Power distribution network toughness optimization method based on improved cuckoo algorithm

Abstract

The invention discloses a power distribution network toughness optimization method based on an improved cuckoo algorithm, and relates to power system toughness analysis operation, and the method comprises the steps: obtaining the historical data of a power grid of a to-be-planned region, and building a power working condition scene; based on the element model and the addressing and sizing planning model, first configuration information is confirmed according to the dynamic characteristics and the total cost of the optical storage load in each electric power working condition scene; on the basis of an island division operation model, according to the first configuration information, determining an island division form in the to-be-planned region under an extreme condition; the first configuration information and the island division form in the to-be-planned region are optimized by using an improved cuckoo algorithm, and the power distribution network toughness of the to-be-planned region is improved. The method can effectively optimize the toughness of the power grid coping with the extreme climate, thereby improving the safety and practical value of the power grid.

Description

TECHNICAL FIELD

[0001] The present application generally relates to the technical field of power system resilience analysis operation, and particularly relates to a power distribution network resilience optimization method based on an improved cuckoo algorithm. BACKGROUND

[0002] Existing research shows that power system blackouts caused by extreme events expose the problem of insufficient resilience of the power grid, especially in the field of distribution networks. Because the automation level of the distribution network is low, the redundancy is insufficient, and it is directly connected to the terminal load, its disaster resistance is weak, which will seriously affect the social and economic operation and energy security.

[0003] The current common means for the recovery ability of the power system in extreme events is to evaluate through a probability model, and at the same time, a method for establishing a resilience evaluation index system is used for research. However, such a method often fails to reproduce the uncertainty of extreme events and cannot accurately make reasonable and comprehensive power planning for the planning area, resulting in ineffective improvement of the resilience of the power grid obtained by planning. SUMMARY

[0004] In view of the above defects or deficiencies in the prior art, it is desirable to provide a power distribution network resilience optimization method based on an improved cuckoo algorithm.

[0005] The present application provides a power distribution network resilience optimization method based on an improved cuckoo algorithm, comprising: obtaining historical data of a power grid in a region to be planned, and establishing power working condition scenarios; based on a component model and a site selection and capacity planning model, according to the dynamic characteristics and total cost of the light storage load under each power working condition scenario, confirming first configuration information; the component model is used to simulate the dynamic characteristics of the light storage load of the power grid in the region to be planned; the first configuration information is the arrangement position and capacity of each power system device inside the region to be planned; based on an island division operation model, according to the first configuration information, confirming the island division form under extreme conditions in the region to be planned; using an improved cuckoo algorithm to optimize the first configuration information and the island division form in the region to be planned, so as to improve the resilience of the power distribution network in the region to be planned.

[0006] According to the technical scheme provided by the present application, the historical data of the power grid in the region to be planned is obtained, and the power working condition scenarios are established, comprising: obtaining historical data of light intensity and load level in the region to be planned, and establishing a probability model followed by the two; generating a plurality of light intensity and load data from the probability model, and calculating the photovoltaic output efficiency and load power based on a plurality of the light intensity and load data; At least one of the power operating condition scenarios is established based on the photovoltaic output efficiency and the load power.

[0007] According to the technical scheme provided in the application, based on the element model and the site selection and capacity planning model, the first configuration information is confirmed according to the dynamic characteristics and total cost of the photovoltaic storage load under each of the power operating condition scenarios, including: The element model is constructed, and the element model includes a photovoltaic output model, an energy storage output model and a load time sequence model. A site selection and capacity total target function is set, and the site selection and capacity total target function is used to calculate the minimum value of the sum of a plurality of cost factors based on the capacity of different devices and the dynamic characteristics of the photovoltaic storage load, so as to confirm the arrangement position and capacity of each power system device in the region to be planned.

[0008] According to the technical scheme provided in the application, based on the island division operation model, the island division form under the extreme condition in the region to be planned is confirmed according to the first configuration information, including: An island division operation target function is set, and the position of any distributed power supply in the region to be planned is taken as the center of a circle, and the capacity is taken as the radius to confirm the power circle range. Each load node in the power circle range is accessed, and it is judged whether each load node can be included in the island area formed by the corresponding power supply according to the first configuration information, until a relevant load point set corresponding to each power supply is obtained. When the load nodes are all accessed, each of the relevant load point sets is traversed, and if there are at least two relevant load point sets containing the same load node, the island areas corresponding to the two relevant load point sets are merged, and finally the island division area of the region to be planned is obtained.

[0009] According to the technical scheme provided in the application, it is judged whether each load node can be included in the island area formed by the corresponding power supply according to the first configuration information, until a relevant load point set corresponding to each power supply is obtained, including: Based on the first configuration information and the corresponding power circle range, the load points meeting the corresponding island condition are confirmed by sequentially accessing the load nodes adjacent to the power supply from the starting point of access. The continuous load power margin and the continuous load energy margin of the photovoltaic storage system of each island area under the extreme condition in the region to be planned are obtained. If the continuous load power margin and the continuous load energy margin are both greater than or equal to a first preset threshold value, it is judged that the division of each island area is stable, and a relevant load point set corresponding to each power supply is generated.

[0010] According to the technical scheme provided in the application, the improved cuckoo algorithm is used to optimize the first configuration information and the island division form in the region to be planned, including: The cuckoo algorithm is improved, and the improved cuckoo algorithm is used to optimize the arrangement position and capacity of each power system device in the region to be planned and the island region; The improvement of the cuckoo algorithm includes: An initial step factor is optimized, and a Skewtent mapping optimization initialization algorithm population is adopted; An iteration equation set by the algorithm is updated to obtain an improved cuckoo algorithm, so as to improve the exploration ability of the cuckoo algorithm.

[0011] According to the technical scheme provided in the application, the improved cuckoo algorithm is used to optimize the arrangement position and capacity of each power system device in the region to be planned, and further includes: The required operation parameters and the basic parameters of the power grid in the region to be planned are input into the improved cuckoo algorithm, the improved cuckoo algorithm is used to initialize the upper nest position, and each nest position is used to represent a standby first configuration information; Based on the step factor and the random step length generated by the flight mechanism configured by the improved cuckoo algorithm, the upper nest position is updated until the optimal solution of the upper nest position that meets the condition is obtained through iteration calculation; The first configuration information is updated based on the optimal solution of the upper nest position, and the island division form in the region to be planned under extreme conditions is determined based on the updated first configuration information.

