Power distribution network power supply self-healing method and system based on hypersphere clustering algorithm

Abstract

The invention discloses a power distribution network power supply self-healing method based on a hypersphere clustering algorithm, and the method comprises the steps: constructing a fault recovery database, and carrying out the secondary kernel cross-integral hypersphere clustering of fault data in the fault recovery database, so as to construct a hypersphere clustering model; determining a clustering maximum range constraint condition of the hyper-sphere clustering model, and determining a cross constraint condition of the hyper-sphere clustering model; when the power distribution network has a fault, matching the fault data of the current fault with the fault data of the historical fault according to the hypersphere clustering model, and if matching succeeds, executing a corresponding power supply recovery scheme in the fault recovery database; and if the matching is unsuccessful, constructing a new power supply recovery scheme, recovering power supply for part of lines or all lines according to the new power supply recovery scheme, adding fault data corresponding to the fault to the hypersphere clustering model, and updating the hypersphere clustering model. According to the design, the speed of calling the fault recovery scheme can be increased, and the power supply self-healing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to a power supply self-healing method for a distribution network, and in particular to a power supply self-healing method and system for a distribution network based on a hypersphere clustering algorithm. Background Art

[0002] The distribution network is the key link between the load and the upper power grid in the power system. Its safety and reliability directly affect the stability of power supply. In the distribution network, there are many facilities such as transformers, switchgear, and protection devices. The diversity of these devices and the complex control strategies make fault management particularly difficult. Especially when a large number of DGs are connected, the intermittent and uncertain nature of DGs significantly increases the volatility of the distribution network operation, and the flow of current becomes more complex and changeable, which places higher demands on the power supply self-healing ability of the distribution network.

[0003] The present invention integrates the historical faults of DG access to the distribution network, builds a historical fault recovery database, and designs secondary core cross-integral hypersphere clustering on this basis to build a hypersphere clustering model. By building a hypersphere clustering model, the speed of calling the fault recovery plan is accelerated and the efficiency of power supply self-healing is improved. Summary of the invention

[0004] The purpose of the present invention is to overcome the problem of insufficient self-healing ability of power distribution network in the prior art, and to provide a method and system for self-healing power distribution network based on a hypersphere clustering algorithm.

[0005] To achieve the above objectives, the technical solution of the present invention is:

[0006] In a first aspect, the present invention provides a power supply self-healing method for a distribution network based on a hypersphere clustering algorithm, comprising:

[0007] S1. Obtain historical operation data of the distribution network, and build a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan;

[0008] S2, performing secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to construct a hypersphere clustering model, and determining the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model;

[0009] S3. When a fault occurs in the distribution network, the fault performance corresponding to the current fault is matched with the fault data corresponding to each historical fault according to the hypersphere clustering model;

[0010] S4. Execute the power supply restoration plan of the distribution network according to the matching results:

[0011] If the match is successful, the power supply recovery plan corresponding to the matched fault data is executed, and the power supply self-healing of the distribution network is completed;

[0012] If the matching is unsuccessful, a new power supply restoration plan is constructed according to the DG output, energy storage, load level, and line flow data in the current distribution network, and power supply is restored to some or all lines according to the new power supply restoration plan, and the new power supply restoration plan is recorded, and then step S5 is entered;

[0013] S5. Add the fault type, fault manifestation, and new power supply restoration plan corresponding to the current fault to the hypersphere clustering model, and update the hypersphere clustering model.

[0014] The historical operation data of the power grid includes: current and voltage in the feeder collected by the protection relay, circuit breaker, FTU, and DTU, and dynamic information of DG load.

[0015] In S4, a new power supply restoration plan is constructed according to the current DG output, energy storage, load level, and line flow data in the distribution network. Restoring power supply to some or all lines according to the new power supply restoration plan includes:

[0016] S41. Calculate the total load of the fault area:

[0017]

[0018] Among them, ω i is the load level, P load,i is the load of feeder i in the fault area, C is the set of fault feeders;

[0019] S42. According to the total load in the fault area, as well as the DG output and energy storage conditions of the distribution network, determine whether the load in the fault area can be fully or partially powered by DG and energy storage. If so, restore full or partial power supply to the fault area.

[0020] S43. If the loads in the faulty area cannot be fully powered by DG and energy storage, determine whether the power can be transferred from the adjacent feeder. If so, transfer power to the faulty area through the adjacent feeder.

[0021] S44. For fault areas where full power supply cannot be restored through DG, energy storage, and adjacent feeder transfer, the line flow in the fault area is calculated, and the loads in the fault area are gradually removed and power supply is restored based on the calculation results.

