,30};

The line set of the prompt radiation shape grid structure that power supply node 70 connects is { 31,32,33,34,35,36,48,40,41,45,46,52,53,58,59,60,61}

To the interconnection switch circuit in decoded line set, { 70,71,72} searches for its two end node with conventional Depth Priority Algorithm successively respectively in prompt radiation shape grid structure again.First search for interconnection 70, because this circuit headend node is connected to the prompt radiation l network of power supply node 1, and the not searched mistake of non-end node, therefore can search for the circuit of not searched mistake in line set after the decoding.The line set searched accordingly is { 70,65,64,63,62,57,66,67,68}; This line set is merged in the prompt radiation l network of power supply node 1.Then the step of repeat search interconnection 70, searches for interconnection 71 and 72 respectively successively.The all searched mistake of two end nodes due to interconnection 71 and 72, therefore disconnects the switch of interconnection 71 and 72, and the corresponding on off state of interconnection in this individual UVR exposure 71 and 72 is set to 0.The line set of the final Radiation Network network of individuality obtained after search terminates is:

The line set of the prompt radiation shape grid structure that power supply node 1 connects is:

{1，2，3，4，5，6，9，10，11，12，13，14，17，18，19，20，21，24，25，26，27，28，29，30，70，65，64，63，62，57，66，67，68}

The line set of the prompt radiation shape grid structure that power supply node 70 connects is:

{31，32，33，34，35，36，48，40，41，45，46，52，53，58，59，60，61}

4. the trend of the Radiation distribution network that individual different electrical power node connects is calculated

The (4)-2)-3. walked after, first by the (4)-2)-3. walk the line set of the Radiation network configuration of trying to achieve, according to the power distribution network basic parameter of input and the active reactive load of each node during fault restoration of trying to achieve, determine each line impedance of the Radiation network that this period different electrical power node connects and each node active reactive load; Use again traditional before push back power flow algorithm and calculate the node voltage of Radiation distribution network and branch current that each power supply node connects.In calculating, when the trend of the Radiation power distribution network that certain power supply node connects does not restrain, terminate whole Load flow calculation.

For Fig. 2 power distribution network, the trend calculating the Radiation distribution network that individual different electrical power node connects is described.To the (4)-2)-the line set that 3. obtains in step, each line impedance of the Radiation network that power supply node 1 is connected with 70 and each node active reactive load (as shown in table 1) when first determining period t=18; Use again traditional before push back power flow algorithm and calculate the node voltage of Radiation distribution network and branch current that each power supply node connects, result is as shown in table 2.

Each node voltage that table 2 Load flow calculation obtains, each branch current

5. individual target function value is calculated

The (4)-2)-4. walked after, when the (4)-2)-4. calculate the Radiation distribution network trend that different electrical power node connects in step and have when not restraining situation, the target function value of individuality is composed a larger value, namely 10 ⁹.When the trend of the Radiation power distribution network that all power supply nodes connect all restrains, calculate individual total out-of-limit value, concrete formula is:

$F_{t,\mathrm{pen}}=\underset{i-1}{\overset{K_S}{\mathrm{\Σ}}}(\underset{k\∈{\mathrm{\α}}_{t,i}}{\mathrm{\Σ}}\frac{U_{\min }-U_{t,k}}{U_B}+\underset{k\∈{\mathrm{\β}}_{t,i}}{\mathrm{\Σ}}\frac{U_{t,k}-U_{\max }}{U_B}+\underset{b\∈{\mathrm{\γ}}_{t,i}}{\mathrm{\Σ}}\frac{I_{t,b}}{I_{b.\max }})---\left(5\right)$

In formula: F _{t, pen}for the out-of-limit value of t period service restoration power distribution network; K _sfor power supply node sum; α _t,i, β _t,iand γ _t,ibe respectively the node voltage more node set of lower limit, the node voltage more node set of the upper limit and the set of fingers of line current overload of the Radiation power distribution network that t period power supply node i connects; U _maxand U _minbe respectively node voltage bound; U _bfor reference voltage; I _b.maxfor the ampacity of circuit b; I _t,b, U _t,kbe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k.

