CN108037414B - A fault location method for distribution network based on hierarchical model and intelligent verification algorithm - Google Patents

Abstract

The invention relates to a distribution network fault location method based on a hierarchical model and an intelligent verification algorithm. The two-port equivalent of each branch of the distribution network fault location model greatly reduces the dimension of the switching function of the location model and greatly simplifies the structure; the location algorithm is divided into two layers: fault port location and fault segment location. The operation dimension of the fault location algorithm is reduced; the absolute reliability of the fault location location is used to feedback and verify the location result of the fault port, which effectively makes up for the defect that the fault identification result of the location algorithm is unstable. The method of the invention has good fault tolerance and stability, and greatly simplifies the fault identification model and improves the efficiency of fault location. The method is especially suitable for fault location of high-penetration and large-scale distribution networks.

Description

A fault location method for distribution network based on hierarchical model and intelligent verification algorithm

technical field

The invention relates to the field of distribution network fault location, isolation and power supply recovery, in particular to a distribution network fault location method based on a hierarchical model and an intelligent verification algorithm.

Background technique

With a large number of cleaner and more efficient distributed power sources (DG) connected to the power grid and the increasing power load, the structure and trend of the power distribution system are becoming larger and more complex. The traditional single-source radiation network fault location method becomes no longer applicable. Aiming at the problem of fault location in distribution network with distributed power generation, a direct location algorithm based on matrix theory and an indirect location method using artificial intelligence technology have been formed. The segment positioning matrix algorithm has the advantages of simple modeling, efficient and accurate positioning, etc., but it also has the defects of low fault tolerance and low generality. The fault location method based on artificial intelligence technology, according to the state approximation idea and the principle of the minimum set of fault diagnosis, adopts the optimization theory to model the fault segment identification model, which has the advantages of strong versatility and high fault tolerance. Among them, the swarm intelligence algorithm is easy to deal with discrete variables in the process of fault identification, and theoretically obtains the global optimal decision, which has become a hot research topic in this field.

There are three main types of existing swarm intelligence algorithms for fault location in distribution networks with distributed power sources. The first type is to directly use a single-layer model and a single intelligent algorithm or its improved algorithm to locate the fault of the distribution network according to the structural characteristics of the distribution network. The construction of the switching function of this type of method is relatively complicated, and it has great limitations in improving the fault tolerance, efficiency and stability of fault location. The second type uses the parallel evolution of multiple populations and information interaction strategies for fault location, which improves the fault tolerance and stability of the intelligent algorithm to a certain level, but still uses a single-layer identification model. In the high-penetration distribution network, the construction of the switching function based on logical OR is still quite complicated, and the fault identification efficiency is low; the algorithm used is still not free from the limitation of unstable fault location results. The third type introduces the idea of zoning into fault location, and reduces the dimension of intelligent algorithms through regional division, which improves the search efficiency of fault location. However, in the face of large-scale, high-penetration distribution networks, the search dimension is still Large, the construction of switching function based on logical OR is still quite complicated, and the identification result of intelligent algorithm still has the problem of instability.

To sum up, the artificial intelligence method of indirect modeling of fault section location in distribution network has achieved fruitful results in theory, but this type of method still has the following problems: 1) Use logical relationships to describe the matching between fault sections and equipment The correlation characteristics make the construction of the fault location model relatively complicated, especially in high-penetration and large-scale distribution networks, the location modeling is extremely complex; 2) The single-layer swarm intelligence algorithm used is used in the location of fault sections in large-scale distribution networks. The computational dimension is huge and the efficiency is too low; 3) There is an essential defect that the identification results are unstable by simply relying on the intelligent algorithm to locate the fault.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a distribution network fault location method based on a hierarchical model and an intelligent verification algorithm. Combined with the structural characteristics of the distribution network, on the basis of analyzing the logic law of the switching function, a method based on multi-branch nodes is proposed. For the boundary and the method of port equivalence for each branch, the distribution network is divided into multiple areas, and then a hierarchical model of fault location is constructed, which greatly reduces the complexity of the construction of the switching function; the first layer of localization The hierarchical positioning strategy of fault area and fault area on the second layer reduces the fault area positioning dimension significantly and improves the efficiency of fault location. Combined with the characteristics of the small number of nodes in the second layer, the 0-1 integer planning The exhaustive method is introduced into the segment location, and the result of the fault location location using the intelligent algorithm is fed back and verified, so that the stability and fault tolerance of the whole fault identification process are greatly improved; the higher the penetration rate, the more the number of nodes. The advantages of hierarchical model and hierarchical intelligent verification algorithm are more obvious in fault location of the distribution network of 2000-2000.