[0022] The performing of secondary kernel cross-integration hypersphere clustering on the data in the fault recovery database comprises:

[0023] S21. According to the fault recovery database, a set N of fault data corresponding to n historical faults is obtained, wherein the fault data corresponding to the i-th historical fault is n i ;

[0024] N={n 1 、n 2 、n 3 ,...,n n};

[0025] S22, divide the fault data into m categories according to the fault type, each category of fault data corresponds to the same fault type, obtain m fault data clusters corresponding to the m fault types, calculate the center of each fault data cluster to obtain the m fault data cluster centers corresponding to the m fault types, wherein the fault data cluster center corresponding to the jth fault type is c j ;

[0026] C={c 1 、c 2 、c 3 , ..., c m};

[0027] S23. For each fault data, extract the fault manifestation corresponding to the fault data, and obtain a set U of fault manifestations corresponding to n historical faults, where the fault manifestation corresponding to the i-th historical fault is u i ;

[0028] U={u 1 、u 2 、u 3 ,...,u n};

[0029] S24, classify the fault manifestations into m categories according to the fault type, each category of fault manifestation corresponds to the same fault type, obtain m fault manifestation clusters corresponding to the m fault types, calculate the center of each fault manifestation cluster to obtain the m fault manifestation cluster centers corresponding to the m fault types, wherein the fault manifestation cluster center corresponding to the jth fault type is g j ;

[0030] G = {g 1 , g 2 , g 3 , ..., g m};

[0031] S25. Update the fault data cluster, fault data cluster center, fault manifestation cluster, and fault manifestation cluster center according to the hypersphere distance between each fault data and the fault data cluster center, and the hypersphere distance between each fault manifestation and the fault manifestation cluster center, so as to bring the center points closer together, perform fitness calculation on the fault manifestation cluster, select the optimal cluster center according to the fitness calculation result, construct the minimum hypersphere according to the optimal cluster center, and obtain the corresponding optimal cluster range.

[0032] The i-th fault data n in the set of fault data N i The fault data cluster center c corresponding to the jth fault type j The hypersphere distance between is calculated according to the following formula:

[0033]

[0034] In the formula, d(n i ,c j ) represents fault data n i and the fault data cluster center c j The hypersphere distance between cov(n i ,c j ) represents fault data n i and the fault data cluster center c j The covariance between σ(n i ),σ(c j ) represent the variance of fault data and fault data cluster center respectively; ε ij Represents the distance calculation error adjustment term, 0≤ε ij ≤1;

[0035] The i-th fault manifestation u in the set M of fault manifestations i The fault manifestation cluster center corresponding to the jth fault type is g j The hypersphere distance between is calculated according to the following formula:

[0036]

[0037] In the formula, d(u i ,g j ) indicates the fault manifestation u i and the fault performance cluster center g j The hypersphere distance between them; cov(u i ,g j ) indicates the fault manifestation u i and the fault performance cluster center g j The covariance between σ(u i ),σ(g j) represent the fault manifestation and the variance of the fault manifestation cluster center respectively.

[0038] The maximum clustering range constraints for the hypersphere clustering model include:

[0039] According to the fault data cluster center and fault manifestation cluster center, weights are assigned to the fault manifestations, and the edge range of the intersecting clusters is calculated in the hypersphere to determine the maximum range constraint of the clustering. The calculation formula is:

[0040]

[0041] Where V Fj is the maximum range constraint of the clustering of the hypersphere corresponding to the jth fault type; ω i is the weight; N i is the fault manifestation u corresponding to the jth fault type i The total number of i According to u i The frequency setting of the corresponding fault type in the fault recovery database.

[0042] The cross-constraints that determine the hypersphere clustering model include:

[0043] According to the maximum range of clustering, cross constraints are set to absorb similar cluster cores. The calculation formula is:

[0044]

[0045] In the formula, is the cross constraint condition of the hypersphere corresponding to the jth fault type.

[0046] In a second aspect, the present invention provides a distribution network power supply self-healing system based on a hypersphere clustering algorithm, comprising:

[0047] A database construction module is used to obtain historical operation data of the distribution network, and to construct a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records the fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan;

[0048] A model building module is used to perform secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to build a hypersphere clustering model, and determine the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model;

[0049] A matching module is used to match the fault performance corresponding to the current fault with the fault data corresponding to each historical fault according to the hypersphere clustering model when a fault occurs in the distribution network;

[0050] The execution module executes the distribution network power supply restoration plan according to the matching results:

[0051] If the match is successful, the power supply recovery plan corresponding to the matched fault data is executed;

[0052] If the matching fails, a new power supply restoration plan is constructed based on the current DG output, energy storage, load level, and line flow data in the distribution network, and power supply is restored to some or all lines according to the new power supply restoration plan;

[0053] The model updating module is used to add the fault type, fault manifestation and new power supply restoration plan corresponding to the current fault to the hypersphere clustering model, and update the hypersphere clustering model.