When the out-of-limit value of individuality is not 0, individual goal functional value is composed a higher value, namely 10 ⁹; When the out-of-limit value of individuality is 0, calculate individual target function value according to formula (1).

For Fig. 2 power distribution network, illustrate and calculate individual target function value.From table 2, node 24-29,55,56, the voltage of 61-67 gets over lower limit, branch road 17,18,24,25 transships, and thus calculating individual total out-of-limit value according to formula (5) is F _{18, pen}=5.0828.Out-of-limit value due to individuality is not 0, is therefore 10 by the target function value assignment of individuality ⁹.

3) Evolution of Population calculates

The (4)-2)-5. walked after, use traditional genetic algorithm to be optimized calculating.First, defining current population is parent population, according to the (4)-2) target function value of individuality of trying to achieve in step calculates fitness value individual in parent population; Secondly, selection operation, interlace operation and mutation operation are carried out to parent population and obtains progeny population; Again, the (4)-2 is adopted) the whole individual goal functional values of step to described progeny population calculate; Then, described progeny population is inserted into parent population and obtains new population; Finally judge that Evolution of Population calculates whether to terminate, when Evolution of Population algebraically does not reach maximum evolutionary generation, evolutionary generation is added 1, and new population is repeated the calculating of this step as current population, until Evolution of Population algebraically reaches in (1) step the maximum evolutionary generation (K inputted _max) till, just obtain coding and the target function value of each individuality in population; Otherwise terminate Evolution of Population to calculate.

For Fig. 2 power distribution network, the evolutionary computation of population is described.Before evolutionary computation, individual UVR exposure is the (4)-2)-1. step in obtain individuality correction coding, wherein position 7,15,22,37,42,43,49,51,54, the element of 56,69 is 0, and the element of all the other positions is 1, individual out-of-limit value be 5.0828 and target function value be 10 ⁹.After being calculated by the fitness value calculation of traditional genetic algorithm, selection, cross and variation and the Evolution of Population that the operation such as heavily to insert, the coding evolution of this individuality is:

[1111111111 1111111111 1111111111 1111111111 1111111111 1110101111 1110101101 01]

In individual UVR exposure after evolution, position 54,56, the element of 64,66,69,71 is 0, and the element of all the other positions is 1.With the (4)-2) the out-of-limit value of individuality that calculates of step be 0 and target function value be 843.18.

4) the rack set of minimum mistake load is determined

The (4)-3) after step completes, from the (4)-3) select the minimum mistake load individuality different with coding the Evolution of Population result of calculation that obtains of step and form the set of candidate's rack.

For Fig. 2 power distribution network, the rack set determining minimum mistake load is described.When period t=18, Evolution of Population calculates that to terminate candidate's rack set of obtaining be that 8 Different Individual being 843.18 by target function value are formed.In these 8 individualities, the element of coding site 7 be respectively 0 and 1 respectively have 4 individualities.In addition, 8 front 61 elements of individuality are all identical with rear 3 elements, unlike the element of position 61-69.The set be made up of the element of 8 individuality 1-61 of putting in place and the element of 70-72 is respectively { 1,1,1,1,1,1,0/1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1} and { 1,0,1}; 8 set formed at the element of position 61-69 are respectively { 0,0,0,1,0,1,1,1}, { 0,1,0,1,0,1,1,1}, { 1,0,0,1,0,1,1,1}, { 1,1,0,1,0,1,1,1}, { 0,0,0,1,0,1,1,0}, { 0,1,0,1,0,1,1,0}, { 1,0,0,1,0,1,1,0}, { 1,1,0,1,0,1,1,0}.

(5) the optimum dynamic power recovery scheme during determining fault restoration

After (4) step completes, in the rack set of the minimum mistake load determined in (4) step, first define the set that dynamic power recovery scheme is made up of the service restoration scheme of each hour during fault restoration.Again according to the hourage that network selection frame number as a child each during fault restoration and fault continue, enumerate the whole dynamic power recovery scheme of combination.Then the switch motion number of times of whole dynamic power recovery scheme is calculated.Minimum first assembled scheme of last selector switch action frequency is as optimum dynamic power recovery scheme.