In order to achieve the above object, the technical scheme of the present invention is: a method for locating faults in a distribution network based on a hierarchical model and an intelligent verification algorithm. Divide and perform two-port equivalent for each branch, each two-port contains one area and one area node; when a fault occurs, the FTU collects the status information of all nodes in the entire distribution network and uploads it to the SCADA system; area positioning; The algorithm reads the status information of all regional nodes from the SCADA system, and locates the fault to a specific area; the section locating algorithm reads the status information of all nodes in the faulty second port, and locates the fault to a specific section; using the section positioning results Verify the area positioning, if it is consistent, output the positioning result; if not, return the segment positioning result as the initial assignment of the second area positioning, and then repeat the positioning process until the result is output.

In an embodiment of the present invention, the method is specifically implemented as follows:

Step S1, perform two-port equivalence on each branch of the distribution network including DG:

In the fault location of the distribution network with DG, the switching function is generally constructed by the following formula based on the logical relationship:

I _j (s)=I _ju (s)-I _jd (s) (3)

I _j (s) represents the switching function, and I _ju (s) and I _jd (s) represent the upstream switching function and the downstream switching function, respectively; respectively represent the state of the section from node j to upstream power source s _u , and from node j to downstream power source s _d , s _u and s _d include three types of main power source S, distributed power source DG, and inductive load L, M′, N' is the number of upstream power sources and the number of downstream power sources, respectively; s _j,d , s _j,u represent the status of all sections between node j and downstream, node j and upstream, respectively, M, N are all upstream The number of sections and the number of all downstream sections; Π represents logical OR, K _u and K _d represent the upstream and downstream power coefficients, respectively, 1 when the power is connected, and 0 when the power is withdrawn;

Taking the three-branch node of the distribution network as an example, the logic rules in the construction of the switching function are analyzed:

It is assumed that the three-branch node of the distribution network includes three branches a, b, and c. 6 three nodes and (4), (5), (6) sections, branch c includes 7, 8, 9 three nodes and (7), (8), (9) sections;

When any one of the sections (7), (8) and (9) on branch c fails alone or any two sections fail at the same time, according to formulas (1), (2), (3), it can be have to:

The switching function of the nodes on branch a satisfies:

I ₁ (s)=I ₂ (s)=I ₃ (s)=1 (4)

All switch functions on branch b satisfy:

I ₄ (s)=I ₅ (s)=I ₆ (s)=-1 (5)

It can be obtained that as long as the fault is on branch c, no matter which section fails or multiple sections fail at the same time, the influence of branch c on the construction of other branch switching functions is the same; according to the equivalence rule, the section of branch c (7), (8), (9) When constructing the switch function, the entire branch is regarded as a passive network, which is equivalent to a two-port externally, and the two terminals are k ₁ , k ₂ respectively; the same is true for branch b When constructing the switch function, it can be equivalent to a two-port externally, and the two terminals are k ₃ and k ₄ ; when the switch function is constructed, the branch a can be externally equivalent to a two-port, and the two terminals are k ₅ respectively. , k ₆ ;

Step S2, build a fault location hierarchical model:

The three equivalent ports are connected in a star shape, the neutral point is the three-branch node, and the three outgoing lines are respectively connected to the main power source S, the distributed power source DG, and the inductive load L to obtain a hierarchical model: three-branch node, three equivalent Two ports a, b, c, three equivalent power sources S, DG, L; among them, the main power source S, distributed power source DG, and inductive load L constitute the first-layer positioning model, and three equivalent two ports a, b, The inner part of c is the second layer positioning model;

Step S3, locating the fault area:

Collect the state code information of each two-port area node, use equations (1), (2), (3) to construct the switching function of the entire distribution network about the area, and use equation (6) to construct the fitness function of fault area location, and then according to BPSOGA algorithm to locate the fault to the fault area;

Among them, fit(n) represents the fitness value of the nth individual, the equivalent number of two ports is D, and the number of nodes in the entire distribution network is T; I _j is the fault current direction information collected by the regional node FTU, I _j (s) is the switch function about the region, s _i is the region state code, and n is the weight coefficient;

Step S4, fault section location:

Inside each fault two-port, the system collects the node state coding information of the segment, uses the dual-source network switch function formula (7) or the single-source switch function formula (8) to construct the switch function about the segment, and uses the formula (9) to construct The fitness function of segment location, and then locate the fault to a specific segment according to the exhaustive method;

