An optimal generation method and terminal for low-frequency load shedding rotation table of power grid

Abstract

The present invention discloses an optimal generation method and terminal for a low-frequency load shedding round table of a power grid. According to the feeder load information, a feeder load property index is proposed to realize dynamic analysis of the feeder load property; by sorting the feeder load property index, a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network is generated; then, with the goal of reducing the comprehensive load property index as much as possible and taking into account reducing the load shedding ratio deviation of each partition of the power grid, a genetic algorithm method is applied to optimize and adjust the number of feeders put into low-frequency load shedding rounds in each partition. In this way, accurate tracking of feeder load property changes is achieved, providing a reasonable and reliable basis for the generation of the low-frequency load shedding round table of the power grid, and being able to take into account the comprehensive load property index of the cut load and the optimal load shedding ratio deviation of the partition under the premise of meeting the load shedding ratio requirements of each low-frequency load shedding round for the entire network, thereby reducing the impact of the power load and the load shedding ratio deviation of each partition.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to an optimal generation method and terminal for a power grid low-frequency load shedding rotation table. Background Art

[0002] Grid low-frequency load reduction is the third line of defense for ensuring the safety of the power system. In multiple rounds, when the power system frequency drops to the corresponding threshold, the feeder load of the corresponding round is cut off, which reduces the system's power generation shortage and helps increase the frequency to normal values.

[0003] The power dispatching organization arranges the low-frequency load shedding schedule for the power grid once a year: first, manually collect feeder load information including public transformers, special transformers, and distributed power access capacity, and analyze the nature of the feeder load; then obtain the overall load level of the power grid and the load level of each feeder through the dispatching automation system; finally, arrange the feeders to be cut in each round based on the load required to be cut in each load shedding round and the nature of the feeder load. This arrangement of low-frequency load shedding schedule has the following problems:

[0004] 1. Manually collect feeder load information and analyze feeder load properties. It is necessary to obtain a large amount of distribution network public and private transformers and distributed power supply ledger data offline, and manually classify and count them, which is labor-intensive and inefficient.

[0005] 2. The load of public and private transformers and the output of distributed power sources in the distribution network will change significantly at different time periods, and it is impossible to always maintain a stable power level; the expansion and transformation of the distribution network will lead to changes in the capacity of public and private transformers and distributed power sources connected to the feeder. The above situations may cause changes in the nature of the feeder load, and the feeder load information collected manually can only reflect the feeder load characteristics of a certain period of time. The low-frequency load reduction rotation table arranged on this basis is obviously unable to adapt to the changes in the nature of the feeder load. The result will inevitably lead to an unreasonable feeder removal sequence, such as cutting off the feeder with a high proportion of livelihood load first; cutting off the feeder with high distributed power output first, further widening the power generation gap.

[0006] 3. The arrangement of the low-frequency load shedding rotation table based on the manually collected feeder load information is difficult to take into account the control requirements of the load shedding ratio deviation in different partitions in the power grid, which may cause a large load shedding ratio deviation between different partitions. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide an optimal generation method and terminal for a low-frequency load shedding round table of a power grid, which can take into account the comprehensive load property indicators of the shedding load and the optimal deviation of the load shedding ratio of each partition under the premise of meeting the requirements of the load shedding ratio of each low-frequency load shedding round on the whole network, so as to improve the generation efficiency of the low-frequency load shedding round table, reduce the impact of power load and the deviation of the load shedding ratio of each partition.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] A method for optimally generating a low-frequency load shedding rotation table for a power grid comprises the following steps:

[0010] Determine the feeder load property index according to the feeder load information, wherein the feeder load information includes the public transformer, the special transformer and the load information thereof connected to the feeder, and the distributed power source and the active power output data thereof;

[0011] Sorting the feeder load property indicators in ascending order, and generating a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network according to the sorting result;

[0012] With the goal of minimizing the comprehensive load property index of low-frequency load shedding rounds and minimizing the load shedding ratio deviation of each partition, a genetic algorithm is used to optimize the number of feeders put into low-frequency load shedding rounds in each partition.

[0013] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0014] A terminal for optimally generating a low-frequency load shedding rotation table of a power grid comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step of the above-mentioned method for optimally generating a low-frequency load shedding rotation table of a power grid is implemented.

[0015] The beneficial effects of the present invention are: according to the feeder load information, feeder load property indicators are proposed to realize dynamic analysis of feeder load properties; by sorting the feeder load property indicators, a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network is generated; then, with the goal of reducing the comprehensive load property indicators as much as possible while taking into account reducing the load shedding ratio deviation of each partition of the power grid, the genetic algorithm method is applied to optimize the number of feeders put into low-frequency load shedding rounds in each partition. In this way, accurate tracking of feeder load property changes is achieved, providing a reasonable and reliable basis for the generation of the low-frequency load shedding round table of the power grid, and being able to take into account the comprehensive load property indicators of the cut load and the optimal load shedding ratio deviation of the partition under the premise of meeting the load shedding ratio requirements of each low-frequency load shedding round for the entire network, thereby reducing the impact of power load and the load shedding ratio deviation of each partition. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of an optimal generation method of a power grid low-frequency load reduction round table according to an embodiment of the present invention;

[0017] Figure 2 A schematic diagram of an optimal generation terminal of a power grid low-frequency load reduction round table according to an embodiment of the present invention;

[0018] Figure 3 The figure is a schematic diagram of feeder load properties according to feeder load property indicators in an embodiment of the present invention.