[0012] According to the technical scheme provided in the application, the improved cuckoo algorithm is used to optimize the island region in the region to be planned, and further includes: The island division region of the region to be planned is obtained, and it is judged that the power supply in each island region supplies power to the load in the island region under the extreme condition of the region to be planned; When the power supply cannot supply power to all primary loads in the corresponding island region, a penalty factor is added to the calculation result; When the power supply can supply power to all primary loads in the corresponding island region, and there are remaining loads that cannot be supplied with power, the power loss cost of all loads is calculated, the upper nest position is optimized based on the power loss cost of all loads, and finally the optimal solution of the arrangement position and capacity of each power system device in the region to be planned, the recovered load amount of each island region under extreme conditions, and the total cost under the current planning are output.

[0013] In summary, the technical scheme specifically discloses a power distribution network resilience optimization method based on an improved cuckoo algorithm, which comprises the following steps: obtaining historical data of a power grid in a region to be planned, and establishing power working condition scenarios; based on a component model and a site selection and capacity planning model, first configuration information is confirmed according to the dynamic characteristics and total cost of the light storage load under each power working condition scenario; the component model is used to simulate the dynamic characteristics of the light storage load in the region to be planned; the first configuration information is the arrangement position and capacity of each power system device in the region to be planned; based on an island division operation model, the island division form under an extreme condition in the region to be planned is confirmed according to the first configuration information; and the first configuration information and the island division form in the region to be planned are optimized by using the improved cuckoo algorithm, so as to improve the resilience of the power distribution network in the region to be planned.

[0014] In the existing power grid planning, a reasonable and comprehensive power planning is made for the region to be planned, so that the resilience of the power grid obtained by the planning cannot be effectively improved. In the present application, the power working condition scenarios are constructed by obtaining the historical data of the power grid in the region to be planned, the arrangement position and capacity of the light storage load device are determined in combination with the component model and the site selection and capacity planning model, the island division form under an extreme condition is determined by using the island division operation model, and finally the configuration and island strategy are optimized by using the improved cuckoo algorithm, so that a multi-dimensional technical breakthrough is realized. Under the premise of taking into account the economy, the problem of insufficient resilience of the power grid caused by the single scene simulation and the static island strategy of the existing technology is effectively solved. BRIEF DESCRIPTION OF DRAWINGS

[0015] Other characteristics, objects and advantages of the present application will become more apparent from the following detailed description of the non-limiting embodiments, made with reference to the accompanying drawings: Figure 1 FIG. 1 is a flowchart of a power distribution network resilience optimization method based on an improved cuckoo algorithm.

[0016] Figure 2 FIG. 2 is an expanded schematic diagram of step S100 of the method.

[0017] Figure 3 FIG. 3 is an expanded schematic diagram of step S200 of the method.

[0018] Figure 4 FIG. 4 is an expanded schematic diagram of step S300 of the method.

[0019] Figure 5 FIG. 5 is an expanded schematic diagram of step S302 of the method. DETAILED DESCRIPTION

[0020] The application will be further described in detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the related application, and not to limit the application. In addition, it should be noted that only the parts related to the application are shown in the drawings for ease of description.

[0021] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0022] Embodiment 1 In order to make the technical solutions of the embodiments of the present application more clear and easy to understand, the application background of the embodiments of the present application is introduced below.

[0023] The current power system is facing double challenges: on the one hand, the frequent occurrence of extreme weather caused by climate warming, on the other hand, the uncertainty risk brought by high proportion of new energy access (such as wind power icing, photovoltaic dust influence). Therefore, it is urgent to improve the system resilience from the aspects of power grid planning, equipment disaster resistance standard, emergency response mechanism, etc., and to build a disaster prevention system suitable for the characteristics of “double high” power system, in order to guarantee energy security and social stable operation.

[0024] Existing research shows that the power system blackout accident caused by extreme events exposes the problem of insufficient resilience of the power grid, especially in the distribution network field. Because the automation level of the distribution network is low, the redundancy is insufficient, and it is directly connected to the terminal load, its disaster resistance is weak, which will seriously affect the social and economic operation and energy security. The current research on the recovery ability of the power system in extreme events is still in the development stage, especially in the aspects of resilience evaluation and planning, there is still a lack of systematic technical support, and the research and application of the resilience improvement technology of urban distribution network need to be strengthened.

[0025] Therefore, the present application proposes a distribution network resilience optimization method based on improved cuckoo algorithm, which combines Figure 1 As can be seen from the flowchart, the present application solves the problems of single scene simulation, static island strategy and low optimization efficiency of the prior art, and provides an effective resilience improvement path for urban distribution network to cope with extreme climate. The method comprises: S100, obtaining historical data of the power grid in the region to be planned, and establishing a power working condition scene; In order to improve the resilience of the power grid in the face of extreme disasters, the first thing to establish is an accurate power working condition scene, which provides accurate data for subsequent planning of each power system equipment in the region to be planned, and ensures the accuracy of the division.

[0026] Therefore, in this step, the power operating condition scene is constructed by historical data, which can cover uncertain factors such as illumination intensity fluctuation and load time sequence change (such as photovoltaic Beta distribution, load normal distribution model), so that the planning scheme can adapt to subsequent actual operation conditions.

[0027] Specifically, the power operating condition scene generally needs to include four typical scenes of “spring, summer, autumn and winter” in terms of planning scale, so as to ensure the rationality of the whole year planning, and the planning purpose is to make the photovoltaic and energy storage devices meet the load recovery needs under fault, so the power operating condition scene is established in the scenes of “spring, summer, autumn and winter” (for example, the summer load is higher in each month of the year, and the photovoltaic changes little in each season of the year), and then it is verified whether the photovoltaic and energy storage devices can meet all the load recovery needs under the condition of power failure caused by extreme disasters.