Among them, fit(n) represents the fitness value of the nth individual, the number of individuals included in each fault area is D ₁ , and the number of nodes in the entire distribution network is T. I _j is the fault current direction information collected by all section nodes FTU inside the fault two port, I _j (s) is the switching function about the section, s _i is the section state code, and n is the weight coefficient;

Step S5, positioning feedback verification:

After segment positioning, judge whether the area positioning result is consistent with the segment positioning result according to the verification criterion; if the results are inconsistent, return the segment positioning result to the area positioning, and use the segment positioning result as the initial assignment to calculate the fitness, If the fitness value is greater than the group optimal fitness value of the first regional positioning, jump directly into the segment positioning; if the fitness value is smaller than the group optimal fitness value of the first regional positioning, the fault area Positioning and section positioning, if the section positioning results are consistent, the positioning results will be output. If they are still inconsistent, enter the next verification cycle until the section positioning results are consistent with the section positioning results.

In an embodiment of the present invention, in the step S5, the method for determining the verification criterion is as follows:

a) If the misjudged area is a single-source network, then the segment node state code is I _j = [000], the switch function is calculated according to equation (8), and the fitness is calculated according to equation (9), when the segment state code is When s _j =[000], the fitness takes the maximum value:

fit _max =2T-(|0|+η·0)=2T (10)

Therefore, it is obtained that the segment state code is s _j =[000], according to which it can be determined that there is no faulty segment in the faulty area;

b) When the misjudged area is a dual-source network,

At this time, the segment node state code is I _j =[111]or[-1-1-1], the switch function is calculated according to equation (7), and the fitness is calculated according to equation (9), when the segment state code is I _j = [100] or [001] that is, when the boundary section fails, the fitness takes the maximum value:

fit _max1 =2T-(|1+1|+η·1)=2T-2.5 (11)

If there is actually a fault in the area, that is, I _j ≠[111]or[-1-1-1], the possible minimum value of the maximum fitness is:

fit _max2 =2T-(|0|+η·2)=2T-1 (12)

It can be seen that although the segment state code cannot be used to check the misjudged area for the dual-source network, it can be checked by the deviation range of the maximum fitness. When the maximum fitness value exceeds [2T-1, 2T+1] When it is within the range, it is determined that there is no fault section in the fault area.

Compared with the prior art, the present invention has the following beneficial effects:

(1) The present invention establishes a hierarchical model of distribution network fault location based on the characteristics of distribution network topology and switching function, which can greatly reduce the dimension established by the switching function and the complexity of the location model;

(2) The hierarchical model greatly reduces the operational dimension of fault area location, and together with the efficient segment location method, the efficiency of the entire fault location is greatly improved;

(3) Use the secondary section positioning results to verify the primary area positioning results, correct the misjudgment of the area positioning, and further improve the fault tolerance of the entire positioning;

(4) Use the absolute stability of the exhaustive method to make up for the defect of the intelligent algorithm that is easy to converge to the local optimum, and improve the stability and accuracy of the entire positioning;

(5) The fault location model and strategy constructed by the present invention are especially suitable for the fault location problem of high penetration rate and large distribution network.

Description of drawings

Figure 1 shows the topology of the T-type distribution network.

Figure 2 shows the equivalent two-port of branch c.

Figure 3 shows the equivalent two-port of branch b.

Figure 4 is an equivalent two-port of branch a.

Figure 5 is a layered model diagram of a T-type distribution network.

Figure 6 is a single source network segment location.

Figure 7 shows dual-source network segment positioning.

Figure 8 is a flowchart of fault location.

Figure 9 is an analysis diagram of a distribution network case.

Figure 10 is a diagram of the first layer positioning model.

Figure 11 is a comparison of the iterative process of the four methods.

Detailed ways

The technical solutions of the present invention will be described in detail below with reference to the accompanying drawings.

The present invention provides a method for locating faults in a distribution network based on a hierarchical model and an intelligent verification algorithm. First, the distribution network with distributed power sources is divided into branches with multi-branch nodes as the boundary, and each branch is divided into branches. The two ports are equivalent, and each two ports contain one area and one area node; when a fault occurs, the FTU collects the status information of all nodes in the entire distribution network and uploads it to the SCADA system; the area location algorithm reads all areas from the SCADA system The state information of the node is used to locate the fault to a specific area; the segment location algorithm reads the state information of all nodes inside the faulty two-port, and locates the fault to a specific segment; uses the segment location result to verify the area location, if the same , the positioning result is output; if it is inconsistent, the segment positioning result is returned as the initial assignment of the second region positioning, and then the positioning process is repeated until the result is output.