[0019] Figure 4 It is a schematic diagram of an implementation scheme for automatic collection of feeder load information and automatic analysis of feeder load properties in an embodiment of the present invention.

[0020] Figure 5 The present invention is a schematic diagram of a distribution network feeder load information file in E format for interaction between a dispatching control system (DCS) and a distribution automation master station system (DMS) in an embodiment of the present invention.

[0021] Figure 6 The present invention is a schematic diagram of a method for searching for a low-frequency load shedding round adjustment solution to solve the problem of over-limit load shedding ratio in the entire network according to an embodiment of the present invention.

[0022] Figure 7 The figure is a functional graph of a fitness function designed when a genetic method is applied to the adjustment and optimization of the number of low-frequency load shedding feeders in each partition in order to take into account both the comprehensive load property index and the optimal deviation of the partition load shedding ratio in an embodiment of the present invention.

[0023] Figure 8 The schematic diagram of the group-generation individual multi-point crossover operation designed when the genetic method is applied to the adjustment and optimization of the number of low-frequency load shedding feeders in each partition in an embodiment of the present invention in order to take into account the comprehensive load property index and the optimal deviation of the partition load shedding ratio.

[0024] Description of labels:

[0025] 1. An optimal generation terminal for a low-frequency load reduction rotation table of a power grid; 2. A memory; 3. A processor. DETAILED DESCRIPTION

[0026] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0027] Please refer to Figure 1 The embodiment of the present invention provides an optimal generation method of a low-frequency load reduction round table of a power grid, comprising the steps of:

[0028] Determine the feeder load property index according to the feeder load information, wherein the feeder load information includes the public transformer, the special transformer and the load information thereof connected to the feeder, and the distributed power source and the active power output data thereof;

[0029] Sorting the feeder load property indicators in ascending order, and generating a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network according to the sorting result;

[0030] With the goal of minimizing the comprehensive load property index of low-frequency load shedding rounds and minimizing the load shedding ratio deviation of each partition, a genetic algorithm is used to optimize the number of feeders put into low-frequency load shedding rounds in each partition.

[0031] From the above description, it can be seen that the beneficial effects of the present invention are: according to the feeder load information, feeder load property indicators are proposed to realize dynamic analysis of feeder load properties; by sorting the feeder load property indicators, a low-frequency load reduction round table that meets the load shedding ratio constraint of the entire network is generated; then, with the goal of reducing the comprehensive load property indicators as much as possible and taking into account reducing the load shedding ratio deviation of each partition of the power grid, the genetic algorithm method is applied to optimize the number of feeders put into low-frequency load shedding rounds in each partition. In this way, accurate tracking of feeder load property changes is achieved, providing a reasonable and reliable basis for the generation of the low-frequency load shedding round table of the power grid, and being able to take into account the comprehensive load property indicators of the cut load and the optimal load shedding ratio deviation of the partition under the premise of meeting the load shedding ratio requirements of each low-frequency load shedding round for the entire network, thereby reducing the impact of power load and the load shedding ratio deviation of each partition.

[0032] Further, the feeder load property index is determined according to the feeder load information, including:

[0033] The feeder load property index of feeder f is defined as:

[0034]

[0035] Where f represents the feeder number, f=1,2,…,N F ; pt represents the public transformer number, st represents the private transformer number, dg represents the distributed power source number; fe(pt) represents the feeder to which the public transformer pt belongs, fe(st) represents the feeder to which the private transformer st belongs, and fe(dg) represents the feeder to which the distributed power source dg belongs; P(pt,t) represents the load carried by the public transformer pt, P(st,t) represents the load carried by the private transformer st, and P(dg,t) represents the generated power of the distributed power source dg; P(f,t) represents the load of the feeder f.

[0036] From the above description, it can be seen that according to the proportion of public transformer load in the load carried by the feeder and the reverse supply situation caused by the access of distributed power sources, the feeder load can be divided into four types of load feeders: low proportion of livelihood load, medium proportion of livelihood load, high proportion of livelihood load and distributed power feeder, and the corresponding feeder load property indicators are calculated. In this way, the dynamic analysis of the feeder load properties is realized.

[0037] Furthermore, a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network is generated according to the sorting result, including:

[0038] A low-frequency load shedding round table is generated according to the sorting results. If the low-frequency load shedding round table does not meet the load shedding ratio constraint of the entire network, a low-frequency load shedding round adjustment scheme that meets the load shedding ratio constraint of the entire network is searched with the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, and the low-frequency load shedding round adjustment scheme is used to adjust the low-frequency load shedding round table.

[0039] From the above description, it can be seen that by sorting the feeder load property indicators, the rapid generation of the low-frequency load shedding round table is achieved; then, in response to the possible problem of over-limit of the load shedding ratio of the entire network, with the goal of optimizing the comprehensive load property indicators of the load shedding round, the adjustment strategy of "sorting by feeder load property indicators, withdrawing feeders that have been put into load shedding rounds from back to front and adding feeders that have been put into load shedding rounds from front to back" is adopted. In this way, it is possible to ensure the generation of a low-frequency load shedding round table that meets the load shedding ratio constraints of the entire network, while avoiding over-adjustment.