[0054] In a third aspect, the present invention provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned distribution network power supply self-healing method based on the hypersphere clustering algorithm when executing the computer program.

[0055] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned distribution network power supply self-healing method based on a hypersphere clustering algorithm.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] In a power supply self-healing method for a distribution network based on a hypersphere clustering algorithm of the present invention, secondary kernel cross-integral hypersphere clustering is performed on data in a fault recovery database to construct a hypersphere clustering model, and the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model are determined. When a fault occurs in the distribution network, the fault performance corresponding to the current fault is matched with the fault data corresponding to each historical fault according to the hypersphere clustering model. If the match is unsuccessful, a new power supply recovery plan is constructed and the fault data corresponding to the current fault is added to the hypersphere clustering model. The hypersphere clustering model dynamically adjusts the strategy according to the change of the distribution network and the system state, so as to improve the adaptive ability of the distribution network to faults. At the same time, the hypersphere clustering model is constructed by secondary kernel cross-integral hypersphere clustering, so as to optimize the fault data structure and improve the post-fault matching efficiency. Therefore, this design optimizes the fault data structure through the quadratic kernel cross-integral hypersphere clustering algorithm and improves the post-fault matching efficiency; at the same time, the hypersphere clustering model dynamically adjusts the strategy according to the changes in the distribution network and the system status, thereby improving the distribution network's adaptability to faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 The present invention provides a flow chart of a method for self-healing power supply in a distribution network based on a hypersphere clustering algorithm.

[0059] Figure 2 The present invention provides a flowchart of restoring power supply to a part of the lines or all the lines.

[0060] Figure 3 It is a structural schematic diagram of a distribution network power supply self-healing system based on a hypersphere clustering algorithm provided in an embodiment of the present invention.

[0061] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention.

[0062] Figure 5 It is a structural diagram of the distribution network test system.

[0063] Figure 6 It is a schematic diagram of the fault location and division results of the distribution network test system. DETAILED DESCRIPTION

[0064] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0065] In related technologies, the distribution network includes a variety of facilities such as transformers, switchgear, and protection devices. The diversity and complex control strategies of these devices make fault management particularly difficult. Especially when a large number of distributed generation (DG) are connected, the intermittent and uncertain nature of DG significantly increases the volatility of the distribution network operation, causing the current flow to become more complex and changeable, which puts higher requirements on the power supply self-healing system of the distribution network.

[0066] In order to solve the above technical problems, the present invention adopts the secondary kernel cross-integral hypersphere clustering method to construct the optimal database for distribution network fault recovery, aiming to improve the power supply self-healing capability of the distribution network after DG access. The specific approach includes: collecting the distribution network fault data that has occurred, designing and establishing a historical fault recovery database; at the same time, combining the emerging new fault characteristics, using the secondary kernel cross-integral hypersphere clustering technology to optimize the database processing performance, thereby achieving rapid self-healing of the distribution network power supply. The embodiment of the present invention provides a distribution network power supply self-healing method based on the hypersphere clustering algorithm, please refer to Figure 1 , Figure 1 It is a flow chart of a distribution network power supply self-healing method based on a hypersphere clustering algorithm provided in an embodiment of the present invention. The method includes steps S1 to S5.

[0067] S1. Obtain historical operation data of the distribution network, and build a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan.

[0068] The historical operation data of the power grid includes: protection relays, circuit breakers, FTUs, and DTUs collect the current, voltage, and DG load dynamic information in the feeder, and use the above information to determine the dynamic information of the distribution network at each moment. According to the historical fault data and the historical power supply restoration plan, the characteristic quantities of the fault components, feeders, and DG output are extracted, and the state of the fault at this time is labeled to build a historical fault recovery database.

[0069] In this embodiment, the fault manifestations include: fault power outage time, DG power, load power, line current, line temperature, etc.

[0070] S2. Perform secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to construct a hypersphere clustering model, and determine the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model.

[0071] The secondary kernel cross-integration hypersphere clustering of the fault data in the fault recovery database includes the steps in S21-S21:

[0072] S21. According to the fault recovery database, a set N of fault data corresponding to n historical faults is obtained, wherein the fault data corresponding to the i-th historical fault is n i ;

[0073] N={n 1 、n 2 、n 3 ,...,n n}.