[0028] Specifically, referring to Figure 2 , the step S100 expands the following steps: S101, obtaining historical data of illumination intensity and load level in the region to be planned, and establishing probability models followed by the two; In order to save the calculation of the scene set and ensure the typicality of the scene establishment in the present application, the Latin Hypercube Sampling (LHS) method is first used to sample the historical data of illumination intensity and load level, which includes two processes of sampling and sorting. In the sampling process, the stratified sampling method is used to cover all the sampling areas and ensure the integrity of the sampling; in the sorting process, the correlation between the sampling value and the random variable is approximately equal by using the change of the permutation order.

[0029] LHS sampling is to sample N times on the probability distribution function P(X) of the random variable, and then divide the interval [0, 1] into subintervals, and the probability in each subinterval is Since this sampling method randomly selects a point in the above interval to obtain the sampling value of the point in the probability distribution function P(X) , the next sampling is randomly sampled in the remaining subinterval, until all intervals are sampled, so in order to realize the specific sampling method of LHS in the embodiment of the present application, it is necessary to first statistically analyze the historical data of illumination intensity and load level in the region to be planned to determine the main parameters of the probability model followed by the two, so as to establish the probability model followed by the two.

[0030] S102, generating a plurality of illumination intensity and load data from the probability model, and calculating the photovoltaic output efficiency and load power based on the plurality of illumination intensity and load data; Specifically, after the probability model generates a plurality of corresponding light intensity and load data respectively, the photovoltaic power generation output data and the load power data can be obtained through the subsequently proposed photovoltaic output model and load time sequence model, and then the photovoltaic output efficiency can be calculated. Since the combination of photovoltaic output efficiency and load power is the core link of constructing the basic scene, this process essentially couples the "power randomness" and "load volatility" to model, to simulate the supply and demand relationship in the real operation of the power distribution network. Therefore, by combining the photovoltaic output efficiency and the load power, the basic scene can be established, and each scene has the same probability, so the overall dimension is 4 dimensions.

[0031] S103, based on the photovoltaic output efficiency and the load power, at least one power working condition scene is established.

[0032] In this process, considering that if all the scene sets are used for calculation, it will have a great impact on the calculation speed, in order to reduce the time used for calculation, and at the same time obtain representative photovoltaic and load data for subsequent research, scene reduction technology needs to be used, so after obtaining a plurality of photovoltaic output efficiency and load power, the embodiments of the present application also adopt K-means The clustering algorithm can quickly and efficiently realize the reduction of the scene while maintaining the diversity of the reduced data when processing multi-dimensional variables and large-scale data; specifically, the clustering analysis is performed on the plurality of light intensity and load data obtained in advance, the diversity of the reduced data is maintained LHS On the basis of the change characteristics of the large-scale time sequence scene after sampling, the simplified power working condition scene is obtained.

[0033] S200, based on the element model and the site selection and capacity planning model, the first configuration information is confirmed according to the dynamic characteristics and the total cost of the light storage load under each power working condition scene; the element model is used to simulate the dynamic characteristics of the light storage load of the power grid in the region to be planned; and the first configuration information is the arrangement position and capacity of each power system device in the region to be planned; Next, after the power working condition scene is confirmed, the step of planning the arrangement position and capacity of each power system device in the region to be planned is entered. Since each planning of the power device cannot be separated from the support of data, in the embodiments of the present application, it is necessary to first establish a basic element model. The element model here includes a photovoltaic output model, an energy storage output model and a load time sequence model, which can reflect the different performance parameters and flexibility changes and consumption costs corresponding to the power grid under different planning. It can be understood that, since the output sizes of the load and the photovoltaic corresponding to each typical scene are different, the dynamic characteristics of the light storage load under different scenes can be accurately simulated by means of the photovoltaic output model, the energy storage output model and the load time sequence model, so as to output corresponding parameters.

[0034] Specifically, referring to Figure 3The step S200 further comprises the following steps: S201, constructing an element model, the element model comprising: a photovoltaic output model, an energy storage output model and a load time sequence model; The photovoltaic output model corresponds to the performance of a photovoltaic cell module, which is a core component of a photovoltaic power generation system. The output power of the photovoltaic cell module is mainly determined by the light intensity of the surrounding environment, and the photovoltaic cell module converts solar light energy into electrical energy by using the photovoltaic effect. Since the light intensity changes with sunrise and sunset, the change rule has certain randomness and intermittency.

[0035] According to a large number of historical measured light intensity data, it is shown that the light intensity distribution in a period of time can be reasonably fitted by a Beta distribution model, and the probability density function is f(r) The formula (1) is as follows: The formula (1) is as follows: Wherein, is a Gamma function; is a real-time light intensity value measured in a period of time in the surrounding environment of a certain area, is the maximum light intensity value measured in the period of time in the area; α and β are positive integers, which are used to construct the shape of the Beta distribution model, and the values thereof can be obtained by calculating the mean value μ and the variance σ of the light intensity in the period of time, and the calculation formula is as follows: The formula (2) is as follows: The formula (3) is as follows: If the photovoltaic power generation module is composed of M photovoltaic panels, the output power of the photovoltaic power generation system is P pv The output power of the photovoltaic power generation system is A , which can be obtained by the product of the total area of the photovoltaic panel r , the received light intensity and the conversion efficiency , and the calculation formula is as follows: The formula (4) is as follows: A Wherein, the total area of the photovoltaic panel and the conversion efficiency can be calculated by the following formula (5) and formula (6): The formula (5) is as follows: A The formula (6) is as follows: Am For the first m The area of ​​each photovoltaic panel; For the first m The conversion efficiency of a single photovoltaic panel.

[0036] The output power of a photovoltaic power generation system also exhibits a Beta distribution, and its calculation formula is as follows (7): Formula (7).

[0037] In formula (7): P pv,max This represents the maximum output power of the photovoltaic power generation system during the period of receiving sunlight.

[0038] The energy storage output model is used to characterize the randomness of the output of distributed generation (DG). Adding an energy storage system (ESS) can smooth out the output uncertainty. In the event of a fault, the ESS and islanded power supply provide power to the internal loads to improve system reliability and power quality.

[0039] Finally, considering charging and discharging efficiency, maximum output power, and capacity limitations, the charging and discharging process should satisfy the constraints of the following formula (8): Formula (8); In formula (8): For the first i load t Load value at any given time; For the first i Photovoltaic t Output value at any given moment; and The first i The charging and discharging capacity of an energy storage device; For the first i The remaining capacity of the energy storage; and The first i Minimum and maximum capacity limits for energy storage; For the island's running time; This represents the number of load nodes within the isolated island; This represents the number of distributed photovoltaic systems connected to the island. and These are the charging and discharging efficiencies of the ESS, respectively.