[0040] Furthermore, an underfrequency load shedding round table is generated according to the sorting result, including:

[0041] Find the smallest integer n0:

[0042] Where P(n0,r,t) represents the total load of the first n0 feeders in the rth round, P ∑ (t) represents the total load of the entire network, Indicates the lower limit of the load shedding ratio of the rth wheel;

[0043] Generate an underfrequency load reduction round table F according to the minimum integer 2.1 ={f|1≤ord(f,r,t)≤n0}, where ord(f,r,t) represents the ranking of feeder load property indicators of feeder f in the rth round.

[0044] It can be seen from the above description that the low-frequency load shedding rotation table is generated by sorting the feeder load property indicators, which can improve the efficiency of generating the low-frequency load shedding rotation table.

[0045] Furthermore, with the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, a low-frequency load shedding round adjustment scheme that meets the load shedding ratio constraint of the entire network is searched, including:

[0046] Define the comprehensive load property index LNI(r,t) of the rth round at time t:

[0047]

[0048] Where rnd(f,t) represents the load shedding round number of feeder f;

[0049] With the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, according to the sorting results of the feeder load property index, when it is necessary to adjust the load shedding ratio downward, the feeders that have been put into the low-frequency load shedding round are withdrawn from the back to the front; when it is necessary to adjust the load shedding ratio upward, new feeders that are put into the load shedding round are added from the front to the back, so as to search for a low-frequency load shedding round adjustment plan that meets the load shedding ratio constraints of the entire network, record the search process, and detect repeated plans.

[0050] It can be seen from the above description that by recording the search process and detecting repeated solutions, the problem of search failure caused by repeated appearance of the same feeder combination solution can be effectively prevented.

[0051] Furthermore, with the goal of minimizing the comprehensive load property index of the low-frequency load shedding round and minimizing the load shedding ratio deviation of each partition, a genetic algorithm is used to optimize the number of feeders in each partition for the low-frequency load shedding round, including:

[0052] Establishing a fitness function that takes into account the load shedding ratio constraint of the entire network, the comprehensive load property index, and the load shedding ratio deviation index optimization of each partition;

[0053] The fitness assignment calculation is performed for different value intervals of the comprehensive load property index and the partition load shedding ratio deviation index, and the number of feeders put into low-frequency load shedding rounds in each partition is optimized and adjusted according to the fitness calculation results.

[0054] From the above description, it can be seen that the design comprehensively considers the load shedding ratio constraint of the whole network and takes into account the optimization of the comprehensive load property index and the partition load shedding ratio deviation index. The fitness assignment calculation is performed for different value intervals of the comprehensive load property index and the partition load shedding ratio deviation index. It can not only effectively distinguish individual differences, but also control the value range of fitness, which can effectively ensure the rationality of the individual fitness evaluation.

[0055] Furthermore, a genetic algorithm is used to optimize and adjust the number of feeders in each zone for low-frequency load shedding rounds, including:

[0056] A multi-point crossover operation and a mutation operation based on mutation probability are set in the genetic algorithm, and new individuals generated by the multi-point crossover operation and the mutation operation are added to the next generation in the iterative calculation of the genetic algorithm.

[0057] From the above description, we can see that designing multi-point crossover operations and mutation operations based on mutation probability, and adding the new individuals generated by crossover and mutation operations to the next generation in iterative calculations can ensure the diversity of individuals in the generation group and achieve the goal of improving the convergence speed of the algorithm.

[0058] Please refer to Figure 2 Another embodiment of the present invention provides an optimal generation terminal for a low-frequency load shedding rotation table of a power grid, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of the above-mentioned optimal generation method for a low-frequency load shedding rotation table of a power grid when executing the computer program.

[0059] The above-mentioned optimal generation method and terminal of the low-frequency load shedding round table of the power grid of the present invention are suitable for taking into account the comprehensive load property index of the shedding load and the optimal deviation of the load shedding ratio of each partition under the premise of meeting the requirements of the load shedding ratio of each low-frequency load shedding round on the whole network, so as to improve the generation efficiency of the low-frequency load shedding round table, reduce the impact of power load and the deviation of the load shedding ratio of each partition. The following is an explanation through a specific implementation method:

[0060] Embodiment 1

[0061] Please refer to Figure 1 , a method for optimally generating a low-frequency load shedding rotation table of a power grid, comprising the steps of:

[0062] 1. Determine the feeder load property index according to the feeder load information, wherein the feeder load information includes the public transformer, special transformer and load information connected to the feeder, as well as the distributed power source and its active output data.

[0063] Specifically, the low-frequency load shedding of the power grid is achieved by cutting off the 10 / 35kV feeder load. In order to reduce the impact of the low-frequency load shedding function on the electricity load for people’s livelihood, it is necessary to accurately grasp the nature of the feeder load. The feeder load nature mentioned here refers to the proportion of the public transformer and special transformer load of the distribution network carried by the feeder. Figure 3 As shown in the figure, according to the proportion of public transformer load in the load carried by the feeder and the reverse supply situation caused by the access of distributed power sources, the feeder load can be divided into four types of load feeders: low proportion of livelihood load, medium proportion of livelihood load, high proportion of livelihood load and distributed power feeder.

[0064] Figure 3 Where LNI(f,t) represents the load property index of feeder f at time t, which is defined as follows:

[0065]

[0066] Where f represents the feeder number, f=1,2,…,NF ; pt represents the public transformer number, st represents the private transformer number, dg represents the distributed power source number; fe(pt) represents the feeder to which the public transformer pt belongs, fe(st) represents the feeder to which the private transformer st belongs, and fe(dg) represents the feeder to which the distributed power source dg belongs; P(pt,t) represents the load carried by the public transformer pt, P(st,t) represents the load carried by the private transformer st, and P(dg,t) represents the generated power of the distributed power source dg; P(f,t) represents the load of the feeder f.