[0074] S22, divide the fault data into m categories according to the fault type, each category of fault data corresponds to the same fault type, obtain m fault data clusters corresponding to the m fault types, calculate the center of each fault data cluster to obtain the m fault data cluster centers corresponding to the m fault types, wherein the fault data cluster center corresponding to the jth fault type is c j ;

[0075] C={c 1 、c 2 、c 3 , ..., c m}.

[0076] S23. For each fault data, extract the fault manifestation corresponding to the fault data, and obtain a set U of fault manifestations corresponding to n historical faults, where the fault manifestation corresponding to the i-th historical fault is u i ;

[0077] U={u 1 、u 2 、u 3 ,...,u n}.

[0078] S24, classify the fault manifestations into m categories according to the fault type, each category of fault manifestation corresponds to the same fault type, obtain m fault manifestation clusters corresponding to the m fault types, calculate the center of each fault manifestation cluster to obtain the m fault manifestation cluster centers corresponding to the m fault types, wherein the fault manifestation cluster center corresponding to the jth fault type is g j ;

[0079] G = {g 1 , g 2 , g 3 , ..., g m}.

[0080] In this embodiment, as shown in Table 1 below, each fault type corresponds to a different fault manifestation.

[0081]

[0082]

[0083] Table 1 Common faults and their characteristics

[0084] For the same type of fault, the corresponding fault manifestation and line data are similar. Therefore, we use the fault type to classify the fault data and classify the data with similar fault factors and line data into one category. Similarly, we can also use the fault manifestation to quickly match the type of the current fault. For example, if the fault feature of "instantaneous current surge" appears, we can quickly match it with "short circuit fault" and provide a fault recovery plan in combination with the feeder flow data.

[0085] S25. Update the fault data cluster, fault data cluster center, fault manifestation cluster, and fault manifestation cluster center according to the hypersphere distance between each fault data and the fault data cluster center, and the hypersphere distance between each fault manifestation and the fault manifestation cluster center, so as to bring the center points closer together, perform fitness calculation on the fault manifestation cluster, select the optimal cluster center according to the fitness calculation result, construct the minimum hypersphere according to the optimal cluster center, and obtain the corresponding optimal cluster range.

[0086] In order to avoid the maximum boundary constraint of the constructed hypersphere and avoid the intersection between hyperspheres, we determined the maximum clustering range constraint of the hypersphere clustering model and the intersection constraint of the hypersphere clustering model.

[0087] The maximum clustering range constraints for determining the hypersphere clustering model include:

[0088] According to the fault data cluster center and fault manifestation cluster center, weights are assigned to the fault manifestations, and the edge range of the intersecting clusters is calculated in the hypersphere to determine the maximum range constraint of the clustering. The calculation formula is:

[0089]

[0090] Where V Fj is the maximum range constraint of the clustering of the hypersphere corresponding to the jth fault type; ω i is the weight; N i is the fault manifestation u corresponding to the jth fault type i The total number of i According to u i The frequency setting of the corresponding fault type in the fault recovery database.

[0091] In this embodiment, the distribution of the weights conforms to the normal distribution. i The frequency setting of the corresponding fault type in the fault recovery database, for u i After counting the frequency of the corresponding fault type in the fault recovery database, i The value of is assigned a weight of 0-1 according to the frequency.

[0092] The cross-constraints that determine the hypersphere clustering model include:

[0093] According to the maximum range of clustering, cross constraints are set to absorb similar cluster cores. The calculation formula is:

[0094]

[0095] In the formula, is the cross constraint condition of the hypersphere corresponding to the jth fault type.

[0096] Through the calculation of the maximum range constraint and cross constraint conditions of the above clustering, we calculated the maximum range of different hypersphere centers and similar hyperspheres, so that there are gaps between adjacent hyperspheres and adjacent hyperspheres do not intersect. This avoids the situation where the data samples corresponding to the new fault appear on the boundary of two hyperspheres when a new fault occurs, which leads to misclassification of the fault and the provision of an incorrect recovery plan.

[0097] S3. When a fault occurs in the distribution network, the fault performance corresponding to the current fault is matched with the fault data corresponding to each historical fault according to the hypersphere clustering model.

[0098] In this embodiment, we use the hypersphere clustering model to match the fault performance data with the historical fault data. If the matching is successful, the historical fault recovery database is called to immediately start the fault recovery plan. If the fault information fails to match the data in the historical database, a new power supply recovery plan needs to be built to deal with this fault.

