Power distribution network switch planning method considering conditional value-at-risk

By introducing a mixed-integer linear model and conditional value of risk into the planning of distribution network switches, the problems of uncertain parameters and investment risks are solved, stability and reliability in fault events are achieved, profits and risks are balanced, and the accuracy and efficiency of planning are improved.

CN118917653APending Publication Date: 2024-11-08LUZHOU POWER SUPPLY COMPANY OF SICHUAN ELECTRIC POWER
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410904128.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-08
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing technologies fail to effectively consider the impact of uncertain parameters and related investment risks in distribution network switch planning, resulting in unstable stability and economy of the distribution network system under different fault events, affecting reliability and investment risk.

Method used

A mixed-integer linear model is adopted. Multiple failure scenarios are generated, and the investment risk is quantified using Conditional Value at Risk (CVaR). The scenarios are reduced by combining fast prior elimination technology. An objective function is set to maximize expected profit and minimize investment risk. The number, type and location of switches are solved by constraints.

Benefits of technology

Maintaining the stability and reliability of the distribution network system under different fault events, balancing expected profits and investment risks, and improving the accuracy of planning results and simulation efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917653A_ABST
    Figure CN118917653A_ABST
Patent Text Reader

Abstract

The invention provides a power distribution network switch planning method considering conditional value-at-risk, and relates to the technical field of power distribution network planning. The method comprises the following steps: generating a plurality of fault scenes to form an original scene set; reducing the original scene set to generate a new scene set; setting a target function and a constraint condition; and solving the objective function based on the constraint condition to obtain the number, the type and the installation position of the power distribution network switches to be installed. According to the method, good stability of the power distribution network system in different fault events can be guaranteed through reasonable switching planning, and the reliability of the power distribution network system is guaranteed. Furthermore, the expected profit and the investment risk can be balanced according to the risk preference, and the investment risk is reduced through reasonable risk preference coefficient and confidence setting.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network planning, and in particular to a distribution network switch planning method considering conditional risk value. Background Art

[0002] In the distribution network system, planning switchgear can improve the reliability of the distribution network system to a certain extent by reducing the fault location time and fault isolation time, but it will also increase investment, installation and maintenance costs, and reduce the economic efficiency of the distribution network system. However, the randomness of fault events makes the effect of installing switchgear unstable, resulting in the distribution network system showing completely opposite stability characteristics in different fault events, affecting the reliability of the distribution network system. In addition, in different fault events, the increase in revenue uncertainty leads to increased investment risks, thereby affecting the value of the switch planning scheme.

[0003] Although the existing technologies provide solutions to the switch planning problem from different aspects, there is no technical solution based on a mixed integer linear model to consider the influence of uncertain parameters and related investment risks in the switch planning problem. Summary of the invention

[0004] In order to solve the above technical problems existing in the prior art, the present invention provides a distribution network switch planning method considering conditional risk value.

[0005] The method specifically comprises the following steps:

[0006] Step S1: Generate multiple fault scenarios to form the original scenario set Ω D ;

[0007] Step S2: Reduce the original scene set Ω D , generate scene set Ω S ;

[0008] Step S3: Set the objective function maxf,

[0009] maxf=(1-β)×P rofit +β×C VaR,α β∈[0,1];

[0010] In the formula, β is the risk preference coefficient, P rofit is the expected profit after installing the switch, C VaR,α is the conditional value at risk under the confidence level α;

[0011]

[0012] In the formula, C w0-RCS is the system interruption cost when no switch is installed in fault scenario w, C w-RCSis the system interruption cost when installing a switch in failure scenario w, C RCS is the present value of the total cost of installing the switch;

[0013]

[0014] In the formula, C w is the set of faults in fault scenario w, J is the set of load points, and K is the set of user types; L f,j,k is the load size of type k user at load point j in feeder f, cdf f,j,k is the unit power outage loss cost of category k user at load point j in feeder f, and The feeder f of the k-type user at load point j is before and after the switchgear is installed. c in i c The duration of the power outage caused by the segmentation fault;

[0015] C RCS =CI RCS +IC RCS +MC RCS ;

[0016]

[0017] In the formula, CI RCS is the total investment cost of the switchgear, IC RCS is the total installation cost of the switch, MC RCS is the total maintenance cost of the switch, X RCS,f,e is a binary variable of whether to install a switchgear at the candidate position e of feeder f, CI RCS,f,e is the unit investment cost of the switchgear, IC RCS,f,e is the installation cost per switch, DR is the discount rate, MC is RCS,f,e The maintenance cost of the switch unit;

[0018]