[0067] When P(f, t) < 0, the feeder is connected to the distributed power source, which leads to reverse power supply to the grid, which is beneficial to frequency stability. Therefore, it cannot be included in the low-frequency load shedding cycle. Therefore, the establishment of the low-frequency load shedding cycle table only considers the feeders with P(f, t) > 0.

[0068] In order to calculate LNI (f, t), it is necessary to accurately grasp the load of the public transformer and special transformer carried by the feeder in the distribution network. In this embodiment, the distribution network topology of the distribution automation master station system and the feeder public and special transformer load collection function of the power consumption information collection system are used to realize the statistics of the feeder load property index through horizontal interaction with the dispatching control system, which is described in detail as follows:

[0069] 1.1、Acquisition of public and private transformer load data in distribution network

[0070] The Electricity Information Collection System (EIAS) can collect the load data of public and private transformers in the distribution network and the output data of distributed generation (DG) by deploying load management terminals (PLMTU). The data density can reach 1 point / 15 minutes, which is sufficient to support the needs of feeder load property analysis and statistics.

[0071] 1.2、Acquisition of information about public and private transformers and distributed power sources connected to the feeder

[0072] The distribution automation master station system (DMS) has a complete medium-voltage distribution network topology wiring analysis function. Through interaction with DMS, it can obtain real-time information on public and private transformers and distributed power sources connected to the feeder.

[0073] 1.3、Implementation scheme of feeder load property analysis function, such as Figure 4 As shown:

[0074] 1) The power consumption information collection system provides 1 point / 15 minutes of distribution network public and private transformer load data and distributed generation (DG) output data, which are sent to the DMS system in E format files;

[0075] 2) The distribution network topology wiring analysis program of the DMS system provides the feeder information of each public transformer, special transformer and distributed power source;

[0076] 3) The feeder load information summary program of the DMS system summarizes the load data of the public and private transformers of the distribution network from the EIAS system, the output data of the distributed generation (DG), and the feeder information of the public transformers, private transformers and distributed generation provided by its own topology wiring analysis program to form a Figure 5 The distribution network feeder load information file in the E format shown is submitted to the feeder load property analysis program of the control system (DCS);

[0077] 4) The feeder load property analysis program of the control system (DCS) calculates the load property index of each feeder and writes it into the power grid model database of the DCS to complete the maintenance of the "load property index" attribute of the 10 / 35kV feeder load;

[0078] 5) The low-frequency load shedding rotation scheduling program of the control system (DCS) obtains the "load property index" attribute of the 10 / 35kV feeder load by accessing the power grid model database, which is used to generate the low-frequency load shedding rotation table.

[0079] 2. Set the objectives and constraints for generating under-frequency load shedding rounds.

[0080] Specifically, the goals of generating low-frequency load shedding rounds are:

[0081] 1) Reduce the impact on people's livelihood electricity load and define the comprehensive load index LNI(r,t) of the rth round at time t:

[0082]

[0083] Where rnd(f,t) represents the load shedding round number of feeder f.

[0084] The purpose of compilation should be to make the LNI(r,t) indicator as small as possible.

[0085] 2) Reduce the deviation of the load shedding ratio p(a, r, t) of each partition, that is, make the overall standard deviation σ(r, t) of the load shedding ratio of each partition as small as possible, a represents the partition number, a=1,2,…,N A .

[0086] in, P ∑ (a, t) represents the load of partition a, P(a, r, t) represents the load of partition a included in the rth round, area(f) indicates the grid partition number where feeder f is located.

[0087] The constraints set are:

[0088] In the formula, p(r,t) represents the proportion of load shedding in the rth round of operation to the total load of the entire network. represents the lower limit of the load shedding ratio in the rth round, It indicates the upper limit of the load shedding ratio in the rth round, which is 1.05 times of the lower limit.

[0089] 3. Based on the comprehensive load property index and the load shedding ratio deviation of each partition, an optimal low-frequency load shedding rotation table is generated.

[0090] 3.1. Initialization.

[0091] 1) Let in(f,r,t)=0(f=1,2,...,N F , r=1,2,...,8), in(f,r,t) indicates whether feeder f is included in the rth round, when in(f,r,t)=1, it indicates yes, when in(f,r,t)=0, it indicates no;

[0092] 2) Let r = 1;

[0093] 3) Let F r =F, Among them, F r represents the set of feeders included in the rth round of scheduling, F represents the set of feeders in the entire network, represents the set of feeders in zone a included in the rth round of scheduling, F a Represents the feeder set of partition a.

[0094] 3.2. Sort the feeder f by its LNI index ord(f,r,t) in the rth round of scheduling to find the feeder combination that satisfies the constraint condition st.

[0095] 3.2.1. There exists a minimum integer n0 that satisfies the following conditions:

[0096]

[0097] Where P ∑ (t) represents the total load of the whole network, P ∑ (t) = ∑P Σ (a, t); P(n0, r, t) represents the sum of the feeder loads before the LNI index ranking in the rth round of scheduling.

[0098] Let F 2.1 ={f1≤ord(f,r,t)≤n0}. The values ​​of in(f,r,t) of each feeder are shown in Table 1:

[0099] Table 1 Feeder in(f,r,t) value table

[0100]

[0101] like If yes, go to step 3.3; otherwise, go to step 3.2.2.