Specifically, a method for locating faults in a distribution network based on a hierarchical model and an intelligent verification algorithm of the present invention includes the following steps:

Step 1: Perform two-port equivalence for each branch of the distribution network including DG.

In the fault location of the distribution network with DG, the switching function is generally constructed by the following formula based on the logical relationship:

I _j (s)=I _ju (s)-I _jd (s) (3)

I _j (s) represents the switching function, and I _ju (s) and I _jd (s) represent the upstream switching function and the downstream switching function, respectively; respectively represent the state of the section from node j to upstream power source s _u , and from node j to downstream power source s _d , s _u and s _d include three types of main power source S, distributed power source DG, and inductive load L, M′, N' is the number of upstream power sources and the number of downstream power sources, respectively; s _j,d , s _j,u represent the status of all sections between node j and downstream, node j and upstream, respectively, M, N are all upstream The number of sections and the number of all downstream sections; Π represents logical OR, K _u and K _d represent the upstream and downstream power coefficients, respectively, 1 when the power is connected, and 0 when the power is withdrawn.

Multi-branch nodes (three or more) are an important part of the distribution network topology, which determines the complexity of the distribution network topology, which also determines the complexity of the switching function construction. Therefore, take the three-branch node (T-type node) of the distribution network as an example, as shown in Figure 1, to analyze the logic law in the construction of the switching function:

1) When the segment (7) on branch c fails, there are s ₇ =1, s _i≠7 =0, according to formulas (1), (2), (3), the node 1 on branch a is obtained The switch function is:

I _1u (s)=(1-s ₁ |s ₂ |s ₃ )|(1-s ₄ |s ₅ |s ₆ )*(s ₇ |s ₈ |s ₉ )=1 (4)

I _1d (s)=(1-s ₇ |s ₈ |s ₉ )*(s ₄ |s ₅ |s ₆ |s ₇ |s ₈ |s ₉ )=0 (5)

I ₁ (s)=I _1u (s)-I _1d (s)=1 (6)

Similarly, the switching functions of nodes 2, 3 on branch a and nodes 4, 5, and 6 on branch b can be obtained as:

I ₂ (s)=I _2u (s)-I _2d (s)=1 (7)

I ₃ (s)=I _3u (s)-I _3d (s)=1 (8)

I ₄ (s)=I _4u (s)-I _4d (s)=-1 (9)

I ₅ (s)=I _5u (s)-I _5d (s)=-1 (10)

I ₆ (s)=I _6u (s)-I _6d (s)=-1 (11)

Obviously, the switching function of the nodes on branch a satisfies:

I ₁ (s)=I ₂ (s)=I ₃ (s)=1 (12)

All switch functions on branch b satisfy:

I ₄ (s)=I ₅ (s)=I ₆ (s)=-1 (13)

2) When the section (8) on branch c fails, construct the formula according to the switching function, and obtain the switching function on branch a as: I ₁ (s)=I ₂ (s)=I ₃ (s)= 1. Equation (12) is still satisfied, and the switching function on branch b is: I ₄ (s)=I ₅ (s)=I ₆ (s)=-1, which still satisfies Equation (13). Similarly, when the section (9) fails, the switching function on branch a satisfies equation (12), and the switching function on branch b satisfies equation (13).

3) When sections (7) and (8) on branch c fail at the same time, the switching function on branch a still satisfies equation (12), and the switching function on branch b also satisfies equation (13). Similarly, when double faults occur in sections (7) and (9), and sections (8) and (9), branches a and b still satisfy equations (12) and (13).

Through the above analysis, the following conclusions can be drawn: as long as the fault is on branch c, no matter which section fails or multiple sections fail at the same time, the influence of branch c on the construction of other branch switching functions is the same. According to the equivalence rule, the sections (7), (8), and (9) of branch c can be synthesized into a "generalized section" when constructing the switching function, that is, the area, and nodes 7, 8, and 9 can be synthesized into a "generalized section". "Segment node" is the regional node, the whole branch is regarded as a passive network, and "externally equivalent to" a two-port, and the two terminals are k ₁ and k ₂ respectively, as shown in Figure 2.

