CN106229964A - A kind of based on the electrical power distribution network fault location method improving binary particle swarm algorithm - Google Patents

Abstract

The present invention proposes a kind of electrical power distribution network fault location method based on improvement binary particle swarm algorithm, improves traditional binary particle cluster algorithm, and is applied to the location of distribution network failure.Comprise the steps: first to determine parameters such as including population scale, maximum iteration time；Secondly form the expectation function of switch according to switch fault information, and construct the fitness function of distribution network failure location；Initialize population, set particle position, and set particle rapidity as 0；Calculate the fitness value of particle according to fitness function and set initial global extremum；Update individual extreme value and initial global extremum；S5: update speed and the position of population；When reaching maximum iteration time, stop calculating, output population global optimum position, i.e. the physical fault state of target power distribution network each feeder line section.The method can overcome the premature problem that traditional method exists, and further increases convergence and stability simultaneously.

Description

A kind of based on the electrical power distribution network fault location method improving binary particle swarm algorithm

Technical field

The present invention relates to a kind of electrical power distribution network fault location method, be specifically related to a kind of based on binary particle swarm algorithm join Electric network fault localization method.

Background technology

Distribution network failure location is one of key content realizing power distribution automation, and its cardinal principle is by each remote feeder Terminal (is called for short FTU), and the fault message reported carries out comprehensive analysis thus judges fault section, extensive for the power supply after fault Condition is provided again.Therefore distribution network failure location to shorten power off time, reduce power failure range and improve power distribution network power reliably The aspects such as property are significant.

At present, distribution network failure method can be divided into matrix method and artificial intelligence method, and wherein matrix method is according to network structure The computing of matrix and fault message matrix forms work out problems, and the method needs FTU to report fault message accurately, therefore Fault-tolerance is poor；Intelligent algorithm can allow a small amount of that distortion information exists, and fault-tolerance is preferable, and diagnosis speed is very fast, It represents algorithm is genetic algorithm, binary particle swarm algorithm etc.；But genetic algorithm calculates operation complexity, computationally intensive, tradition Binary particle swarm algorithm there is the problems such as Premature Convergence.

Summary of the invention

For overcoming above-mentioned algorithm to there is the problem of Premature Convergence, and improving convergence further, the present invention uses Electrical power distribution network fault location method based on improvement binary particle swarm algorithm, the method can overcome the precocity that traditional algorithm exists Convergence problem, further increases convergence and stability simultaneously.

A kind of electrical power distribution network fault location method based on improvement binary particle swarm algorithm, comprises the following steps:

S1: determine population parameter: include population scale M, maximum iteration time t_max, Studying factors c₁And c₂, inertia Weights omega_maxAnd ω_min, determining nodes and the feeder line sector number of target power distribution network, particle dimension is equal to feeder line section sum, i.e. Determine the dimension of particle；

S2: structure fitness function: form the expectation function of switch according to switch fault information, and construct distribution network failure The fitness function of location；

S3: initialization population: population particle position X_i=(x_i1, x_i2..., x_in), set particle rapidity V_i=(v_i1, v_i2..., v_in) it is 0；Calculate the fitness value of particle according to fitness function, be recorded as initial individuals extreme value P_{Best, i}, Simultaneously by all P_{Best, i}One of intermediate value minimum is assigned to initial global extremum g_best；

S4: update P_{Best, i}And g_best: the fitness value and before of each particle is first calculated according to fitness function Individual ratio of extreme values relatively, updates P_{Best, i}, simultaneously by entirety P_{Best, i}And g_bestCompare, thus update g_best；

S5: update speed and the position of population；

S6: if algorithm reaches maximum iteration time, then stop calculating, output population global optimum position, i.e. target The physical fault state of power distribution network each feeder line section；Otherwise return step S4, again iterative computation.

Further, in step S2, the expectation function value of power distribution network switch is determined by the feeder line section at its rear portion, works as network When middle any zone breaks down, the switch near mains side all will experience fault current；The fitness of distribution network failure location Function is:

$F_{it}\left(S_B\right)=\underset{j=1}{\overset{N}{\Σ}}|I_j-I_j^{*}\left(S_B\right)|+\mathrm{\μ}\underset{j=1}{\overset{N}{\Σ}}\left|S_B(j,i)\right|$