[0040] The load time series model is used to characterize the load in a power system, which is also a time-varying random variable. Its fluctuation characteristics can be approximated by a normal distribution, and the probability density function can be expressed as: Equation (9); In Equation (9), and are the active power and the reactive power of the load, respectively; and are the active standard deviation and the reactive standard deviation of the load, respectively; and are the active expectation value and the reactive expectation value of the load, respectively.

[0041] S202, a total site and capacity selection target function is set, and the total site and capacity selection target function is used to calculate the minimum value of the sum of a plurality of cost factors based on the capacity of different devices and the dynamic characteristics of the load and storage, so as to confirm the arrangement position and capacity of each power system device in the region to be planned.

[0042] After the component model is obtained, the total site and capacity selection target function is used to calculate the cost of each item based on the current total site and capacity selection target function. Under the total site and capacity selection target function, the position and capacity planning that meets the conditions and saves the cost can be obtained; It should be noted that if the established power scenario is multiple, the different loads and photovoltaic changes presented under different scenarios need to be brought into the total cost of the planning target function, and finally the total cost is used as a guiding theory to plan the position and capacity of the photovoltaic, energy storage, SOP and other devices. Here, it will not be described in detail.

[0043] It should be noted that the objects participating in the site and capacity selection include but are not limited to distributed power sources (such as photovoltaic power stations, wind power stations), energy storage systems, transformers, switch stations, distribution lines and other devices that need to determine the installation position in geographical space. Finally, under the following constraint conditions, the final first configuration information is obtained.

[0044] The photovoltaic investment and operation and maintenance cost C1 is as follows: Equation (10); In Equation (10), b is the discount rate; is the full life cycle of the photovoltaic; is the investment cost of the unit capacity photovoltaic; is the installed capacity of the photovoltaic at the node i ; is the set of photovoltaic installation nodes of the line; is the unit operation and maintenance cost of the photovoltaic; is the active power of the photovoltaic at the node t at the moment i ; The time interval of the photovoltaic continuous power generation is generally 1 h.

[0045] The energy storage investment and operation and maintenance cost C2 is as follows formula (11): Formula (11); In formula (11): is the full life cycle of the energy storage system; is the set of ESS installation nodes; is the unit investment capacity cost of the ESS; is the investment capacity of the ESS at node i; i is the unit investment power cost of the ESS; is the maximum charge and discharge power of the ESS at node i; is the unit operation and maintenance cost of the ESS; , is the charge and discharge power of the ESS at node i at time t; is the time interval of the ESS continuous power generation, which is generally 1 h. t i The time interval of the photovoltaic continuous power generation is generally 1 h.

[0046] The SOP investment and operation and maintenance cost C3 is as follows formula (12): Formula (12); In formula (12): is the operation life of the SOP; is the unit installation capacity cost of the SOP; is the installation capacity of the i th SOP; n is the number of installed SOPs; is the annual maintenance cost coefficient. The network loss cost C4 is as follows formula (13):

[0047] Formula (13); The load loss cost C5 is as follows formula (14): Formula (14); In formula (14): is the set of load reduction nodes; is the unit power reduction compensation cost of the load; is the load reduction amount of node i at time t; is the load reduction time of node i. t i i

[0048] ​​​​​​Based on the calculations of formulas (10)-(14) and the overall objective function of site selection and capacity determination, we can obtain the planning scheme with the minimum total cost after comprehensively considering the investment, operation and maintenance costs of photovoltaic and energy storage, network loss costs, and load loss costs. At the same time, in order to obtain the first configuration information more accurately, it is also necessary to perform photovoltaic capacity constraints, energy storage capacity constraints, SOP capacity constraints, curtailment rate constraints, and penetration rate constraints based on the dynamic characteristics of photovoltaic, energy storage and load under the corresponding power operating conditions and reflected by the model of each component.

[0049] The photovoltaic capacity constraint is as follows (15): Formula (15); In formula (15): P pv,i Let be the photovoltaic capacity at node i; P pv,i,max This represents the maximum photovoltaic capacity at node i. P pv,max This represents the maximum total installed capacity of distributed photovoltaic power that can be connected to the power grid.

[0050] Energy storage capacity constraints are as follows (16): Formula (16); In formula (16): P ess,i For nodes i The capacity of energy storage; P ess,i,max For nodes i The maximum capacity of the energy storage at that location; P ess,max This represents the maximum installed capacity of energy storage connected to the power grid.

[0051] Formula (17); In formula (17): For nodes i The active power output at the point; For nodes i The reactive power output at the location; For nodes j The active power output at the point; For nodes j The reactive power output at the location; For nodes ij The SOP installation capacity between them.

[0052] Waste light rate constraint: Formula (18); In formula (18): For nodes iphotovoltaic power generation amount of the node photovoltaic power generation amount of the node i photovoltaic power generation amount of the node

[0053] formula (19); in formula (19): photovoltaic power generation amount of the node i penetration rate of photovoltaic power at the node maximum penetration rate of photovoltaic power

[0054] S300, based on the island division operation model, according to the first configuration information, confirming the island division form under extreme conditions in the region to be planned; Firstly, island division of the region to be planned means that when the power distribution network encounters extreme disasters or failures, the strategy of maintaining key load power supply by reasonably dividing the "island" region for independent operation, using distributed power sources (such as photovoltaic, energy storage) and flexible interconnection devices (SOP). The core goal is to maximize the amount of load recovery and guarantee system stability when the power grid is partially disabled. Therefore, after knowing the location and capacity of each photovoltaic and energy storage device, the specific island strategy and recovery strategy under extreme working conditions can be planned to ensure the resilience of the risk resistance capability.

[0055] Specifically, referring to Figure 4 The step S300 further includes the following steps: S301, setting an island division operation objective function, and taking the location of any distributed power source in the region to be planned as the center of a circle and the capacity as the radius to confirm the power circle range; Firstly, the island division operation model is built-in island division operation objective function, and in the embodiment of the present application, the depth first search (depth first search, DFS) algorithm is used for preliminary island partitioning. This algorithm gives priority to the point closest to the starting point, expands the search depth layer by layer, and stops searching when a certain depth is reached. If the algorithm does not find a target that meets the conditions at the current depth, or reaches the end of the graph, the algorithm will backtrack to the last layer of nodes and reconsider other possibilities.