Set single fault and double fault on branch b. According to the above conclusions, construct the switching functions of other non-faulty branches a and c. It can be found that branch b can also be "externally equivalent" to a two-port when constructing the switching function. , the two terminals are respectively k ₃ and k ₄ , as shown in Figure 3.

In the same way, set a single fault and a double fault on branch a, and construct the switching functions of other non-faulty branches b and c. It can be concluded that branch a can be "externally equivalent" to a two-port when constructing a switching function, The two terminals are respectively k ₅ and k ₆ , as shown in FIG. 4 .

Step 2: Build a fault location hierarchical model.

The three equivalent ports are connected in a star shape, the neutral point is the three-branch node, and the three outgoing lines are respectively connected to the main power source S, the distributed power source DG, and the inductive load L. The hierarchical model of Figure 1 is shown in Figure 5. Three-branch nodes, three equivalent two-ports (a, b, c), three equivalent power sources (main power S, distributed power DG, and inductive load L) constitute the first-layer positioning model, and three equivalent two-ports Inside is the second layer positioning model.

Step 3: Locate the fault area.

The system first collects the state code information of each two-port area node, uses equations (1), (2), (3) to construct the switching function of the entire distribution network about the area, and uses equation (14) to construct the fitness function of fault area location, Then according to the BPSOGA algorithm (Jin Tao, Li Hongnan, Liu Duan. BPSOGA-based distribution line fault location with wind turbines [J]. Electric Power Automation Equipment, 2016, 36(06): 27-33.), the fault location is to the fault area.

Among them, fit(n) represents the fitness value of the nth individual, the number of equivalent two ports is D, and the number of nodes in the entire power distribution network is T. I _j is the fault current direction information collected by the area node FTU, I _j (s) is the switching function about the area, s _i is the area state code, and η is the weight coefficient, which is usually set to 0.5.

Step 4: Locate the fault zone.

Inside each fault two-port, the system collects the node state coding information of the segment, and uses the dual-source network switch function formula (15) or the single-source switch function formula (16) to construct the switch function about the segment, and uses the formula (17) to construct The fitness function of the segment location, and then according to the exhaustive method, the fault is located to the specific segment.

Among them, fit(n) represents the fitness value of the nth individual, the number of individuals included in each fault area is D ₁ , and the number of nodes in the entire distribution network is T. I _j is the fault current direction information collected by the FTUs of all section nodes inside the fault two port, I _j (s) is the switching function about the section, and _si is the section state code. η is the weight coefficient, which is usually set to 0.5.

Step 5: Positioning feedback verification.

In order to improve the fault tolerance and accuracy of the entire fault location and overcome the “immature convergence” of the intelligent algorithm, a feedback verification mechanism is introduced into the algorithm. Whether the result is consistent with the segment location result; if the result is inconsistent, return the segment location result to the segment location, and use the segment location result as the initial assignment to calculate the fitness. If the fitness value is greater than the first segment location, the group is the most If the fitness value is smaller than the optimal fitness value of the group in the first regional positioning, then the fault area positioning and section positioning are performed. If the section positioning results are consistent, then Output the positioning result. If it is still inconsistent, enter the next verification cycle until the regional positioning result is consistent with the segment positioning result.

The method of determining the verification criterion is as follows:

a) If the misjudged area is a single-source network, as shown in Figure 6. At this time, the segment node (switch) state code is I _j = [000], the switch function is calculated according to equation (16), and the fitness is calculated according to equation (17). When the segment state code is s _j =[000], the adaptive Take the maximum value:

fit _max =2T-(|0|+η·0)=2T (18)

Therefore, it is obtained that the segment state code is s _j =[000], according to which it can be determined that there is no faulty segment in the faulty area.

b) When the misjudged area is a dual-source network, as shown in Figure 7:

At this time, the segment node state code is I _j =[111]or[-1-1-1], the switch function is calculated according to equation (15), and the fitness is calculated according to equation (17), when the segment state code is I _j = [100] or [001] that is, when the boundary section fails, the fitness takes the maximum value:

fit _max1 = 2T-(|1+1|+η·1)=2T-2.5 (19)

If there is actually a fault in the area, that is, I _j ≠[111]or[-1-1-1], the possible minimum value of the maximum fitness is:

fit _max2 =2T-(|0|+η·2)=2T-1 (20)

According to this, it can be found that although the segment state code cannot be used to check the misjudged area for the dual-source network, it can be checked by the deviation range of the maximum fitness. When the maximum fitness value exceeds [2T-1,2T +1] range, it is determined that there is no fault zone in the fault area.