For Fig. 2 power distribution network, the optimum dynamic power recovery scheme during determining fault restoration is described.Period t=18 during using (4) step to obtain fault restoration, 19,20, candidate's rack set when 21, the set of each candidate's rack comprises 8 respectively, 8,16 racks different with 8 kind.Enumerate the whole dynamic power recovery scheme of combination according to the hourage that these candidate's rack numbers and fault continue, and the grid structure be combined in after the Fault Isolation obtained in (3) step calculates the switch motion number of times of each dynamic power recovery scheme.First of switch motion least number of times assembled scheme is chosen as final scheme.The final scheme switch motion number of times of trying to achieve is 9 times; Period t=18, the set that the individual UVR exposure of 19,20,21 is formed at the element of position 1-61 is { 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,0,1,1,1,1,1}, the line sectionalizing switch of wherein 0 element representation isolated fault disconnection of position 54 and 56; The set formed at the element of position 65-72 is { 1,0,1,1,0,1,0,1}, wherein the element 1 of position 65,67 and 68 represents that circuit does not install switch, the element 0 of position 66 represents the block switch of open-circuit line 66, and the element 0 of position 69 and 71 represents that the interconnection switch of the interconnection 69 and 71 disconnected, the element 1 of position 70 and 72 represent the interconnection switch of closed interconnection 70 and 72; 4 set formed at the element of position 62-64 are respectively { 1,1,0}, { 1,0,1}, { 0,1,1} and { 1,1,1}.The on off state corresponding with the element of position 62-64 and 66 in optimum dynamic power recovery scheme and the load of recovery as shown in table 3,

The on off state of circuit 62-64 and 66 and the load of recovery in the optimum dynamic power recovery scheme of table 3

From Fig. 2 and table 3, when normally running, the switch of interconnection 69-72 is all disconnect, and whole block switch is all closed.When circuit 55 fault, after the block switch isolated fault by open-circuit line 54 and 56, the load of node 55,56,61-67 is in fault outage state.In 4 periods of optimum dynamic power recovery scheme, the switch of interconnection 69,71 and the block switch of circuit 66 all disconnect, and the switch of interconnection 70 and 72 is all closed.Because the block switch of 4 period circuits 66 all disconnects, avoid in service restoration process and form looped network; The switch of interconnection 72 is all closed, and the load of node 65-67 is restored electricity by interconnection 72.In 4 periods, the switch of interconnection 70 is all closed, and the load of node 55,56,61-64 can be restored electricity by interconnection 70.But because the load of period 18 is maximum, can not recover whole loads of these nodes, the block switch of this period open-circuit line 64, only the load of recovery nodes 63,64 is powered.Along with the continuous reduction of load, in the period 19, the block switch of open-circuit line 63, the block switch of closed circuit 64, the load of node 62 is powered and is also restored; In the period 20, the block switch of open-circuit line 62, the block switch of closed circuit 63, the power supply of the load of further recovery nodes 61; In 21 periods, the block switch of closed circuit 62, recovery nodes 55, the load of 56 is powered.

Experimental result

For Fig. 2 electric power system, carry out service restoration effectiveness comparison with the service restoration expert system method (method 2) of method of the present invention (method 1) and existing consideration load variations.When carrying out service restoration by method 2, require that the service restoration scheme formulated can not occur that during whole fault restoration line current is out-of-limit.Due to during the fault restoration of period 18 to period 21, the load of period 18 is maximum, therefore the grid structure determining service restoration scheme of the load of period 18.The service restoration scheme obtained by method 2 is, the switch of closed interconnection 70 and 72, the block switch of open-circuit line 64 and 66; The load point restored electricity has 63,64,65,66,67, and grid structure is constant within the recovery scheme enforcement time.The load of the feeder line 2 in period 19 to the period 21, Fig. 2 reduces the main feeder of feeder line 2 or branch feeder head end current-carrying surplus are increased gradually gradually, add the possibility recovering power failure load further, but method 2 can not adapt to this change.Can take into full account load variations by method 1, be the rational grid structure of selection per hour during fault restoration.From table 3, when during fault restoration, load reduces gradually, the characteristic that rack changes flexibly progressively can recover the power supply of the node 62,61,55 and 56 in Fig. 2 ladderingly.Therefore, the interruption duration of sub-load can be shortened by adjustment grid structure.