In formula: N is the total number of network breaker in middle；I_jRepresenting the fault current information of jth switch FTU storage, value is 1 table Showing that this switch experienced by fault current, value is 0 expression non-faulty current；Represent the expectation state of switch, it is desirable to state It is when 1, to represent that switch is normal, it is desirable to state is to represent switch fault when 0；S_BRepresent the state of equipment, its value 1 in power distribution network Or 0, represent that equipment fault is with normal respectively；Represent faulty equipment sum in network；μ is fault diagnosis weight Coefficient, between value [0,1]；

Further, in step S5, preset a mutation probability, in the every one-dimensional tax one [0,1] of each particle Between random number, if this random number less than mutation probability, then carries out xor operation to the most one-dimensional of particle；If random number is more than Equal to mutation probability, then particle keeps constant；Wherein, particle variations probability calculation formula is as follows:

$v^{\′}=|2*\[\left(\frac{1}{1-e^{-v}}\right)-0.5\]|$

In formula: v ' is mutation probability, v is the present speed of particle.

Further, in step S5, the speed of particle more new formula is as follows:

$V_i^{k+1}={\mathrm{\ωv}}_i^k+c_1{r}_1(P_i^k-X_i^k)+c_2{r}_2(P_g^k-X_i^k)$

In formula,WithSpeed after updating when k+1 iteration for particle i and position, ω is inertial factor, c₁And c₂It is Studying factors,The position that when searching for for kth time, the history optimal solution of particle i is corresponding,During for kth time search The position that globally optimal solution in all particles is corresponding, r₁And r₂It it is the random number between 0 to 1；

In step S5, the location updating formula of particle is as follows:

$\left\{\begin{array}{cc}{x}_{in}^{k+1}=1& r_{in}^{k+1}<sigmoid\left(v_{in}^{k+1}\right)\\ x_{in}^{k+1}=0& r_{in}^{k+1}\&GreaterEqual;sigmoid\left(v_{in}^{k+1}\right)\end{array}\right.$

In formula,It it is the random number between [0,1]；

Further,The mathematic(al) representation of function is as follows:

$sigmoid\left(v_{in}^{k+1}\right)=\left\{\begin{array}{cc}0.98& v_{in}^{k+1}>L\\ \frac{1}{1+e^{-v_{in}^{k+1}}}& -L\&le;v_{in}^{k+1}\&le;L\\ -0.98& v_{in}^{k+1}<-L\end{array}\right.$

In formula,Speed for particle+1 iteration of kth；L-value is preferably 4.

Further, the speed more new formula of particle is improved by introducing contraction factor:

Wherein:

C=c₁+c₂

In formula:For contraction factor, preferred value is 4.1.

Further, inertia weight coefficient formulas is as follows:

$\mathrm{\&omega;}={\mathrm{\&omega;}}_{\max }-t\frac{{\mathrm{\&omega;}}_{\max }-{\mathrm{\&omega;}}_{\min }}{t_{\max }}$

In formula, ω_maxAnd ω_minBeing respectively the maxima and minima of ω, t is current iterations, t_maxChange for maximum Generation number；Wherein, ω_maxPreferred value is 0.9, ω_minPreferred value is 0.4.

Beneficial effect:

Traditional binary particle cluster algorithm is improved by the present invention, and is applied to the location of distribution network failure；By drawing Enter contraction factor to reset particle rapidity with new formula, take into account optimizing ability and the convergence of population；According to inertia The characteristic of weight coefficient, uses the inertia weight coefficient dynamically adjusted, and focuses on the ability of searching optimum event of particle when algorithm is initial Use bigger inertia weight value, focus on convergence during algorithm ending therefore use less inertia weight value, improve grain with this The optimizing ability of subgroup and convergence capabilities；For overcoming the premature convergence problem of traditional algorithm, also introduce mutagenic factor, add particle The multiformity of population, the ability of searching optimum of algorithm increases, thus avoids being absorbed in local extremum.

Accompanying drawing explanation

The simple distribution network of Fig. 1

Fig. 2 flow chart based on the distribution network failure location improving binary particle swarm algorithm

Fig. 3 single supply 12 node radial distribution networks

Detailed description of the invention

For open loop operation distribution network failure position, each switching node of power distribution network is equipped with terminal, can more than Report fault message, 0 represents that non-faulty current flows through, and 1 represents that faulty electric current flows through.The position of particle represents feeder line in power distribution network The state of section, the dimension of particle represents distribution feeder section sum.Feeder line section normal condition represents with 0, malfunction Represent with 1.The sneak condition of the N-dimensional positional representation power distribution network N section feeder line section of each particle, by solving N-dimensional population Optimum results, it is possible to obtain the state of power distribution network N section feeder line section, thus judge fault section, it is achieved the event of power distribution network Barrier location.