[0102] 3.2.2, in F r Search for the feeder subset F in 2.2 .

[0103]

[0104] Although for the feeder subset F 2.2 The search is based on solving the problem of load shedding ratio exceeding the upper limit of the whole network. However, as the search progresses, it may be necessary to adjust the load shedding ratio downward (i.e., the so-called "over-regulation"). In order to solve this problem, the following is designed. Figure 6 The search method shown is described in detail below:

[0105] 3.2.2.1. Define the sorting function for the feeder set F r Sort.

[0106] The feeders with in(f,r,t) values ​​of "1" and "0" are sorted from small to large according to the LNI index, which are referred to as "1" sorting and "0 sorting" below. The following sorting functions O1(f) and O0(f) are defined.

[0107] 1) O1(f):

[0108] When in(f,r,t) takes the value of "1", O1(f) takes the value of the feeder f in the "1" ranking; when in(f,r,t) takes the value of "0", O1(f) takes the value of 0. The maximum ranking of the "1" ranking is O 1,max .

[0109] 2) O0(f):

[0110] When in(f,r,t) takes the value of “0”, O0(f) takes the value of the feeder f in the “0” sorting; when in(f,r,t) takes the value of “1”, O0(f) takes the value of 0.

[0111] According to the method described in the comments, sort the feeder LNI indicators by "1" and "0", and determine the O1(f), O0(f) and O 1,max .

[0112] 3.2.2.2. Define and initialize variables m and n.

[0113] The function of the variable m is to determine the feeder with the largest ranking for load shedding ratio adjustment in the "1" ranking when the load shedding ratio needs to be adjusted downward, and initialize its value to 0.

[0114] The function of the variable n is: when the load shedding ratio needs to be adjusted upward, the feeder with the smallest ranking for load shedding ratio adjustment is determined in the "0" ranking, and its value is initialized to 0.

[0115] 3.2.2.3. Search feeder subset F by adopting the policy of "giving priority to feeders with small LNI index when the load shedding ratio needs to be adjusted upward, and giving priority to feeders with large LNI index when the load shedding ratio needs to be adjusted downward" 2.2 .

[0116] In order to prevent the search process from being invalid due to the repeated appearance of the same feeder combination scheme, each new combination scheme that appears in the search process is first recorded. In the subsequent search process, when it is found that the found combination scheme is repeated with the recorded scheme, the search is performed again by changing the values ​​of m and n.

[0117] 1) When the combination ratio needs to be adjusted downward

[0118] First search for the feeder with the largest rank in the current "1" sort. Obviously, the value of O1(f) of this feeder is O. 1,max -m, change the value of in(f,r,t) of the feeder to "0", and check whether the combination scheme obtained at this time has appeared in the previous search process: if the scheme obtained in the previous search process appears repeatedly, increase the value of the variable m by 1, and continue to search for other feeders whose LNI index is ranked before this feeder; otherwise, keep the new combination scheme obtained in this search.

[0119] 2) When the combination ratio needs to be adjusted upward

[0120] First search for the feeder with the smallest rank in the current "0" ranking. Obviously, the value of O0(f) of this feeder is n+1. Modify the value of in(f,r,t) of this feeder to "1" and check whether the combination scheme obtained at this time has appeared in the previous search process: if the scheme obtained in the previous search process appears repeatedly, add 1 to the value of variable n and continue to search for other feeders whose LNI index ranking is after this feeder; otherwise, keep the new combination scheme obtained in this search.

[0121] 3.2.2.4、Calculate the values ​​of O1(f), O0(f) and O 1,max .

[0122] 3.2.2.5. Determine whether the load shedding ratio of the whole network of the new load shedding combination scheme meets the requirements. If the load shedding ratio of the whole network still does not meet the requirements, return to step 3.2.2.3 and continue searching for a new round of load shedding combination schemes.

[0123] After step 3.2.2, the values ​​of iin(f,r,t) of each feeder are updated as shown in Table 2:

[0124] Table 2 Feeder in(f,r,t) value update table

[0125]

[0126] 3.3、Optimize the deviation of the load shedding ratio of each partition

[0127] From step 3.2, a feeder combination scheme for the rth round of load shedding can be obtained. From this scheme, the load shedding ratio p(a, r, t) of each partition and the load shedding ratio p(r, t) of the entire network can be known. This step will adjust the feeders included in the rth round of each partition in the scheme to make σ(r, t) as small as possible.

[0128] 3.3.1. Similar to step 3.2.2.1, define the ranking function for the feeder set F r a Sort.

[0129] The feeders with in(f,r,t) values ​​of "1" and "0" are sorted from small to large according to the LNI index, which are referred to as the "1" sorting of partition a and the "0" sorting of partition a. The following sorting function O is defined as 1,a (f) and O 0,a (f).

[0130] 1)O 1,a (f), a=1,2…,N A :

[0131] When in(f,r,t) takes the value of "1", O 1,a The value of (f) is the rank of feeder f in the “1” ranking; when in(f,r,t) is “0”, O 1,a The value of (f) is 0. The maximum order of "1" is 0. 1,a,max .

[0132] 2)O 0,a (f), a=1,2…,N A :

[0133] When in(f,r,t) takes the value of "0", O 0,a The value of (f) is the rank of feeder f in the “0” ranking; when in(f,r,t) is “1”, O 0,a The value of (f) is 0.