Increase when fault occurs in load, and when the main feeder that carries out restoring electricity or branch feeder head end can not be caused to occur that electric current is out-of-limit, voltage out-of-limit appears in end, method 1 is better than method 2 equally.Such as, between the based model for load duration rise period, there is circuit 55 fault in the period 7, the period 10 repairs fault.Because method 2 requires that the service restoration scheme formulated can not occur that during whole fault restoration line current is out-of-limit, therefore the service restoration rack obtained only can recover the power supply of load bus 63-67 during whole fault restoration, load bus 55,56,61,62 will continue to have a power failure to the period 10 from the period 7.The individuality of the optimum dynamic power recovery scheme that employing method 1 obtains is identical with the coding of 8 in the period 7, and the period 9 is identical with the coding of 10.The difference of two codings is only, the former element in position 64 is 1, and the element of the latter in position 64 is 0.The set that element is formed is { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1/0, 1, 0, 1, 1, 0, 1, 0, 1}, wherein the element 0 of position 54 and 56 represents the line switching that isolated fault disconnects, the element 0 of position 69 and 71 represents the interconnection switch disconnected, the element 1 of position 70 and 72 represents closed interconnection switch, switch motion number of times is 4 times.The on off state corresponding with the element of position 64 and 66 in optimum dynamic power recovery scheme and the load of recovery as shown in table 4,

The on off state of circuit 64 and 66 and the load of recovery in the optimum dynamic power recovery scheme of table 4

From table 4, the load that method 1 can recover all power failure nodes 55,56 and 61-67 in the period 7 and 8 is powered.Because load will cause the overload of circuit in the growth of period 9 and 10, therefore the block switch on period 9 open-circuit line 64, excision load bus 55,56,61, the user on 62, avoids occurring that line current transships.At fault early period of origination, can expect in the period 9, the block switch on open-circuit line 64 is caused load bus 55,56,61 according to load curve, the user of 62 has a power failure, and these users that thus can notify early reduce loss of outage.Visible, method 1 ratio method 2 can recover customer power supply as much as possible, and can limit user power utilization before there is circuit overload, by early informing the user outage information thus reducing the unnecessary loss that user brings because having a power failure suddenly.

Table 5 gives the interruption duration of the mistake load bus between two kinds of method age at failures.In table 5, the interruption duration that recovery scheme implements the node 62 of 18 ~ period of period 21 shortened to 1 hour from 4 hours, and the interruption duration of node 61 shortened to 2 hours from 4 hours, and the interruption duration of node 55 and 56 shortened to 3 hours from 4 hours.Implement in 7 ~ period of period 11 at recovery scheme, node 55,56, the interruption duration of 61,62 all shortened to 2 hours from 4 hours.Because the method 1 ratio method 2 more can shorten the interruption duration of certain customers.

Mistake load point interruption duration (h) between table 5 two kinds of method age at failures

Table 6 gives the electricity recovered between two kinds of method age at failures.From table 6, the electricity that the electricity ratio method 2 that method 1 is recovered is recovered is many, shows that method 1 is reducing the electricity sales amount adding Utilities Electric Co. while user has a power failure.

The electricity recovered between table 6 two kinds of method age at failures

Method 1 can also be avoided occurring because the final recovery scheme not meeting security constraint becomes the situation of inefficient solution.Such as, break down at period 18 circuit 25, fault terminated as the period 20.As seen from Figure 2, load point 25,26, therefore 27,28,29 can have a power failure.Power failure load point 25,26,27,28 is made, 29 schemes restored electricity with method 2 switch can formulated on closed circuit 70.After making service restoration scheme, carry out Load flow calculation when verifying its security constraint, find that the minimum voltage of system in the period 18 is lower than the lower voltage limit of system requirements, therefore invalid by the recovery scheme that method 2 is formulated, load point 25,26,27,28,29 all by power failure 4 hours.The optimal power recovery scheme formulated by method 1 is as shown in table 7.In process due to candidate's rack set of each hour minimum mistake load during determining fault restoration, consider the security constraint of branch current and node voltage, therefore the optimum dynamic power recovery scheme finally determined can not become inefficient solution because security constraint disobeys limit.