Concrete, the invention provides a kind of electrical power distribution network fault location method based on improvement binary particle swarm algorithm, Shown in the flow chart of accompanying drawing 2, comprise the following steps:

S1: determine population parameter: include population scale M, maximum iteration time t_max, Studying factors c₁And c₂, inertia Weights omega_maxAnd ω_min, determining nodes and the feeder line sector number of target power distribution network, particle dimension is equal to feeder line section sum, i.e. Determine the dimension of particle；

S2: structure fitness function: form the expectation function of switch according to switch fault information, and construct distribution network failure The fitness function of location；

S3: initialization population: population particle position X_i=(x_i1, x_i2..., x_in), set particle rapidity V_i=(v_i1, v_i2..., v_in) it is 0；Calculate the fitness value of particle according to fitness function, be recorded as initial individuals extreme value P_{Best, i}, Simultaneously by all P_{Bsst, i}One of intermediate value minimum is assigned to initial global extremum g_best；According to particle fitness function, calculate particle Fitness value, fault message deviation should be uploaded with actual based on waiting to ask information corresponding under a feeder line section virtual condition Minimum principle, particle fitness value is the least, and corresponding solution is the most excellent.

S4: update P_{Best, i}And g_best: the fitness value and before of each particle is first calculated according to fitness function Individual ratio of extreme values relatively, updates P_{Best, i}, simultaneously by entirety P_{Best, i}And g_bestCompare, thus update g_best；

S5: update speed and the position of population；

S6: if algorithm reaches maximum iteration time, then stop calculating, output population global optimum position, i.e. target The physical fault state of power distribution network each feeder line section；Otherwise return step S4, again iterative computation.

In above-mentioned steps:

In step S, the expectation function value of 2 power distribution network switches is determined by the feeder line section at its rear portion, when any zone in network When breaking down, the switch near mains side all will experience fault current；Distribution network as shown in Figure 1, its power distribution network switchs Expectation function value computing formula as follows:

$\left\{\begin{array}{c}{I}^{*}\left({\mathrm{CB}}_1\right)=L_1\left|\right|L_2\left|\right|L_3\left|\right|L_4\left|\right|L_5\\ I^{*}\left(S_1\right)=L_2\left|\right|L_3\left|\right|L_4\left|\right|L_5\\ I^{*}\left(S_2\right)=L_3\left|\right|L_5\\ I^{*}\left(S_3\right)=L_4\\ I^{*}\left(S_4\right)=L_5\end{array}\right.$

In formula: CB₁For lead-in circuit breaker；S₁, S₂, S₃, S₄For block switch；L₁, L₂, L₃, L₄, L₅For feeder line sector number.

In step S2, the fitness function of distribution network failure location is:

$F_{it}\left(S_B\right)=\underset{j=1}{\overset{N}{\&Sigma;}}|I_j-I_j^{*}\left(S_B\right)|+\mathrm{\&mu;}\underset{j=1}{\overset{N}{\&Sigma;}}\left|S_B(j,i)\right|$

In formula: N is the total number of network breaker in middle；I_jRepresenting the fault current information of jth switch FTU storage, value is 1 table Showing that this switch experienced by fault current, value is 0 expression non-faulty current；Represent the expectation state of switch, it is desirable to state It is when 1, to represent that switch is normal, it is desirable to state is to represent switch fault when 0；S_BRepresent the state of equipment, its value 1 in power distribution network Or 0, represent that equipment fault is with normal respectively；Represent faulty equipment sum in network；μ is fault diagnosis weight Coefficient, between value [0,1]；

In step S5, the speed of particle more new formula is as follows:

$V_i^{k+1}={\mathrm{\&omega;v}}_i^k+c_1{r}_1(P_i^k-X_i^k)+c_2{r}_2(P_g^k-X_i^k)$

In formula,WithSpeed after updating when k+1 iteration for particle i and position, ω is inertial factor, c₁And c₂It is Studying factors, is usually taken to be 2,The position that when searching for for kth time, the history optimal solution of particle i is corresponding,It is The position that during k search, the globally optimal solution in all particles is corresponding, r₁And r₂It it is the random number between 0 to 1；

The location updating formula of particle is as follows:

$\left\{\begin{array}{cc}{x}_{in}^{k+1}=1& r_{in}^{k+1}<sigmoid\left(v_{in}^{k+1}\right)\\ x_{in}^{k+1}=0& r_{in}^{k+1}\&GreaterEqual;sigmoid\left(v_{in}^{k+1}\right)\end{array}\right.$