[0099] S4. Execute the distribution network power supply restoration plan according to the matching results.

[0100] If the match is successful, the power supply recovery plan corresponding to the fault data that successfully matched in the fault recovery database is executed, and the power supply self-healing of the distribution network is completed;

[0101] If the match is unsuccessful, a new power supply restoration plan is constructed based on the current DG output, energy storage, load level, and line flow data in the distribution network. According to the new power supply restoration plan, power supply is restored to some or all lines, and the new power supply restoration plan is recorded. For details, please refer to Figure 2 , Figure 2 is a schematic diagram of a process for restoring power to a part of the lines or all the lines provided by an embodiment of the present invention, such as Figure 2 As shown, the method of restoring power supply to part of the lines or all the lines provided in this embodiment includes steps S21 to S26.

[0102] In S4, a new power supply restoration plan is constructed according to the current DG output, energy storage, load level, and line flow data in the distribution network. Restoring power supply to some or all lines according to the new power supply restoration plan includes:

[0103] S41. Calculate the total load of the fault area.

[0104]

[0105] Among them, ω i is the load level, P load,i is the load of feeder i in the fault area, C is the set of fault feeders;

[0106] S42. According to the total load in the fault area, as well as the DG output and energy storage conditions of the distribution network, determine whether the load in the fault area can be fully or partially powered by DG and energy storage. If so, restore full or partial power supply to the fault area.

[0107] S43. If the loads in the faulty area cannot be fully powered by DG and energy storage, determine whether the power can be transferred from the adjacent feeder. If so, transfer power to the faulty area through the adjacent feeder.

[0108] S44. For fault areas where full power supply cannot be restored through DG, energy storage, and adjacent feeder transfer, the line flow in the fault area is calculated based on the line flow data, and the loads in the fault area are gradually removed and power supply is restored based on the calculation results.

[0109] In the matching process, the feature quantity of paired clusters is selected, the fault information is matched and analyzed, the optimal cluster center and optimal cluster range of the fault solution are determined, and the optimal database for fault recovery is integrated to reduce data redundancy. For example, if the fault feature matching fails and the fault handling solution has not been stored in the historical fault database, the fault feature and its solution are collected and integrated.

[0110] S5. Add the fault type, fault manifestation, and new power supply restoration plan corresponding to the current fault to the hypersphere clustering model, and perform secondary kernel cross-integration hypersphere clustering on all fault data including the newly added data to update the hypersphere clustering model.

[0111] In summary, the present invention discloses a distribution network power supply self-healing method based on a hypersphere clustering algorithm, which constructs a historical fault recovery database by extracting historical fault state characteristics and fault handling solutions in the distribution network; performs secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to construct a hypersphere clustering model; matches the fault characteristics with the data in the hypersphere clustering model, and formulates corresponding recovery decision plans based on historical power supply restoration plans; next, for fault characteristics that fail to match, calls the DG output and energy storage conditions to restore partial or full load power supply to the fault area, and stores the fault handling plan in the fault recovery database; then, based on the secondary kernel cross-integral hypersphere clustering newly proposed by the present invention, the new fault characteristics and the new fault handling plan are integrated into the hypersphere clustering model, the data is again subjected to secondary kernel cross-integral hypersphere clustering, similar fault characteristics are merged, the weights of the fault occurrence frequencies are updated, the hypersphere clustering model is updated, and the structure of the optimal fault recovery database is optimized, so that the distribution network can respond to faults quickly, realize rapid self-healing of the power supply system, and improve the overall reliability and resilience.

[0112] We conducted power supply self-healing and reliability evaluation for the test system based on the main feeder F4 of the IEEE RBTSBU S6 system. Figure 5 As shown, Figure 5This is a structural diagram of the distribution network test system, which includes: 23 fuses, 23 distribution transformers, and 23 load points (each load point supplies multiple loads). The failure rate of distributed power supply is 5 times / year, and the average repair time is 50h. The loads in the distribution network are divided into two categories: agricultural loads and urban loads. The line data and load data of the distribution network test system are shown in Table 2 and Table 3 below:

[0113]

[0114] Table 2 Line data

[0115]

[0116] Table 2 Line data

[0117]

[0118] Table 3 Load data

[0119] In order to analyze the impact of the access location and access capacity of distributed resources on the reliability of the distribution network, present more comprehensive data support, and determine the feasibility of the research content, the average power outage duration of the system, the average power supply availability, the average power outage frequency of the system, the system power shortage, and the average power shortage of the system are used as evaluation indicators.