[0134] 3.3.2. Determine the maximum number of feeders to be adjusted in each zone

[0135] 1) Yes According to the LNI index of feeder f in the rth round arrangement, orda(f,r,t), the strategy of "withdrawing feeders from the load shedding round from the back to the front" can be adopted to determine the maximum number of feeders to be reduced in zone a. The following conditions must be met:

[0136]

[0137] Obviously, for this type of partition, the reduction in the number of feeders should meet the following requirements: Where N ad (a,r,t) represents the number of adjusted feeders in zone a included in round r.

[0138] 2) Yes According to orda(f,r,t), the strategy of "putting feeders into load shedding rounds from front to back" can be adopted to determine the maximum number of feeders to be added in zone a. The following conditions must be met:

[0139]

[0140] Obviously, for this type of partition, the number of feeders to be increased should meet the following requirements:

[0141] 3) Yes The partition can either increase or decrease the feeders included in the rth round. In this case, the maximum number of feeders to be increased needs to be calculated separately. and the maximum number of feeders to be reduced The method for determining is consistent with the conditions in 1) and 2) and will not be repeated here.

[0142] If the first category is partitioned and 2) partition If the feeder number ranges for the above three types of partitions are "0", they can be written uniformly as follows:

[0143] 3.3.3 Method for determining the number of feeders to be adjusted in each partition based on GA (Genetic Algorithm) method

[0144] 1) The feeder adjustment scheme will be determined by the combination of the number of adjusted feeders in each partition. Different adjustment schemes are regarded as different GA group generation individuals, each of which is composed of N A The number of adjusted feeders corresponding to different partitions is composed of. Assume that the size of the GA group generation is N G , then the individuals in each generation can be expressed as:

[0145]

[0146] N i,j,ad (k,r,t) represents the number of adjusted feeders in partition k of individual j in the i-th generation. According to the description of step 3.3.2, it is obvious that: 1≤k≤N A , 1≤j≤N G , i=1,2,…

[0147] 2) Design of fitness function

[0148] According to the low-frequency load shedding round compilation objectives and constraints described in step 2, the following GA method individual fitness function is designed: J = J p ·(J LNI +J σ ), the fitness function is given by J p , J LNI , J σ Three components.

[0149] J p : is a penalty term, which reflects the deviation between the load shedding ratio of the entire network after feeder adjustment and the required load shedding ratio range. When J p The denominator of the term will always be equal to The value of the fitness function depends on the other two incentive items; on the contrary, if the adjusted load shedding ratio of the entire network is not in the interval In the range, J p The denominator of the term will be greater than And the load shedding ratio of the entire network deviates from the range The farther away, the larger the denominator, so J p The smaller the value is. p The item is used to control the feeder adjustment scheme generated by the GA method to meet the requirements of the load shedding ratio of the entire network.

[0150]

[0151] J LNI , J σ : is an incentive term, which reflects the overall property index LNI(r,t) of the load shedding in the rth round and the deviation σ(r,t) of the load shedding ratio in each partition. The calculation formula is as follows:

[0152]

[0153]

[0154] J LNI , J σThe design of takes into account the value characteristics of LNI and σ. Under normal circumstances, LNI is mostly above 10%, and it is less likely to be below 10%; while σ is mostly above 1%, and it is less likely to be below 1%; therefore, it is considered to combine the power function (power is a negative number) with the linear function, and use the power function to significantly distinguish the superiority and inferiority of individuals in the range of values ​​where LNI and σ appear more frequently. In the range of values ​​where LNI and σ appear less frequently, the linear function is used, firstly, to continue to compare the superiority and inferiority of individuals; secondly, it can effectively limit J LNI , J σ The upper limit of the value of J LNI or J σ If the value of is too large, it will affect the overall evaluation of the individual; thirdly, J LNI and J σ The values ​​of are controlled at the same order of magnitude, which is conducive to the overall evaluation of the individual advantages and disadvantages. Figure 7 J is given in LNI , J σ The function graph of .

[0155] 3) Initial group generation method

[0156] The method of generating the initial group generation G0(t) is as follows:

[0157]

[0158] Where rnd(j,k,t) is a random number in the interval [0,1], 1≤k≤N A , 1≤j≤N G .

[0159] In the method of generating the initial group generation, in each individual of the initial group generation, each partition will obtain a random adjustment feeder number between its upper limit of increase and decrease. It should be noted that and When N i,j,ad When (k,r,t) takes a positive value, it means that the k partition increases by N. i,j,ad (k,r,t) feeders are included in the rth round of load shedding. i,j,ad When (k,r,t) takes a negative value, it means that the k partition is reduced by |N i,j,ad (k,r,t)| feeders included in the rth round of load shedding.

[0160] The following introduces the design of individual selection, crossover and mutation operations in the GA method. For the convenience of description, the following symbols are defined:

[0161] ①Remember G i The individual in (t) is ind i,j ,

[0162] ② For Gi The individuals in (t) perform the selection operation to generate the mating pool S for the crossover operation i (t), and the individual is denoted as ind S,i,j ;

[0163] ③ For S i The individuals in (t) perform a crossover operation to generate a mutation pool C for mutation operation. i (t), and the individual is denoted as ind C,i,j ;

[0164] ④To C i The individuals in (t) perform mutation operations to generate the individual set M after mutation operations. i (t), and the individual is denoted as ind M,i,j ;

[0165] ⑤M i The individuals in (t) constitute a new group G i+1 (t).