[0056] The island division operation objective function set in the embodiment of the present application is as follows formula (20): formula (20) in formula (20): total load amount of all island recovery , , respectively, the first-level load, the second-level load, and the third-level load node j int active power at time t; , , respectively the number of nodes of primary load, secondary load, and tertiary load; , , respectively the weight coefficients of nodes of primary load, secondary load, and tertiary load; is a variable representing the number of nodes of secondary load and tertiary load j In t 0-1 variable of curtailment state at time t, 0 represents that the load is curtailed, and 1 represents that the load is not curtailed; is the fault duration.

[0057] The core of the above island division operation objective function is to sum the power of each level of load to obtain the total load amount of all island restoration, and maximize it. In this way, it can ensure that as much load as possible is restored to supply power when the distribution network fails to form an island operation, reducing the impact of power outages; at the same time, different levels of load have different requirements for power supply reliability, and the island division operation objective function sets different weight coefficients to reflect the importance differences of primary load, secondary load, and tertiary load, ensuring the power supply restoration sequence of the load in extreme conditions of the power grid, which means that in the optimization process, the primary load with higher weight will be prioritized to supply power, and after meeting the power supply demand of the primary load, the secondary load will be considered, and finally the tertiary load. For example, hospitals, communication base stations and other primary loads will be prioritized to supply power during island division and load restoration to maintain the continuity of key services.

[0058] When planning an island, first take the location of any distributed power source in the planned area as the center, and its capacity as the radius to confirm the power circle range. The load nodes within the power circle range represent the range that can be powered by the power source under rated capacity, and the load nodes beyond the radius cannot be supported by the power source alone.

[0059] S302, by accessing each load node within the power circle range, using the first configuration information to determine whether each load node can be included in the island area formed by the corresponding power source, until a relevant load point set corresponding to each power source is obtained; After the power circle range is divided, each load node in the power circle range can be accessed, and the access is used to determine whether each load node can be included in the island range of the power supply; in the determination process, the first configuration information obtained in the foregoing process needs to be combined; after the distributed power supply node is searched, a new distributed power supply node needs to be selected as a starting node, and the depth-first search is continued; when all distributed power supplies are confirmed by the power circle range according to the above manner, the load points are accessed and determined in turn, and until all nodes are accessed, a relevant load point set corresponding to each power supply is obtained, and the relevant load point set includes load points included in the island region of the corresponding power supply.

[0060] Specifically, referring to Figure 5 The step S302 includes the following steps: S3021, based on the first configuration information and the corresponding power circle range, a load node adjacent to the power supply is accessed as a starting point in turn, and load points meeting the corresponding island condition are confirmed; The selected distributed power supply is taken as a starting node, the positions of the load nodes set based on the first configuration information are accessed in the power circle range, and the nodes adjacent to the starting node and not accessed are accessed in priority, if the node meets the island condition, the node is included in the island. When the distributed power supply node is searched, a new distributed power supply node is selected as a starting node, and the depth-first search is continued, and the island range is continuously expanded, the new node is included in the island, and the depth-first search strategy is maintained.

[0061] In combination with the foregoing "if the node meets the island condition, the node is included in the island", in the embodiment of the application, the island condition can be set as "power supply output ≥ load demand in the island + network loss", and the subsequent energy storage constraint condition also needs to be combined for determination.

[0062] S3022, the continuous load power margin and the continuous load energy margin of the light storage system of each island region in the extreme case in the region to be planned are obtained. S3023, if the continuous load power margin and the continuous load energy margin are greater than or equal to the first preset threshold, it is determined that the island region division is stable, and a relevant load point set corresponding to each power supply is generated.

[0063] The rationality of the dynamic islanding is evaluated by two indicators: "continuous load power margin" and "continuous load energy margin". When the continuous load power margin and continuous load energy margin of the power supply (photovoltaic energy storage system) in each islanding area are both greater than the first preset threshold (the first preset threshold can be set to 0, and can be adjusted according to the actual situation), it means that the system has a strong ability to continuously supply power to the load under extreme fault conditions and the power supply in the islanding area is stable.

[0064] (1) Continuous load power margin Under extreme operating conditions and during fault periods, the photovoltaic-storage system can achieve a margin for increased load supply while ensuring power balance within the island. The calculation formula is as follows: Formula (21): Formula (21); In formula (21): This refers to load forecasting error; This refers to a specific moment during the fault. This is the expected time when the fault will end; K This refers to the operating range of a microgrid island.

[0065] (2) Continuous load capacity margin M w ( t During a fault, the photovoltaic-storage system can achieve a margin for increased load supply while maintaining energy balance within the island. The calculation formula is as follows (22): Formula (22); In formula (22): For this load range The remaining energy stored after the time period; the sum of the initial energy of all ESS.

[0066] Using the above two metrics, each island can incorporate as much node load as possible while ensuring its stability. Based on this, the set of relevant load points corresponding to each power source can be accurately generated. When , Furthermore, when the energy storage constraints are met, the island is in a stable state, and the first preset threshold here is 0.

[0067] S303. After all load nodes have been visited, traverse each relevant load point set and determine if there are at least two relevant load point sets that contain the same load node. Then merge the island regions corresponding to the two relevant load point sets to obtain the island division region of the area to be planned.

[0068] When two or more island power circles cover a common load node, it indicates that these islands can be combined through interconnection devices (such as SOP) to share power resources, so if at least two relevant load point sets contain the same load node, the island regions of the two power sources can be combined, and after the combination, the redundant capacity of the two power sources can be utilized to improve the power supply reliability of C and reduce the loss of abandoned light or energy storage.

[0069] The aforementioned energy storage constraints are described in detail as follows: power flow constraints, system security constraints, node connection constraints, island connectivity constraints, island radial constraints, SOP active power constraints, and SOP reactive power constraints. The specific formulas are as follows: (1) The power flow constraints are as follows in formula (23): Formula (23); In formula (23): denotes the set of branch end nodes with j as the head node; denotes the set of branch head nodes with j as the end node; , and , are the active and reactive powers on the line at time t ; is the reactance value of the line; and are the active and reactive powers at node t at time j ; is the voltage value at node t at time j .