[0120] See also Figure 5 , distributed power sources with different capacity ratios are connected to 13 and 27 in the distribution network, and the reliability evaluation results of the distribution network are shown in Table 4.

[0121]

[0122] Table 4 Reliability index analysis

[0123]

[0124] Table 4 Reliability index analysis

[0125] It can be seen from the data in the above table that with the access of DG to the distribution network, the system reliability has been greatly improved and tended to be stable. The reliability index mainly depends on the load parameters of the fault area. For the non-fault area, the load basically does not lose power after the fault is isolated.

[0126] Nodes 11 and 22 in the distribution network are responsible for a large number of load operations, and a certain proportion of DG is connected to the load group to provide source and load consumption. In addition, when the distribution network feeder or equipment fails, it can share the power supply pressure for the distribution network, forming an island to supply power to the load, which can improve the power supply self-healing ability of the distribution network.

[0127] like Figure 6 As shown, Figure 6It is a schematic diagram of the fault location and division results of the distribution network test system. When two power supply line faults occur in the distribution network, the line power supply is affected. By analyzing the distribution network faults, a fault recovery database is constructed, and a hypersphere clustering model is constructed through secondary kernel cross-integral hypersphere clustering to quickly restore the safe power supply operation of the downstream load.

[0128]

[0129] Table 5 Branch fault decision table

[0130] According to the index sorting, when the faults of branch 11-12 and branch 24-26 occur, the downstream tie switch verifies the flow of the feeder to check whether the node voltage exceeds the limit. For the nodes that exceed the limit, active voltage management will be implemented to control the resources such as adjusting the capacitor bank or transformer ratio in the distribution network. Through the secondary kernel cross-integration hypersphere clustering processing, the hypersphere clustering model is called to effectively match the faults.

[0131] In order to better evaluate the performance of the secondary kernel cross-integration hypersphere clustering proposed in the present invention, multimodal clustering and dynamic clustering with faster current calculation speed and better clustering effect are selected for index comparison analysis, and the performance comparison data is shown in Table 6. As can be seen from Table 6, under the evaluation of the Silhouette Coefficient, DBI, WSS, RI, ARI, FM Index, and iteration time performance indicators, the performance of the secondary kernel cross-integration hypersphere clustering is in a leading state, which fully proves the excellent characteristics of the secondary kernel cross-integration hypersphere clustering.

[0132]

[0133] Table 6 Clustering algorithm performance comparison

[0134] Based on the above analysis, with the continuous improvement of the hypersphere, the hypersphere clustering model can be called quickly and the equipment utilization rate can be improved, which can effectively shorten the power supply recovery time of the distribution network and reduce the load loss caused by faults, thereby improving the overall power supply reliability.

[0135] According to the method described in the above embodiment, this embodiment will be further described from the perspective of a distribution network power supply self-healing system based on a hypersphere clustering algorithm. The distribution network power supply self-healing system based on a hypersphere clustering algorithm can be implemented as an independent entity or integrated into an electronic device, such as a terminal. The terminal may include a mobile phone, a tablet computer, etc.

[0136] See also Figure 3 , Figure 3is a structural diagram of a distribution network power supply self-healing system based on a hypersphere clustering algorithm provided by an embodiment of the present invention, such as Figure 3 As shown, the distribution network power supply self-healing system based on the hypersphere clustering algorithm provided by the embodiment of the present invention includes:

[0137] A database construction module is used to obtain historical operation data of the distribution network, and to construct a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan. The database construction module is used to execute step S1.

[0138] The model building module is used to perform secondary kernel cross-integration hypersphere clustering on the data in the fault recovery database to build a hypersphere clustering model, and determine the clustering maximum range constraint conditions of the hypersphere clustering model and the cross constraint conditions of the hypersphere clustering model. The model building module is used to execute step S2.

[0139] The matching module is used to match the fault performance corresponding to the current fault with the fault data corresponding to each historical fault according to the hypersphere clustering model when a fault occurs in the distribution network. The matching module is used to execute step S3.

[0140] The execution module is used to execute the power supply restoration plan of the distribution network according to the matching results:

[0141] If the match is successful, the power supply recovery plan corresponding to the successfully matched fault manifestation in the fault recovery database is executed;

[0142] If the matching is unsuccessful, a new power supply restoration plan is constructed according to the DG output, energy storage, load level, and line flow data in the current distribution network. Power supply is restored to some or all lines according to the new power supply restoration plan, and the new power supply restoration plan is recorded. The execution module is used to execute step S4.