[0166] 4) Design of the method for selecting individuals in the group generation in the GA method

[0167] Perform selection operations on individuals of the previous generation according to the fitness ratio, ind S,i,j Number of copies:

[0168]

[0169] In order to prevent i,j After rounding, this results in The following sequential selection and copying method is adopted:

[0170] ① For G i The individuals in (t) are sorted from large to small according to their fitness, and the individuals with the highest ranking are selected for replication first, and the number of replications is allocated according to the above formula;

[0171] ② Each time an individual is selected to perform a replication operation, the number of copies allocated is recorded;

[0172] ③ When selecting individuals of a certain order to replicate, if the number of copies of the individuals of this order is found to be "0" after rounding, in order to prevent the number of individual copies from failing to reach the group generation scale, the remaining number of copies will be allocated to the individuals of this order.

[0173] 5) Design of the group-generation individual crossover method in the GA method

[0174] Let individual ind S,i,u and ind S,i,v as follows:

[0175] Ni,u,ad (1,r,t),…,N i,u,ad (α,r,t),…,N i,u,ad (β,r,t),...,N i,u,ad (γ,r,t),...,N i,u,ad (N A ,r,t)

[0176] N i,v,ad (1,r,t),...,N i,v,ad (α,r,t),…,N i,v,ad (β,r,t),…,N i,v,ad (γ,r,t),…,N i,v,ad (N A ,r,t)

[0177] The multi-point crossover method is used to increase the individual changes brought about by crossover. Three crossover points α, β, and γ are introduced respectively. Figure 8 Individual crosses shown.

[0178] Let the crossover probability be p c , that is, in S i (t) Select p c ·N G Individuals perform crossover operations. i Each individual in (t) generates a random number p to determine whether to participate in the crossover operation. S,i,j , p S,i,j ∈[0,1]. When p S,i,j ≤p c , then individual ind S,i,j Participate in the crossover operation. After executing the above crossover operation, C is generated i (t).

[0179] 6) Design of the method of individual variation in the GA method

[0180] C i (t) Perform mutation operation, let the mutation probability be p m , then C i The number of individuals that mutate in (t) is p m ·N G ·N A . For C i Each bit of each individual in (t) (corresponding to the number of adjusted feeders in a certain partition) generates a random number p for whether to participate in the mutation operation C,i,j,k , p C,i,j,k ∈[0,1]. When p C,i,j,k ≤p m , then individual ind C,i,jThe kth position of is mutated (i.e., the number of adjusted feeders of partition k is randomly changed, and the operation is the same as the initial group generation method described in 3). After executing the above mutation operation, .

[0181] 3.4. Determine whether to continue the next round of scheduling

[0182] Step 3.3 can determine the feeder set F(r) included in the rth round and the feeder set F(r,a) included in the rth round in each partition.

[0183] 3.4.1. If r=8, then the low-frequency load shedding round arrangement for the current time t is terminated; otherwise, the next load shedding round arrangement for the current time t is continued, and step 3.4.2 is entered.

[0184] 3.4.2. Modify the feeder set and feeder property index ranking for the next round of scheduling:

[0185] 1) Let F r+1 =F r -F(r),

[0186] 2) Let r = r + 1

[0187] 3) Recalculate F according to LNI(f,t) r , Sort the feeders in the feeder and determine the ord(f,r,t), orda(f,r,t) of each feeder

[0188] 4) Return to step 3.2 and start planning a new load shedding round.

[0189] In this way, the present embodiment can realize the automatic collection of load information of low-frequency load shedding feeders and the automatic analysis and statistics of feeder load properties, avoiding the heavy workload brought by manual collection, analysis and calculation, and improving the efficiency of scheduling low-frequency load shedding rounds; it can realize accurate tracking of changes in feeder load properties, and provide a reasonable and reliable basis for the scheduling of low-frequency load shedding rounds of the power grid; and it can take into account the comprehensive load property indicators of the cut load and the optimal deviation of the partitioned load shedding ratio on the premise of meeting the requirements of each low-frequency load shedding round for the load shedding ratio of the entire network.

[0190] Embodiment 2

[0191] Please refer to Figure 2 A terminal 1 for optimally generating a low-frequency load shedding rotation table of a power grid comprises a memory 2, a processor 3 and a computer program stored in the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, each step of an optimally generating method for a low-frequency load shedding rotation table of a power grid in embodiment 1 is implemented.

[0192] In summary, the present invention provides an optimal generation method and terminal for a low-frequency load reduction rotation table of a power grid, proposes a feeder load property index, and realizes a dynamic analysis of the feeder load property. Through horizontal interaction with the distribution automation master station system (DMS) and the electricity information acquisition system (EIAS), on the one hand, the public and private transformers and distributed power source information connected to the feeder can be automatically obtained; on the other hand, the public and private transformer load data and distributed power source output data can be automatically obtained. On this basis, a feeder load property analysis program is developed to realize a dynamic analysis of the feeder load property, thereby providing a reliable basis for the arrangement of the low-frequency load reduction rotation table.

[0193] A comprehensive load property index characterizing the characteristics of the load shedding in low-frequency load shedding rounds is also proposed. First, by sorting the feeder load property index, the low-frequency load shedding round table is quickly generated; then, in view of the possible problem of over-limit load shedding ratio in the entire network, with the goal of optimizing the comprehensive load property index of the load shedding round, an adjustment strategy of "sorting by feeder load property index, withdrawing feeders that have been put into load shedding rounds from back to front and adding feeders that have been put into load shedding rounds from front to back" is adopted to search for low-frequency load shedding round adjustment plans that meet the requirements of the load shedding ratio of the entire network. By adopting the method of "search process recording and repeated plan detection", the search failure problem caused by the repeated appearance of the same feeder combination plan is effectively prevented.