[0070] (2) The system security constraints are as follows in formula (24): Formula (24); In formula (24): and are the upper and lower limits of the voltage at node i; is the maximum current value for the safe operation of the line ij .

[0071] (3) The node connection constraints are as follows in formula (25): Formula (25); In formula (25): denotes whether node i belongs to the island at time t . uA 0-1 variable, which is 1 if it belongs to a certain category and 0 if it does not belong to a certain category; n The total number of nodes; This represents the total number of isolated islands. This constraint stipulates that each node can belong to only one isolated island.

[0072] (4) The connectivity constraint of isolated islands is given by the following formula (26): Formula (26); In formula (26): for t Time-isolated island u Middle node i The set of parent nodes. This constraint indicates that if an island... u There are nodes i Then at least one node must exist. i parent node k Belongs to an isolated island u This is how power node-to-node connections are formed in the isolated island. i Pathway.

[0073] (5) The island radial constraint is given by the following formula (27): Formula (27); In formula (27): To indicate being on an isolated island u The lines with China as the node and the first and last node are in t The connection status is represented by a 0-1 variable, where 1 indicates a connected line and 0 indicates a disconnected line. This constraint ensures that the difference between the number of nodes and the number of lines in each island is 1, guaranteeing that the island operates radially.

[0074] (6) The active power constraint of SOP is given by the following formulas (28) and (29): Formula (28); Formula (29); In formula (29): and They are respectively t Time period connected to node i and j The active power transmitted at the SOP; and They are respectively t Time period nodes i and nodes j Active power loss during SOP transmission; and These are the loss coefficients of the SOP at node i and node j, respectively; and respectively t periods are connected to nodes i and j reactive power of SOP transmission.

[0075] (7) The SOP reactive power constraint is as follows formula (30): formula (30); In formula (30): , and , respectively i and node j transmission minimum and maximum value of reactive power.

[0076] S400, using the improved cuckoo algorithm to optimize the first configuration information, the island division form in the region to be planned, has promoted the distribution network resilience of the region to be planned.

[0077] The following embodiments of the application also use an improved cuckoo algorithm to find its optimal solution to achieve optimal extreme value solving under different constraint conditions.

[0078] First, in the cuckoo algorithm in this application, the Lévy flight mechanism is introduced, which can significantly enhance the global search ability of the algorithm, making it more efficient in searching for the optimal solution. Specifically, Lévy flight is a non-Gaussian random process, also a random walking mechanism with heavy-tailed distribution, and is related to Lévy stability. Lévy stable distribution is a continuous distribution in probability theory, proposed by French mathematician Pual Pierre Lévy. Lévy stable is represented by scale σ , characteristic number α , displacement μ and skewness parameter β , and its characteristic function φ(t) continuous Fourier transform can be used to define Lévy distribution, as follows formula (31): formula (31); wherein: formula (32); The probability density function of the random step length distribution of Lévy flight λ(x) decays as follows formula (33): formula (33); The basic idea of the cuckoo algorithm is the nest parasitic behavior of the cuckoo and the Lévy flight behavior of the bird. In the algorithm, the Lévy flight is used to update the solution, which has very strong global search ability.

[0079] Specifically, in the embodiment of the present application, the improvement of the cuckoo search algorithm includes the following steps: Step one, optimize the initial step factor, and use Skewtent mapping to optimize the initialization algorithm population; The original cuckoo search algorithm is only suitable for solving continuous optimization problems. For discrete, constrained, combinatorial, and multi-objective optimization problems, the CS algorithm needs to be constantly improved. In the optimization process, if the step factor α is always large, the algorithm has strong exploration ability, but it is difficult to obtain the high-precision global optimal solution required in the formula; if the step factor α is always small, the cuckoo algorithm has to iterate many times to achieve the appropriate accuracy. Drawing on the idea of group information sharing in the particle swarm algorithm, the embodiment of the present application proposes a parameter α adaptive improvement formula (34) as follows: Formula (34); In formula (34), the maximum number of iterations of the cuckoo search algorithm is maxcs ; the current number of iterations is t ; and the parameter α increases with the increasing number of iterations, α starting from 0.6 and gradually decreasing until it approaches 0.001. The step factor α is large in the early stage, which has strong search ability, and the step factor α is small in the later stage, which has higher search accuracy.

[0080] In the process of function optimization of the cuckoo search algorithm, the position of the bird nest is usually randomly initialized, so that the initial parameters cannot be distributed throughout the space, which will lead to falling into local optimum and slow down the convergence speed. Chaos motion is a bounded, ergodic, internally random and statistically unstable finite constant motion, which can make the initial population more evenly distributed in space, and is widely used in artificial intelligence algorithms that are extremely sensitive to initial values, such as particle swarm optimization, grey wolf optimization, etc.

[0081] In order to distribute the initial bird nest in the edge area, the Skewtent mapping is used in this paper, and the Skewtent algorithm formula is as follows formula (35): Formula (35); Wherein φ , x∈[0,1], usually φ =0.5.

[0082] Step two, update the iteration equation set by the algorithm to obtain the improved cuckoo algorithm to improve the exploration ability of the cuckoo algorithm.

[0083] Because the Lévy flight trajectory of the classic cuckoo algorithm has a certain randomness, the convergence speed and the optimal solution found in the later iteration are not ideal, and people can only run the program to find the optimal solution. The traditional cuckoo algorithm uses the strategy of fighting and random migration in each iteration, which makes the historical optimal position of each bird nest as the final solution, resulting in a decrease in the exploration ability of other bird nests. In order to enhance the efficiency of search, the embodiments of the present application learn from the particle swarm algorithm, add "self-learning part" and "social learning part", and the improved bird nest position formula is as follows formula (36): Formula (36); In formula (36): X pi For each bird nest historical optimal position, the self-learning ability of position updating can be enhanced. X pg For all bird nest historical optimal positions, the social learning ability of position updating can be enhanced. r 1 is a learning factor for each bird nest historical optimal position, indicating the self-learning ability. r 2 is a learning factor for all bird nest historical optimal positions, a distance parameter for the global optimal solution.