In formula,It it is the random number between [0,1]；

The mathematic(al) representation of function is as follows:

$sigmoid\left(v_{in}^{k+1}\right)=\left\{\begin{array}{cc}0.98& v_{in}^{k+1}>L\\ \frac{1}{1+e^{-v_{in}^{k+1}}}& -L\&le;v_{in}^{k+1}\&le;L\\ -0.98& v_{in}^{k+1}<-L\end{array}\right.$

In formula,Speed for particle+1 iteration of kth；Above formula is the location updating formula of particle, because power distribution network Fault location is Discretization, therefore by particle position more new formula discretization, in formulaIt is random between [0,1] Number.For preventingFunction is saturated, generally the speed of particle is limited between [-4,4], i.e. L is preferably 4.

Further, the speed more new formula of particle is optimized by introducing contraction factor:

Wherein:

C=c₁+c₂

In formula:For contraction factor, generally for ensureing that algorithm is smoothed out, takeIt is 4.1.

Wherein, inertia weight coefficient formulas is as follows:

$\mathrm{\&omega;}={\mathrm{\&omega;}}_{\max }-t\frac{{\mathrm{\&omega;}}_{\max }-{\mathrm{\&omega;}}_{\min }}{t_{\max }}$

In formula, ω_maxAnd ω_minBeing respectively the maxima and minima of ω, t is current iterations, t_maxChange for maximum Generation number；Preferably, ω_maxIt is 0.9, ω_minIt is 0.4.

Further, for the precocious phenomenon in traditional binary particle cluster algorithm, it is contemplated that particle is in current optimum position Put down, in fact it could happen that better position, mutagenic factor is introduced in innovatory algorithm.The most in step s 5, a change is preset Different probability, the random number between the every one-dimensional tax one [0,1] of each particle, if this random number is less than mutation probability, the most right The the most one-dimensional of particle carries out xor operation；If random number is more than or equal to mutation probability, then particle keeps constant；Wherein, particle becomes Different probability calculation formula is as follows:

$v^{\&prime;}=|2*\&lsqb;\left(\frac{1}{1-e^{-v}}\right)-0.5\&rsqb;|$

In formula: v ' is mutation probability, v is the present speed of particle.

As it is shown on figure 3, in single supply 12 node radial distribution networks, join based on improvement binary particle swarm algorithm Electric network fault localization method carries out Example Verification, and result shows that innovatory algorithm just has preferable convergence and stability, and overcomes The premature problem of traditional algorithm.

Design parameter is provided that Studying factors c₁=c₂=2.05, inertia weight ω_max=0.9, ω_min=0.4, shrink The factorPopulation scale is set to 30, and particle dimension is set to 12, and maximum iteration time is 120 times.

The present embodiment sets the fault type of power distribution network as phase fault, the switching node fault that input FTU reports Information, the faulty electric current of switching node flows through, is 1, and non-faulty current flows through, and is 0.Output result is 1, represents its correspondence Feeder line section fault, output result is 0, represents the feeder line section fault-free of its correspondence.To power distribution network Single Point of Faliure, multiple spot event Hindering, calculate with or without failure conditions such as distortion information, program is run 50 times continuously, and result of calculation is as shown in table 1:

Table 1 result of calculation

During Single Point of Faliure, as feeder line section 6 occurs phase fault, FTU to report as [111001000000], display joint Point switch 1,2,3,6 experienced by fault current, undistorted information, and improved BPSO algorithm calculates, and output result is [000001000000], show feeder line section 6 fault, accurately realize fault section location.In the case of same fault, when on FTU The information of report exists when distorting on a small quantity [100001000000], i.e. switchs 2,3 wrong reports, and finally output is still that [000001000000], show feeder line section 6 fault, be still accurately positioned fault section.

During multipoint fault, as feeder line section 4,8,12 occurs phase fault, FTU reports as [111101110011], Switch 1,2,3,4,6,7,8,11,12 experienced by fault current, undistorted information, and improved BPSO algorithm calculates, output For [000100010001], show feeder line section 4,8,12 fault, accurately achieve fault section location.Same fault situation Under, when FTU reporting information exists distortion information [100101110011], i.e. switch 2,3 wrong reports, be finally output as [000100010001], show feeder line section 4,8,12 fault, still realize fault section and be accurately positioned.

Therefore, improving binary particle swarm algorithm can be under a small amount of FTU distortion information, it is achieved fault fast and accurately Location, can meet the requirement of distribution network failure location.