The flowchart of the entire fault location is shown in Figure 8.

Example:

Build a distribution line model containing wind turbines according to Figure 9. The model has 30 feeder nodes and 30 sections. The specific numbers are shown in the figure. S is the main power supply of the system, DG1 and DG2 are wind turbines, and L1 and L2 are inductive loads.

First, the distribution network is equivalent to a combination of ten two-ports with the multi-branch nodes as the boundary, each two-port contains an area and an area node, and the first-layer positioning model is constructed, as shown in Figure 10. Each two-port belongs to the second-layer positioning model, and the segment nodes and segments included in each two-port are shown in Table 1.

Table 1 Nodes and Sections Contained by Two Ports

When the section 5 of the setting area 3 fails and all the distributed power sources are put into operation, the fault direction information of all nodes uploaded by the FTU [11111-1-1-1-1-1-1-1-1-1-1 -1-1000-1-1-1-1-1-1-1-1-1-1]. The BPSOGA algorithm first reads all nodes in the fault area [1(1)2(2)3(3)4(6)5(10)6(13)7(18)8(21)9(23)10(28) ] Uploaded fault direction information: [111-1-1-10-1-1-1], and then locate the fault area, the maximum fitness function value of the fault area location is 59.5, and the corresponding area status is: [ 0010000000], so it is judged that the area 3 is faulty.

The exhaustive method reads the fault direction information of all nodes in fault area three (345) according to the positioning result of the BPSOGA algorithm: (111), and then locates the fault section. When the maximum value of the segment fitness function is 59.5, the corresponding segment state is: [001], and the exhaustive method determines that segment 5 is faulty.

The verification mechanism checks that the regional positioning results are consistent with the section results, and outputs the positioning results.

It is set that section 3 of area 3 and section 10 of area 5 fail at the same time. At this time, the fault direction information uploaded by the FTU of node 2 is distorted from 1 to 0, and the BPSOGA algorithm misjudged, and the maximum fitness value obtained is 57.5 , the corresponding area status is: [0 1 1 0 1 0 0 0 0 0], it is determined that the two, three and five areas are faulty. Therefore, the three fault areas are located at the same time, and the results of the segment location are shown in Table 2:

Table 2 Segmentation results

As can be seen from the above table, there is no faulty section in area 2, and the area location is inconsistent with the segment location result, so the segment location result is returned and the area location is performed again. The initial value of the area state is [0010100000], and the result of the area location is the three and five faults in the area. Further, it is concluded that the segment location results are faults in segments 3 and 10, the two location results are consistent, and the results are output.

The above simulation results confirm that the layered model proposed by the present invention reduces the operation dimension from 30 to 10, simplifies the positioning model, and improves the positioning efficiency. At the same time, it is proved that the verification mechanism can effectively avoid the "immature convergence" problem of the intelligent algorithm, and improve the stability and fault tolerance of the entire positioning algorithm.

In addition, the fault location simulation was carried out considering three conditions, namely, the number of faults, the number of distributed power sources, and the distortion of data uploaded by the FTU. The simulation results are shown in Tables 3 and 4.

Table 3 Single fault simulation results

Table 4. Double fault simulation results

It can be seen from Tables 3 and 4 that the model and method proposed by the present invention have relatively high positioning error tolerance and accuracy.

In order to verify the advantages of the method proposed in the present invention in the positioning model, the layered model HBPSOGA proposed in the present invention is compared with three single-layer models of GA, BPSO and BPSOGA. The populations of the four models are all set to 100, and the maximum number of iterations is 100. The number of iterations of the exhaustive method of HBPSOGA is subject to the actual situation. Setting a single fault at section 3, the iterative optimization process of the four methods is shown in Figure 11. Among them, HBPSOGA1 represents regional localization and HBPSOGA2 represents segmental localization. For the four methods, single fault and double fault are set in 3 different sections. Each fault is run 30 times.