[0194] In addition, the GA method is applied to optimize the number of feeders put into low-frequency load shedding rounds in each partition, with the goal of reducing the comprehensive load property index as much as possible and taking into account the reduction of the load shedding ratio deviation of each partition of the power grid. A fitness function is designed that comprehensively considers the load shedding ratio constraint of the entire network and takes into account the optimization of the comprehensive load property index and the partition load shedding ratio deviation index. The fitness assignment calculation is performed for different value intervals of the comprehensive load property index and the partition load shedding ratio deviation index, which can effectively distinguish individual differences and control the value range of fitness, and can effectively ensure the rationality of individual fitness evaluation; multi-point crossover operation and mutation operation based on mutation probability are designed, and the new individuals generated by crossover and mutation operations are added to the next group generation in the iterative calculation, which can ensure the diversity of individuals in the group generation and achieve the goal of improving the convergence speed of the algorithm.

[0195] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for optimally generating a low-frequency load shedding rotation table for a power grid, characterized in that: Includes steps: Determine the feeder load property index according to the feeder load information, wherein the feeder load information includes the public transformer, the special transformer and the load information thereof connected to the feeder, and the distributed power source and the active power output data thereof; Sorting the feeder load property indicators in ascending order, and generating a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network according to the sorting result; With the goal of minimizing the comprehensive load property index of the low-frequency load shedding round and minimizing the load shedding ratio deviation of each partition, a genetic algorithm is used to optimize the number of feeders in each partition for low-frequency load shedding rounds. Determine feeder load property indicators based on feeder load information, including: Define Feeder f The feeder load property index is: In the formula, Indicates the feeder number, ; Indicates the public variable number, Indicates the special transformer number. Indicates the distributed power source number; Indicates public change The feeder to which it belongs, Indicates special change The feeder to which it belongs, Distributed power supply The feeder to which it belongs; Indicates public change The load carried, Indicates special change The load carried, Distributed power supply The power generation capacity; Indicates feeder Load; With the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, search for a low-frequency load shedding round adjustment plan that meets the load shedding ratio constraint of the entire network, including: definition Moment Comprehensive load property index of wheel : In the formula, Indicates feeder The load shedding round number; With the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, according to the sorting results of the feeder load property index, when it is necessary to adjust the load shedding ratio downward, the feeders that have been put into the low-frequency load shedding round are withdrawn from the back to the front; when it is necessary to adjust the load shedding ratio upward, new feeders that are put into the load shedding round are added from the front to the back, so as to search for a low-frequency load shedding round adjustment plan that meets the load shedding ratio constraints of the entire network, record the search process, and detect repeated plans.

2. The optimal generation method of a power grid low-frequency load reduction rotation table according to claim 1 is characterized in that: According to the sorting results, a low-frequency load shedding round table that meets the load shedding ratio constraint of the entire network is generated, including: A low-frequency load shedding round table is generated according to the sorting results. If the low-frequency load shedding round table does not meet the load shedding ratio constraint of the entire network, a low-frequency load shedding round adjustment scheme that meets the load shedding ratio constraint of the entire network is searched with the goal of minimizing the comprehensive load property index of the low-frequency load shedding round, and the low-frequency load shedding round adjustment scheme is used to adjust the low-frequency load shedding round table.

3. The optimal generation method of a power grid low-frequency load reduction rotation table according to claim 2 is characterized in that: Generate an underfrequency load reduction round table based on the sorting results, including: Find the smallest integer : ; In the formula, Indicates Before sorting in round The total feeder load is Represents the total load of the entire network. Indicates Lower limit of wheel shear load ratio; Generate an underfrequency load reduction round table according to the minimum integer , where Indicates feeder In the The feeder load property index of the wheel is ranked.

4. The optimal generation method of a power grid low-frequency load reduction rotation table according to claim 1 is characterized in that: With the goal of minimizing the comprehensive load property index of the low-frequency load shedding round and minimizing the load shedding ratio deviation of each partition, a genetic algorithm is used to optimize the number of feeders put into the low-frequency load shedding round in each partition, including: Establishing a fitness function that takes into account the load shedding ratio constraint of the entire network, the comprehensive load property index, and the load shedding ratio deviation index optimization of each partition; The fitness assignment calculation is performed for different value intervals of the comprehensive load property index and the partition load shedding ratio deviation index, and the number of feeders put into low-frequency load shedding rounds in each partition is optimized and adjusted according to the fitness calculation results.

5. The optimal generation method of a power grid low-frequency load reduction rotation table according to claim 1 is characterized in that: Genetic algorithms are used to optimize the number of feeders in each zone for low-frequency load shedding rounds, including: A multi-point crossover operation and a mutation operation based on mutation probability are set in the genetic algorithm, and new individuals generated by the multi-point crossover operation and the mutation operation are added to the next generation in the iterative calculation of the genetic algorithm.

6. A terminal for optimally generating a low-frequency load shedding rotation table of a power grid, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, each step of the method for optimally generating a power grid low-frequency load reduction rotation table as described in any one of claims 1 to 5 is implemented.

Images