The preferred embodiment of the present invention is only intended to help to illustrate the present invention, and it is all of that preferred embodiment does not has detailed descriptionthe Details, is also not intended to the detailed description of the invention that this invention is only described.Obviously, according to the content of this specification, can make a lot Modifications and variations.These embodiments are chosen and specifically described to this specification, is to preferably explain the principle of the present invention and reality Border is applied, so that skilled artisan can utilize the present invention well.

Claims (10)

1. an electrical power distribution network fault location method based on improvement binary particle swarm algorithm, it is characterised in that include following step Rapid:

S1: determine population parameter: include population scale M, maximum iteration time t_max, Studying factors c₁And c₂, inertia weight ω_maxAnd ω_min, the nodes of target power distribution network and feeder line sector number, particle dimension is equal to feeder line section sum；

S2: structure fitness function: form the expectation function of switch according to switch fault information, and construct distribution network failure location Fitness function；

S3: initialization population: population particle position X_i=(x_i1, x_i2..., x_in), set particle rapidity V_i=(v_i1, v_i2..., v_in) it is 0；Calculate the fitness value of particle according to fitness function, be recorded as initial individuals extreme value P_{Best, i}, Simultaneously by all P_{Best, i}One of intermediate value minimum is assigned to initial global extremum g_best；

S4: update P_{Best, i}And g_best: fitness value and the individuality before of each particle is first calculated according to fitness function Ratio of extreme values relatively, updates P_{Best, i}, simultaneously by entirety P_{Best, i}And g_bestCompare, thus update g_best；

S5: update speed and the position of population；

S6: if reaching maximum iteration time, then stop calculating, and output population global optimum position, i.e. target power distribution network is each The physical fault state of feeder line section；Otherwise return step S4, again iterative computation.

Electrical power distribution network fault location method the most according to claim 1, it is characterised in that: the phase of power distribution network switch in step S2 Hoping that functional value is determined by the feeder line section at its rear portion, when in network, any zone breaks down, the switch of close mains side is all Fault current will be experienced；The fitness function of distribution network failure location is:

In formula: N is the total number of network breaker in middle；I_jRepresenting the fault current information of jth switch FTU storage, value is that 1 expression should Switch experienced by fault current, and value is 0 expression non-faulty current；Represent the expectation state of switch, it is desirable to when state is 1 Represent that switch is normal, it is desirable to state is to represent switch fault when 0；S_BRepresent the state of equipment in power distribution network, its value 1 or 0, divide Biao Shi equipment fault and normal；Represent faulty equipment sum in network；μ is fault diagnosis weight coefficient, Between value [0,1].

3. the electrical power distribution network fault location method described in claim 1, it is characterised in that: in step S5, preset a variation Probability, the random number between the every one-dimensional tax one [0,1] of each particle, if this random number is less than mutation probability, then to grain The the most one-dimensional of son carries out xor operation；If random number is more than or equal to mutation probability, then particle keeps constant；Wherein, particle variations Probability calculation formula is as follows:

In formula: v ' is mutation probability, v is the present speed of particle.

Electrical power distribution network fault location method the most according to claim 1, it is characterised in that: in step S5, the speed of particle is more New formula is as follows:

In formula,WithSpeed after updating when k+1 iteration for particle i and position, ω is inertial factor, c₁And c₂ It is Studying factors,The position that when searching for for kth time, the history optimal solution of particle i is corresponding,For all grains during kth time search The position that globally optimal solution in son is corresponding, r₁And r₂It it is the random number between 0 to 1；

The location updating formula of particle is as follows:

In formula,It it is the random number between [0,1].

Electrical power distribution network fault location method the most according to claim 4, it is characterised in that:

The mathematic(al) representation of function is as follows:

In formula,Speed for particle+1 iteration of kth.

Electrical power distribution network fault location method the most according to claim 5, it is characterised in that: L-value is 4.

Electrical power distribution network fault location method the most according to claim 4, it is characterised in that: enter one by introducing contraction factor The speed more new formula of step improvement particle:

Wherein:

C=c₁+c₂

In formula:For contraction factor.

Electrical power distribution network fault location method the most according to claim 4, it is characterised in that: contraction factorValue is 4.1.

9. according to the electrical power distribution network fault location method described in claim 4 to 8 any one, it is characterised in that: inertia weight system Number computing formula is as follows:

In formula, ω_maxAnd ω_minBeing respectively the maxima and minima of ω, t is current iterations, t_maxFor greatest iteration time Number.

Electrical power distribution network fault location method the most according to claim 9, it is characterised in that: inertia weight ω_maxIt is 0.9, ω_minIt is 0.4.