Table 5 Performance comparison table of four methods

From the above simulation results, it can be found that since the dimension of the region positioning of the hierarchical model is only 10, the positioning time is 0.94s; in the section positioning, the operation dimension is only 3, and the number of iterations using the exhaustive method is: The time-consuming is 0.01s; the whole positioning process takes no more than 1s, and the total number of iterations is 13. The average time-consuming and average convergence times of the hierarchical model of the present invention are significantly less than those of the other three single-layer models. The proposed method has obvious advantages in simplifying the positioning model and positioning speed;

In order to verify the advantages of the proposed method in the positioning algorithm, the algorithm HBPSOGA+CM (Check Mechanism) of the feedback check mechanism is compared with the algorithm HBPSOGA of the section check mechanism and the algorithm BPSOGA without the check mechanism. Among them, BPSOGA only locates the fault area of the hierarchical model, HBPSOGA only uses segment location for verification, and HBPSOGA+CM adds feedback links on the basis of HBPSOGA. The three methods uniformly reduce the population and the number of iterations to 30, set single faults and double faults in different sections, and run each fault 30 times. shown in Table 6.

Table 6 Statistics of BPSOGA regional positioning results

In addition, the particle swarm algorithm with feedback verification mechanism and genetic algorithm HBPSO+CM, GA+CM and hierarchical particle swarm optimization with segment verification mechanism and genetic algorithm HBPSO, HGA and particles without verification mechanism are also given. The comparison results of swarm algorithm and genetic algorithm BPSO and GA are shown in Tables 7 and 8.

Table 7 Statistics of BPSO regional positioning results

Table 8 Statistics of GA regional positioning results

From Tables 6, 7, and 8, it can be seen that in terms of the number of accurate positioning, most of the regional positioning can be corrected by simply relying on the section positioning verification, because the results of the "immature convergence" of the intelligent algorithm in most cases include " As a result of "mature convergence", segment positioning can filter out the misjudged regions in the region positioning, and only allow the correctly positioned regions to be output. The verification mechanism with feedback link not only has a "filtering" corrective effect, but also has a "supplementary" corrective effect, which can correct the "immature convergence" results that do not cover and partially cover the "mature convergence" results. Although the mechanism increases the number of iterations and time-consuming to a certain extent, its impact on the quickness of fault location is negligible.

The above are the preferred embodiments of the present invention, all changes made according to the technical solutions of the present invention, when the resulting functional effects do not exceed the scope of the technical solutions of the present invention, belong to the protection scope of the present invention.

Claims (2)

1. a kind of electrical power distribution network fault location method based on hierarchical mode and intelligent checking algorithm, it is characterised in that: first to containing The power distribution network of distributed generation resource carries out branch k-path partition by boundary of multi-branch node, equivalent to every branch progress Two-port netwerk, often A Two-port netwerk contains a region and an Area Node；When an error occurs, all node shapes of entire power distribution network of FTU acquisition State information, and upload SCADA system；Zone location algorithm reads the status information of all areas node from SCADA system, will Fault location is to specific region；The status information of all nodes, failure is determined inside section location algorithm read failure Two-port netwerk Specific section is arrived in position；Zone location is verified using section positioning result, if unanimously, exporting positioning result；If different It causes, section positioning result is returned and the initial assignment as second of zone location, then resetting process, until output As a result；This method is implemented as follows,

Step S1, equivalent to each branch of power distribution network containing DG progress Two-port netwerk:

In the positioning of distribution network failure containing DG, switch function is generally used and is constructed below based on the formula of logical relation:

I_j(s)=I_ju(s)-I_jd(s) (3)

I_j(s) switch function, I are indicated_ju(s)、I_jd(s) upstream switch function and downstream switches function are respectively indicated；s_j,su、s_j,sd It respectively indicates from node j to upstream power supply s_u, node j to downstream power supply s_dBetween section state, s_uAnd s_dIncluding main power source S, Distributed generation resource DG, inductive load L three types, M ', N ' are respectively the number and downstream power supply number of upstream power supply；s_j,d、 s_j,uNode j is respectively indicated to downstream, the state of node j to sections all between upstream, M, N are respectively all sections in upstream The number of all sections of number and downstream；Π indicates logic or, K_u、K_dThe power supply coefficient of upstream and downstream is respectively indicated, power supply connects Entering then is 1, and it is then 0 that power supply, which exits,；

By taking three branch node of power distribution network as an example, the logical laws in switch function building are analyzed:

If three branch node of power distribution network includes three branches a, b, c, branch a includes 1,2,3 three node and (1), (2), (3) Section, branch b include 4,5,6 three nodes and (4), (5), (6) section, branch c include 7,8,9 three nodes and (7), (8), (9) section；

When section (7), (8), (9) any zone on branch c individually break down or wantonly two section while when breaking down, root According to formula (1), (2), (3), can obtain:

The switch function of institute's node meets on branch a:

I₁(s)=I₂(s)=I₃(s)=1 (4)

All switch functions on branch b meet:

I₄(s)=I₅(s)=I₆(s)=- 1 (5)

It can thus be concluded that: as long as failure, on branch c, whichever section fault or multiple section simultaneous faults, branch c is to it His influence of branch switch function building is identical；According to equivalent rule, the section (7) of branch c, (8), (9) are constructing switch letter When number, entire branch regards passive network as, externally equivalent at a Two-port netwerk, and two terminals are respectively k₁, k₂；Similarly branch b , can be externally equivalent at a Two-port netwerk when constructing switch function, two terminals are respectively k₃, k₄；Branch a is switched in building , can be externally equivalent at a Two-port netwerk when function, two terminals are respectively k₅, k₆；

Step S2, fault location hierarchical mode is constructed:

Three equivalent ports are subjected to star-like connections, neutral point is three branch nodes, three outlets be separately connected main power source S, Distributed generation resource DG, inductive load L obtain hierarchical mode: three branch nodes, three equivalent Two-port netwerk a, b, c, three equivalent electricity Source S, DG, L；Wherein, main power source S, distributed generation resource DG, inductive load L constitute first layer location model, three equivalent Two-port netwerks It a, is second layer location model inside b, c；

Step S3, fault zone positions:

Each Two-port netwerk Area Node state encoding information is acquired, constructs entire power distribution network about area using formula (1), (2), (3) The switch function in domain, using the fitness function of formula (6) building fault zone positioning, then according to BPSOGA algorithm, by failure Navigate to fault zone；

Wherein, fit (n) indicates the fitness value of n-th of individual, and equivalent Two-port netwerk number is D, the number of nodes of entire distribution network For T；I_jFor the fault current directional information of Area Node FTU acquisition, I_jIt (s) is switch function about region, s_iFor area-shaped State coding, η is weight coefficient；

Step S4, fault section location:

Inside each failure Two-port netwerk, system acquisition section node state encoded information is public using double source network switching function Formula (7) or single source switch function formula (8) construct the switch function about section, utilize the adaptation of formula (9) building section positioning Function is spent, then according to the method for exhaustion, by fault location to specific section；

Wherein, fit (n) indicates that the fitness value of n-th of individual, the number of individuals that each fault zone includes are D₁, entire power distribution network Node number be T；I_jFor the fault current directional information of section node FTU acquisitions all inside failure Two-port netwerk, I_j(s) it is About the switch function of section, s_iFor sector status coding, η is weight coefficient；

Step S5, position feedback verifies:

It is whether consistent according to verification criterion critical region positioning result and section positioning result after section positioning；If result is not Unanimously, the result return area by section positioning positions, and using section positioning result as initial assignment, calculates fitness, if this is suitable It answers angle value to be greater than group's adaptive optimal control angle value of first time zone location, then directly jumps into section positioning；If this fitness value is small In group's adaptive optimal control angle value of first time zone location, then fault zone positioning and section positioning are carried out, if section positioning knot Fruit is consistent, then exports positioning result, if still inconsistent, into verification next time circulation, until zone location result and section are fixed Position result is consistent.

2. a kind of electrical power distribution network fault location method based on hierarchical mode and intelligent checking algorithm according to claim 1, It is characterized by: the determination method for verifying criterion is as follows in the step S5:

If a) region judged by accident is single source network, section node state is encoded to I at this time_j=[000] calculates according to formula (8) and switchs Function calculates fitness according to formula (9), when sector status is encoded to s_jWhen=[000], fitness is maximized:

fit_max=2T- (| 0 |+η 0)=2T (10)

Then show that sector status is encoded to s_j=[000] can be determined that there is no fault sections for the fault zone accordingly；

B) when the region of erroneous judgement is double source network,

Section node state is encoded to I at this time_j=[111] or [- 1-1-1] calculates switch function according to formula (7), according to formula (9) Fitness is calculated, when sector status is encoded to I_j=[100] or [001] i.e. boundary section fault when, fitness is maximized:

fit_max1=2T- (| 1+1 |+η 1)=2T-2.5 (11)

If the region physical presence failure, that is, I_j≠ [111] or [- 1-1-1], then the possibility minimum value of maximum adaptation degree are as follows:

fit_max2=2T- (| 0 |+η 2)=2T-1 (12)

It, can be by most although being verified as it can be seen that cannot be encoded with sector status for double source network to erroneous judgement region The deviation range of big fitness is verified, and when maximum adaptation angle value exceeds [2T-1,2T+1] range, determines the faulty section Fault section is not present in domain.