[0084] In the step of "optimizing the layout position and capacity of each power system equipment in the to-be-planned area by using the improved cuckoo algorithm", there are two aspects of upper optimization and lower optimization. The upper optimization here specifically refers to the optimization of site selection and capacity planning to obtain more reasonable first configuration information, and through the update of the first configuration information, the lower layer is re-guided to perform island division.

[0085] Specifically, the upper optimization includes the following steps: Step R1, input the required running parameters and the basic parameters of the power grid in the to-be-planned area into the improved cuckoo algorithm, initialize the upper bird nest position by using the improved cuckoo algorithm, and each bird nest position is used to represent a standby first configuration information; First, input the required parameters of the algorithm, for example, including the bird nest population size N , the minimum and maximum values of the step factor α min and α max , the self-learning factor r 1 and the social learning factor r 2, the host discovery probability P a , and the dimension of spaceD , maximum number of iterations T max , and the basic data of the power grid. In the initialization process, the skew tent chaotic mapping method is used to generate the initial bird nest position.

[0086] Step R2, update the upper bird nest position based on the step factor and the random step generated by the flight mechanism configured by the improved cuckoo algorithm, until the optimal solution of the upper bird nest position that meets the conditions is obtained through iterative calculation; According to the planned range, initialize the position of the upper bird nest, and the position of each bird nest is a potential solution (first configuration information), which contains the installation position and the corresponding capacity size of the device; after initialization, set the number of iterations y =0, and update the bird nest position using the Lévy flight mechanism, which generates long and short random steps through its heavy-tailed characteristics, realizes dynamic updating of the solution and efficient exploration of the search space, and the number of iterations y = y +1.

[0087] It should be noted that the improved cuckoo algorithm introduces self-learning and social learning mechanisms in position updating, balances individual experience and group wisdom by adjusting the learning factor, and improves search efficiency; and variable boundary check and operation constraint verification are performed in each iteration to ensure that the optimization result meets the device parameter limit and power grid operation requirement.

[0088] Step R3, update the first configuration information based on the optimal solution of the upper bird nest position, and confirm the island division form under the extreme condition in the region to be planned with the updated first configuration information.

[0089] Finally, the optimal solution of the upper bird nest position is converted into the first configuration information, the update is realized, and the upper site selection and capacity determination result is substituted into the lower layer. The nodes of the distribution network are preliminarily islanded by the depth-first search algorithm. The light storage position and capacity in the first configuration information are substituted into each island, and when the continuous load power margin M p and the continuous load energy margin M w are greater than 0, it means that the island division meets the requirements.

[0090] Step R4, get the island division area of the region to be planned, and judge whether the power supply in each island area in the region to be planned under the extreme condition is the state of supplying power to the load in the island area. After the island division of the region to be planned is obtained, the planning result thereof needs to be further judged in the application by using the improved cuckoo algorithm, that is, whether the planning result is adapted to the fault recovery strategy; the power supply mentioned here is the state of the load power supply in the island region, and the details are as follows: Suppose that the region to be planned is in an extreme fault condition, and the fault recovery strategy is as follows: if the important load in an island region is restored, the remaining power is transmitted through the SOP to continue to restore the un-restored load in the remaining island, if the load in the island needs to be cut, the third-level load is preferentially cut, if the third-level load is cut off, the power supply still cannot supply power, then the second-level load is cut off to ensure that all first-level loads are supplied with power.

[0091] Step R5, when the power supply cannot supply power to all first-level loads in the corresponding island region, a penalty factor is added to the calculation result; It is found by calculation that, in the extreme fault condition, the first-level load in a certain island is also in power failure under the current island planning, which means that the first-level load cannot be restored under the planning, and a large penalty factor is added to the corresponding calculation result, which degrades the fitness value of the current solution to ensure that the algorithm preferentially avoids such infeasible schemes.

[0092] Step R6, when the power supply can supply power to all first-level loads in the corresponding island region, and there are remaining loads that cannot be supplied with power, the power loss cost of all loads is calculated, and the bird nest position in the upper layer is optimized based on the power loss cost of all loads, and finally the optimal solution of the arrangement position and capacity of each power system device in the region to be planned, the amount of restored load in the extreme condition of each island region, and the total cost under the current planning are output.

[0093] Based on the above description, it is known that there is a case under the island planning, that is, the first-level load power supply is restored, but there are some first-level or third-level loads that cannot be supplied with power, so the power failure time of all loads under the island planning needs to be calculated, and the power loss cost of all loads is calculated to return to the upper layer based on the loss cost, and the minimum amount of lost load (the least amount of power failure load) is used as a guide to perform iteration, the fitness of the upper bird nest under the iteration optimization is calculated and sorted, and in the process, when it is judged that the upper optimization has not reached the maximum iteration number, the bird nest position is updated by using the Lévy flight mechanism; if the maximum iteration number is reached, the optimal solution of the arrangement position and capacity of each power system device in the region to be planned, the amount of restored load in the extreme condition of each island region, and the total cost under the current planning can be directly output.

[0094] Based on the above description, the application proposes a power distribution network resilience optimization method based on an improved cuckoo algorithm, which comprises: obtaining historical data of the to-be-planned regional power grid, and establishing a power working condition scene; based on the component model and the site selection and capacity planning model, according to the dynamic characteristics and total cost of the light storage load under each power working condition scene, the first configuration information is confirmed; the component model is used to simulate the dynamic characteristics of the light storage load of the to-be-planned regional power grid; the first configuration information is the arrangement position and capacity of each power system device in the to-be-planned region; based on the island division operation model, according to the first configuration information, the island division form under extreme conditions in the to-be-planned region is confirmed; the improved cuckoo algorithm is used to optimize the first configuration information and the island division form in the to-be-planned region, so as to improve the resilience of the power distribution network in the to-be-planned region.