[0143] The model updating module is used to add the fault manifestation corresponding to the current fault and the new power supply restoration plan to the hypersphere clustering model, and update the hypersphere clustering model. The model updating module is used to execute step S5.

[0144] Also, see Figure 4 , Figure 4 is a schematic diagram of a structure of an electronic device provided by an embodiment of the present invention, such as Figure 4 As shown, the electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned distribution network power supply self-healing method based on the hypersphere clustering algorithm are implemented.

[0145] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions, or by controlling related hardware through instructions, and the instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. To this end, an embodiment of the present invention provides a storage medium, which stores a plurality of instructions, and the instructions can be executed by a processor to implement the steps of the power supply self-healing method for a distribution network based on a hypersphere clustering algorithm provided in the above embodiments.

[0146] Generally speaking, the computer instructions for implementing the method of the present invention may be carried in any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media, except for the signal itself that is temporarily propagating.

[0147] Computer-readable storage media, for example, may be, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or devices, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with a system, device or device for executing instructions.

[0148] One or more programming languages ​​or their combinations can be used to write computer program codes for performing the operation of the present invention. These programming languages ​​include object-oriented programming languages, such as Java, Smalltalk, C++, and also include conventional procedural programming languages, such as C language or similar programming languages. In particular, Python language suitable for neural network calculations and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through various types of networks, including a local area network (LAN) or a wide area network (WAN), or an Internet connection is performed through an Internet service provider.

[0149] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A power supply self-healing method for a distribution network based on a hypersphere clustering algorithm, characterized in that: The power distribution network self-healing method comprises: S1. Obtain historical operation data of the distribution network, and build a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan; S2, performing secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to construct a hypersphere clustering model, and determining the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model; S3. When a fault occurs in the distribution network, the fault performance corresponding to the current fault is matched with the fault data corresponding to each historical fault according to the hypersphere clustering model; S4. Execute the power supply restoration plan of the distribution network according to the matching results: If the match is successful, the power supply recovery plan corresponding to the matched fault data is executed; If the matching fails, a new power supply restoration plan is constructed based on the current DG output, energy storage, load level, and line flow data in the distribution network, and power supply is restored to some or all lines according to the new power supply restoration plan; S5. Add the fault type, fault manifestation, and new power supply restoration plan corresponding to the current fault to the hypersphere clustering model, and update the hypersphere clustering model.

2. The method for self-healing power supply of a distribution network based on a hypersphere clustering algorithm according to claim 1, characterized in that: The power grid historical operation data includes: feeder current, voltage and DG load dynamic information collected by protection relays, circuit breakers, FTUs and DTUs.

3. The method for self-healing power supply in a distribution network based on a hypersphere clustering algorithm according to claim 1, characterized in that: In S4, a new power supply restoration plan is constructed according to the DG output, energy storage, load level, and line flow data in the current distribution network. Restoring power supply to some or all lines according to the new power supply restoration plan includes: S41. Calculate the total load of the fault area: Among them, ω i is the load level, P load,i is the load of feeder i in the fault area, C is the set of feeders in the fault area; S42. According to the total load in the fault area, as well as the DG output and energy storage conditions of the distribution network, determine whether the load in the fault area can be fully or partially powered by DG and energy storage. If so, restore full or partial power supply to the fault area. S43. If the loads in the faulty area cannot be fully powered by DG and energy storage, determine whether the power can be transferred from the adjacent feeder. If so, transfer power to the faulty area through the adjacent feeder. S44. For fault areas where full power supply cannot be restored through DG, energy storage, and adjacent feeder transfer, the line flow in the fault area is calculated, and the loads in the fault area are gradually removed and power supply is restored based on the calculation results.