Optimum dynamic power recovery scheme when table 7 faulty line is 25

Know from above-mentioned experimental result and com-parison and analysis:

(1) the present invention is when considering load variations, can formulate service restoration rack, can utilize load curve more careful and fully for payload hourly during fault restoration.

(2) dynamic power restoration methods of the present invention, can adjust grid structure according to the needs of load variations, scheme has time and space dynamic in good time.

(3) the present invention compared with the conventional method, can reduce customer outage hours further, recovers more electricity.Meanwhile, the present invention can also reduce the loss that brings because having a power failure suddenly of user by prior notice customer outage hours.

(4) the present invention compared with the conventional method, in the process of each hour minimum mistake load candidate rack set during determining fault restoration, consider the security constraint of branch current and node voltage, the dynamic power recovery scheme therefore determined can be avoided occurring that final recovery scheme becomes the situation of inefficient solution owing to not meeting security constraint.

Claims (1)

1. based on a power distribution network dynamic power restoration method for load curve, utilize computer, calculated by program, effectively formulate the optimum dynamic power recovery scheme during fault restoration, it is characterized in that the concrete steps of described method are as follows:

(1) basic parameter is inputted

First, basic parameter needed for input calculates, comprise the basic parameter of power distribution network and the parameter of traditional genetic algorithm, the basic parameter of power distribution network comprises circuit sum (M), circuit number, circuit first and last end node numbering, line resistance (R), line reactance (X), line energizing flow amount, line switching type, namely do not install switch, installing block switch, installing interconnection switch, circuit place feeder line is numbered; Node total number (N), power supply node sum (K _s), power supply node numbering, node burden with power (P _l), node load or burden without work (Q _l), the meritorious daily load rate curve of node load type, various load type; The node voltage upper limit (U _max), node voltage lower limit (U _min); Faulty line numbering, time of failure (t ₁), fault end time (t ₂); Reference voltage (U _b), reference power (S _b), the parameter of traditional genetic algorithm comprises population scale (P _p), maximum evolutionary generation (K _max), generation gap (P _g), crossing-over rate (p _c), aberration rate (p _m);

(2) calculate meritorious, the load or burden without work of each node each hour and determine grid structure when normally running

After (1) step completes, first according to the burden with power (P of each node of (1) step input _l), load or burden without work (Q _l) and the meritorious daily load curve of various load type, calculate the burden with power of each hour of each node and load or burden without work according to the principle that power factor is constant, then according to the position of the interconnection switch installed in power distribution network, determine grid structure when normally running;

(3) grid structure after the supply path of fault section and mains side and Fault Isolation is determined

After (2) step completes, in grid structure when the normal operation that (2) step is determined, calculate for ease of optimizing and reduce computing time, first failure definition section is centered by fault point and the line set being border with the switch nearest from fault point; The supply path of failure definition section mains side is the line set between power supply node to fault section, then uses conventional Depth Priority Algorithm, whether installs switch determination fault section according to faulty line head end; The supply path of fault section mains side is determined from the power supply node of faulty line place feeder line, the boundary switch isolated fault of the grid structure fault section when power distribution network finally determined according to (2) step normally runs, determines the grid structure after Fault Isolation;

(4) candidate's grid structure set of each hour minimum mistake load during using traditional genetic algorithm determination fault restoration

After (3) step completes, in the grid structure after the Fault Isolation that (3) step is determined, for obtaining the grid structure of the power distribution network each hour minimum mistake load during fault restoration after Fault Isolation, to determine candidate's grid structure set of dynamic power recovery scheme, during using the fault restoration obtained based on load curve, each hour load solves the optimal models of corresponding power distribution network service restoration, the target function of this model is minimum mistake load, constraints is branch current, the constraints such as node voltage and Radiation Network shelf structure, concrete formula is as follows:

$\begin{array}{cc}\min & P_{t,c}\left(X_{t,r}\right)=\underset{n\∈{\mathrm{\Ω}}_N}{\mathrm{\Σ}}{P}_{t,n}-\underset{k\∈{\mathrm{\Ω}}_K}{\mathrm{\Σ}}{P}_{t,k}---\left(1\right)\end{array}$

s.t.I _t,b≤I _b.max(2)

U _min≤U _t,k≤U _max(3)

g _t,i∈G (4)

In formula: t is the period, and the time of one day is divided into 24 hours, it it within every 1 hour, is 1 period; P _t,cfor the mistake load of period t; X _t,rfor the line switching state vector of the service restoration power distribution network of period t, X _t,r=[W _{t, r1}..., W _{t, rm}... W _{t, rM}]; M is circuit sum; W _{t, rm}for X _t,rthe on off state of m article of circuit, be taken as 1 when the switch that circuit does not install switch or installing is in closure state, be taken as 0 when the switch installed is in off-state; Ω _nand P _t,nbe respectively the node set of the power distribution network before service restoration and the node n burden with power at period t; Ω _kand P _t,kbe respectively the node set of service restoration power distribution network and the node k burden with power at period t; N and K is respectively the node total number of power distribution network before and after service restoration; I _t,b, U _t,kand g _t,ibe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k and grid structure; I _b.maxfor the ampacity of circuit b; U _maxand U _minbe respectively the bound of node voltage; G is the set of Radiation Network shelf structure;

During solving fault restoration by traditional genetic algorithm, the concrete steps of the grid structure of the minimum mistake load that each hour load is corresponding are as follows:

1) initial population is formed

After (3) step completes, according to the population scale (P of (1) step input _p), with { 0,1} binary code, the individuality that the random number produced is population scale is encoded, form initial population, in initial population, the code length of each individuality is circuit sum, the circuit head end on off state of each coding stands power distribution network, when being encoded to 0, represent that this line switching is in off-state; When being encoded to 1, represent that this line switching is in closure state;

2) target function value individual in population is calculated

The (4)-1) after step completes, for evaluating the (4)-1) quality of individual performance in the initial population that formed of step, the target function value of individuality each in this population is calculated, individual target function value is less illustrates that individual performance is more excellent, as follows to the concrete steps of individual calculating target function value each in described population:

1. individual UVR exposure is revised

The (4)-1) after step completes, to the (4)-1) step produce initial population in individuality carry out code modification, namely first the line switching state of the circuit and fault section mains side supply path of not installing switch in individual UVR exposure is set to 1; Again the on off state of fault section border circuit in individual UVR exposure is set to 0, realizes Fault Isolation;

2. individuality is decoded

The (4)-2)-1. walked after, according to the (4)-2)-1. walk the individual UVR exposure after revising to decode, namely according to the track data of revised individual UVR exposure and input, determine that individual UVR exposure breaker in middle state is the track data of 1 correspondence, realize individual decoding;

3. the Radiation distribution network that individual different electrical power node connects is formed

The (4)-2)-2. walked after, for obtaining the Radiation Network shelf structure that each power supply node connects, to the (4)-2)-2. walk the Depth Priority Algorithm of decoded individual line data application routine, first from different power supply node with interconnection switch place circuit for border carries out searching for the prompt radiation shape grid structure forming each power supply node and connect; Two end nodes that the circuit of interconnection switch is housed are searched for again respectively successively in prompt radiation shape grid structure, when interconnection two end node is all searched out-of-date, disconnect this interconnection switch and on off state corresponding for this interconnector be set to 0 in this individual UVR exposure to meet the requirement of Radiation Network shelf structure constraints formula (4), otherwise the decoded individual line data of the prompt radiation shape grid structure that search interconnection headend node or endpoint node connect and not searched mistake, form individual final Radiation Network shelf structure;