[0095] On the one hand, the method constructs a power working condition scene by obtaining historical data of the to-be-planned regional power grid, determines the arrangement position and capacity of the light storage load device in combination with the component model and the site selection and capacity planning model, and finally optimizes the configuration and island strategy by means of the improved cuckoo algorithm, realizing a multi-dimensional technical breakthrough. On the other hand, the scheme covers the full working condition operation demand by means of data-driven scene modeling, balances the economy and reliability through the collaborative configuration of light storage load, strengthens the key load power supply capacity under disaster by means of SOP flexible interconnection and dynamic island division, and improves the global optimization efficiency relying on the improved algorithm, forming a double-layer collaborative closed loop of "planning-verification-iteration"; compared with the traditional scheme, the above-mentioned problems such as single scene simulation, static island strategy and low optimization efficiency in the prior art are fundamentally solved, and the resilience improvement planning which is quantifiable, has higher practical value and higher economy is provided for the city power distribution network to cope with extreme weather.

[0096] The above description is only the preferred embodiment of the application and the explanation of the applied technical principles. Those skilled in the art should understand that the scope of the application disclosed in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above-mentioned features are replaced with the technical features disclosed in the present application (but not limited to) having similar functions to form a technical solution.

Claims

1. A power distribution network resilience optimization method based on improved cuckoo algorithm, characterized in that, The method comprises the following steps: acquiring historical data of a power grid in a region to be planned, and establishing power working condition scenarios; based on an element model and a site selection and capacity planning model, according to the dynamic characteristics and total cost of the photovoltaic and energy storage load under each of the power working condition scenarios, confirming first configuration information; the element model is used to simulate the dynamic characteristics of the photovoltaic and energy storage load of the power grid in the region to be planned; the first configuration information is the arrangement position and capacity of each power system device inside the region to be planned; based on an island division operation model, according to the first configuration information, confirming the island division form under extreme conditions in the region to be planned; using an improved cuckoo algorithm to optimize the first configuration information and the island division form in the region to be planned, so as to improve the resilience of the distribution network in the region to be planned.

2. The method of claim 1, wherein the historical data of the power grid in the region to be planned is acquired, and the power working condition scenarios are established, comprising: acquiring historical data of the intensity of light and the level of load in the region to be planned, and establishing a probability model followed by the two; generating a plurality of light intensity and load data from the probability model, and calculating the photovoltaic output efficiency and the load power based on the plurality of light intensity and load data; based on the photovoltaic output efficiency and the load power, at least one of the power working condition scenarios is established.

3. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 1, characterized in that, based on an element model and a site selection and capacity planning model, according to the dynamic characteristics and total cost of the photovoltaic and energy storage load under each of the power working condition scenarios, confirming first configuration information, comprising: constructing the element model, which includes a photovoltaic output model, an energy storage output model, and a load time sequence model; setting a site selection and capacity total objective function, which is used to calculate the minimum sum of a plurality of cost factors based on the capacity of different devices and the dynamic characteristics of photovoltaic and energy storage load, so as to confirm the arrangement position and capacity of each power system device inside the region to be planned.

4. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 1, characterized in that, based on an island division operation model, according to the first configuration information, confirming the island division form under extreme conditions in the region to be planned, comprising: setting an island division operation objective function, and taking the position of any distributed power supply in the region to be planned as the center and its capacity as the radius to confirm the power circle range; by accessing each load node in the power circle range, using the first configuration information to determine whether each load node can be included in the island area formed by the corresponding power supply, until a relevant load point set corresponding to each power supply is obtained; after the load nodes are accessed, traversing each of the relevant load point sets, if there are at least two relevant load point sets containing the same load node, merging the island areas corresponding to the two relevant load point sets, and finally obtaining the island division area of the region to be planned.

5. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 4, characterized in that, using the first configuration information to determine whether each load node can be included in the island area formed by the corresponding power supply, until a relevant load point set corresponding to each power supply is obtained, comprising: Based on the first configuration information and the corresponding power circle range, the load nodes adjacent to the power supply are sequentially accessed from the access starting point, and the load points meeting the corresponding island condition are confirmed; Obtain the continuous load power margin and the continuous load energy margin of the optical storage system of each island region in the planning area under extreme conditions; If the continuous load power margin and the continuous load energy margin are greater than or equal to the first preset threshold, it is determined that the division of each island region is stable, and a related load point set corresponding to each power supply is generated.

6. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 1, characterized in that, Optimize the first configuration information and the island division form in the planning area by using the improved cuckoo algorithm, including: The cuckoo algorithm is improved, and the improved cuckoo algorithm is used to optimize the arrangement position and capacity of each power system equipment in the planning area and the island region; The cuckoo algorithm is improved, and the improved cuckoo algorithm is used to optimize the arrangement position and capacity of each power system equipment in the planning area and the island region; The initial step factor is optimized, and the Skewtent mapping optimization initialization algorithm population is adopted; The iteration equation set by the algorithm is updated to obtain the improved cuckoo algorithm, so as to improve the exploration ability of the cuckoo algorithm.

7. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 6, characterized in that, The arrangement position and capacity of each power system equipment in the planning area are optimized by using the improved cuckoo algorithm, and further include: The required operation parameters and the basic parameters of the power grid in the planning area are input into the improved cuckoo algorithm, and the improved cuckoo algorithm is used to initialize the upper bird nest position. Each bird nest position is used to represent a standby first configuration information; Based on the step factor and the random step length generated by the flight mechanism configured by the improved cuckoo algorithm, the upper bird nest position is updated until the optimal solution of the upper bird nest position meeting the condition is obtained through iteration calculation; Based on the optimal solution of the upper bird nest position, the first configuration information is updated, and the island division form under extreme conditions in the planning area is confirmed based on the updated first configuration information.

8. The power distribution network resilience optimization method based on improved cuckoo algorithm according to claim 7, characterized in that, The island region in the planning area is optimized by using the improved cuckoo algorithm, and further include: The island division region of the planning area is obtained, and it is judged that the power supply in each island region supplies power to the load in the island region under extreme conditions in the planning area; When the power supply cannot supply power to all primary loads in the corresponding island region, a penalty factor is added to the calculation result; When the power supply can supply power to all primary loads in the corresponding island region, and there are remaining loads that cannot be powered, the outage loss cost of all loads is calculated, and the upper bird nest position is optimized based on the outage loss cost of all loads. Finally, the optimal solution of the arrangement position and capacity of each power system equipment in the planning area, the recovered load amount of each island region under extreme conditions, and the total cost under the current planning are output.

Images