4. A power supply self-healing method for a distribution network based on a hypersphere clustering algorithm according to claim 1, 2 or 3, characterized in that: The performing of secondary kernel cross-integration hypersphere clustering comprises: S21, obtain a set N of fault data corresponding to n historical faults, where the fault data corresponding to the i-th historical fault is n i ; <h2 style=";text-align:left;direction:ltr">N = {n1, n2, n3,..., n<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr">}; S22, divide the fault data into m categories according to the fault type, each category of fault data corresponds to the same fault type, obtain m fault data clusters corresponding to the m fault types, calculate the center of each fault data cluster to obtain the m fault data cluster centers corresponding to the m fault types, wherein the fault data cluster center corresponding to the jth fault type is c j ; C={c1、c2、c3、...、c m }; S23. For each fault data, extract the fault manifestation corresponding to the fault data, and obtain a set U of fault manifestations corresponding to n historical faults, where the fault manifestation corresponding to the i-th historical fault is u i ; U={u1、u2、u3、...、u n }; S24, classify the fault manifestations into m categories according to the fault type, each category of fault manifestation corresponds to the same fault type, obtain m fault manifestation clusters corresponding to the m fault types, calculate the center of each fault manifestation cluster to obtain the m fault manifestation cluster centers corresponding to the m fault types, wherein the fault manifestation cluster center corresponding to the jth fault type is g j ; G={g1、g2、g3、...、g m }; S25. Update the fault data cluster, fault data cluster center, fault manifestation cluster, and fault manifestation cluster center according to the hypersphere distance between each fault data and the fault data cluster center, and the hypersphere distance between each fault manifestation and the fault manifestation cluster center, so as to bring the center points closer together, perform fitness calculation on the fault manifestation cluster, select the optimal cluster center according to the fitness calculation result, construct the minimum hypersphere according to the optimal cluster center, and obtain the corresponding optimal cluster range.

5. The method for self-healing power supply of a distribution network based on a hypersphere clustering algorithm according to claim 4 is characterized in that: The i-th fault data n in the set of fault data N i The fault data cluster center c corresponding to the jth fault type j The hypersphere distance between is calculated according to the following formula: In the formula, d(n i ,c j ) represents fault data n i and the fault data cluster center c j The hypersphere distance between cov(n i ,c j ) represents fault data n i and the fault data cluster center c j The covariance between σ(n i ),σ(c j ) represent the variance of fault data and fault data cluster center respectively; ε ij Represents the distance calculation error adjustment term, 0≤ε ij ≤1; The i-th fault manifestation u in the set M of fault manifestations i The fault manifestation cluster center corresponding to the jth fault type is g j The hypersphere distance between is calculated according to the following formula: In the formula, d(u i ,g j ) indicates the fault manifestation u i and the fault performance cluster center g j The hypersphere distance between them; cov(u i ,g j ) indicates the fault manifestation u i and the fault performance cluster center g j The covariance between σ(u i ),σ(g j ) represent the fault manifestation and the variance of the fault manifestation cluster center respectively.

6. The method for self-healing power supply of a distribution network based on a hypersphere clustering algorithm according to claim 5, characterized in that: The maximum clustering range constraints for the hypersphere clustering model include: According to the fault data cluster center and fault manifestation cluster center, weights are assigned to the fault manifestations, and the edge range of the intersecting clusters is calculated in the hypersphere to determine the maximum range constraint of the clustering. The calculation formula is: Where V Fj is the maximum range constraint of the clustering of the hypersphere corresponding to the jth fault type; ω i is the weight; N i is the fault manifestation u corresponding to the jth fault type i The total number of i According to u i The frequency setting of the corresponding fault type in the fault recovery database.

7. The method for self-healing power supply of a distribution network based on a hypersphere clustering algorithm according to claim 3 is characterized in that: The cross-constraints that determine the hypersphere clustering model include: According to the maximum range of clustering, cross constraints are set to absorb similar cluster cores. The calculation formula is: In the formula, is the cross constraint condition of the hypersphere corresponding to the jth fault type.

8. A distribution network power supply self-healing system based on a hypersphere clustering algorithm, comprising: A database construction module is used to obtain historical operation data of the distribution network, and to construct a fault recovery database based on the historical operation data of the distribution network. The fault recovery database records the fault data corresponding to each historical fault, and the fault data includes: fault type, fault manifestation, and power supply recovery plan; A model building module is used to perform secondary kernel cross-integral hypersphere clustering on the data in the fault recovery database to build a hypersphere clustering model, and determine the clustering maximum range constraint condition of the hypersphere clustering model and the cross constraint condition of the hypersphere clustering model; A matching module is used to match the fault performance corresponding to the current fault with the fault data corresponding to each historical fault according to the hypersphere clustering model when a fault occurs in the distribution network; The execution module executes the distribution network power supply restoration plan according to the matching results: If the match is successful, the power supply recovery plan corresponding to the matched fault data is executed; If the matching fails, a new power supply restoration plan is constructed based on the current DG output, energy storage, load level, and line flow data in the distribution network, and power supply is restored to some or all lines according to the new power supply restoration plan; The model updating module is used to add the fault type, fault manifestation and new power supply restoration plan corresponding to the current fault to the hypersphere clustering model, and update the hypersphere clustering model.

9. An electronic device, characterized in that: The electronic device comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, and the processor implements the steps in the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 7 are implemented.

Images