4. the trend of the Radiation distribution network that individual different electrical power node connects is calculated

The (4)-2)-3. walked after, first by the (4)-2)-3. walk the line set of the Radiation network configuration of trying to achieve, according to the power distribution network basic parameter of input and the active reactive load of each node during fault restoration of trying to achieve, determine each line impedance of the Radiation network that this period different electrical power node connects and each node active reactive load; Use again traditional before push back power flow algorithm and calculate the node voltage of Radiation distribution network and branch current that each power supply node connects, in calculating, when the trend of the Radiation power distribution network that certain power supply node connects does not restrain, terminate whole Load flow calculation;

5. individual target function value is calculated

The (4)-2)-4. walked after, when the (4)-2)-4. calculate the Radiation distribution network trend that different electrical power node connects in step and have when not restraining situation, the target function value of individuality is composed a larger value, namely 10 ⁹, when the trend of the Radiation power distribution network that all power supply nodes connect all restrains, calculate individual total out-of-limit value, concrete formula is:

$F_{t,\mathrm{pen}}=\underset{i-1}{\overset{K_S}{\mathrm{\Σ}}}(\underset{k\∈{\mathrm{\α}}_{t,i}}{\mathrm{\Σ}}\frac{U_{\min }-U_{t,k}}{U_B}+\underset{k\∈{\mathrm{\β}}_{t,i}}{\mathrm{\Σ}}\frac{U_{t,k}-U_{\max }}{U_B}+\underset{b\∈{\mathrm{\γ}}_{t,i}}{\mathrm{\Σ}}\frac{I_{t,b}}{I_{b.\max }})---\left(5\right)$

In formula: F _{t, pen}for the out-of-limit value of t period service restoration power distribution network; K _sfor power supply node sum; α _t,i, β _t,iand γ _t,ibe respectively the node voltage more node set of lower limit, the node voltage more node set of the upper limit and the set of fingers of line current overload of the Radiation power distribution network that t period power supply node i connects; U _maxand U _minbe respectively node voltage bound; U _bfor reference voltage; I _b.maxfor the ampacity of circuit b; I _t,b, U _t,kbe respectively in the service restoration power distribution network of period t and flow through the electric current of circuit b, the voltage of node k;

When the out-of-limit value of individuality is not 0, individual goal functional value is composed a higher value, namely 10 ⁹; When the out-of-limit value of individuality is 0, calculate individual target function value according to formula (1);

3) Evolution of Population calculates

The (4)-2)-5. walked after, traditional genetic algorithm is used to be optimized calculating, first, defining current population is parent population, according to the (4)-2) target function value of individuality of trying to achieve in step calculates fitness value individual in parent population; Secondly, selection operation, interlace operation and mutation operation are carried out to parent population and obtains progeny population; Again, the (4)-2 is adopted) the whole individual goal functional values of step to described progeny population calculate; Then, described progeny population is inserted into parent population and obtains new population; Finally judge that Evolution of Population calculates whether to terminate, when Evolution of Population algebraically does not reach maximum evolutionary generation, evolutionary generation is added 1, and new population is repeated the calculating of this step as current population, until Evolution of Population algebraically reaches in (1) step the maximum evolutionary generation (K inputted _max) till, just obtain coding and the target function value of each individuality in population; Otherwise terminate Evolution of Population to calculate;

4) the rack set of minimum mistake load is determined

The (4)-3) after step completes, from the (4)-3) select the minimum mistake load individuality different with coding the Evolution of Population result of calculation that obtains of step and form the set of candidate's rack;

(5) the optimum dynamic power recovery scheme during determining fault restoration

After (4) step completes, in the rack set of the minimum mistake load determined in (4) step, first define the set that dynamic power recovery scheme is made up of the service restoration scheme of each hour during fault restoration, again according to the hourage that network selection frame number as a child each during fault restoration and fault continue, enumerate the whole dynamic power recovery scheme of combination, then calculate the switch motion number of times of whole dynamic power recovery scheme, minimum first assembled scheme of last selector switch action frequency is as optimum dynamic power recovery scheme.