[0019] Where ξ is the conditional risk value under the confidence level α, η w is a dummy auxiliary variable for scenario w;

[0020] Step S4: setting constraints, including calculating the load point outage time;

[0021] If there is an automatic switch between the load point and the fault area, then

[0022]

[0023] In the formula, is the load point outage time, T RCS is the action time of the automatic switch;

[0024] If there is a manual switch between the load point and the fault area,

[0025]

[0026] Where, T MCS is the action time of the manual switch, X RCS,f,e A binary variable indicating whether to install an automatic switch at the candidate position e of feeder f, v f A binary variable indicating whether there is a connection between feeder f and other feeders;

[0027] If there is no switch between the load point and the fault area, then

[0028]

[0029] In the formula, is the time required to repair the fault, X MS,f,e A binary variable indicating whether an automatic switch is installed at the candidate location e of feeder f.

[0030] Step S5: Solve the objective function to obtain the number, type and installation location of the distribution network switches to be installed.

[0031] Preferably, step S1 comprises:

[0032] Step S11: traverse each component, generate a first random number through standard uniform distribution, and combine it with the first probability density function to sample the fault time of the corresponding component;

[0033] Step S12: traverse the components in the fault state, generate a second random number through standard uniform distribution, and combine it with the second probability density function to sample the repair time of the component;

[0034] Step S13: repeating steps S11 and S12 until the fault time and the repair time cover the entire planning period;

[0035] Step S14: Find the first faulty component and determine its repair time, and find the next faulty component after repair, until the entire planning period is covered and combined into a fault scenario;

[0036] Step S15: Repeat steps S11 and S15 to generate multiple fault scenarios to form an original scenario set Ω D .

[0037] Optionally, a fault scenario may also be generated by a Latin hypercube sampling method.

[0038] Preferably, step S2 includes using a fast previous generation elimination technique to reduce the fault scenario;

[0039] Step S21: Calculate the weighted distance d of scene w w ,

[0040]

[0041] In the formula, v w,s is the Euclidean distance between scene w and scene s, v w and v s are the total system costs of scenario w and scenario s respectively, ||·|| represents the Euclidean distance, π s is the occurrence probability of scene s, n is the original scene set Ω D The number of scenes in

[0042] Step S22: Select a preset number of scenes from small to large according to the weighted distance and add them to the empty scene set Ω S , and from the original scene set Ω D Delete it.

[0043] Optionally, k-means clustering may be used to reduce the number of fault scenarios.

[0044] Further, step S4 also includes setting switch installation constraints,

[0045]

[0046] Further, step S4 also includes setting a switch budget constraint,

[0047]

[0048] Where N is the maximum planned number of switchgear that the power grid company can afford.

[0049] The technical solution provided by the present invention can ensure that the distribution network system maintains good stability in different fault events and ensure the reliability of the distribution network system through reasonable switch planning. Furthermore, the expected profit and investment risk can be balanced according to the risk preference, and the investment risk can be reduced by setting a reasonable risk preference coefficient and confidence level; the use of fast previous generation elimination technology can ensure the accuracy of fault scenario reduction, and simulation can be performed for a sufficient period of time on the basis of ensuring the accuracy of the planning results; setting switch installation constraints and switch budget constraints can further enhance the above effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The figure is a brief flowchart of the distribution network switch planning method of the present invention.

[0051] Figure 2 This is a schematic diagram of a typical distribution network structure.

[0052] Figure 3 This is a schematic diagram of the RBTS4 (Roy Billinton Test System 4) distribution network structure.

[0053] Figure 4 Schematic diagram of the impact of risk preference coefficient on switch planning results.

[0054] Figure 5 Schematic diagram of the impact of confidence on switch planning results. DETAILED DESCRIPTION

[0055] At present, the solutions to the distribution network switch planning problem can be roughly divided into two categories: heuristic algorithms and mathematical programming methods. For example, by analyzing the sensitivity of distribution network power supply reliability and feeder automation terminal configuration, the failure mode and consequence analysis method is used to propose a distribution terminal planning method based on the distribution network reliability development goal; or considering the distribution network reliability assessment method of distribution automation, a scientific planning method for distribution automation terminal layout is proposed and solved by genetic algorithm; or a multi-objective switch optimization layout method for radial distribution network based on particle swarm algorithm provides a complete set of solutions; or with the goal of minimizing the sum of user outage cost and switch investment cost, the immune algorithm is used to obtain the best switch planning scheme for the distribution network. Although the above methods can use heuristic algorithms to obtain a reasonable number and location of switch planning, they cannot guarantee the global optimal solution, and there will be a large computational burden when implemented in large-scale systems. Therefore, using mathematical programming methods to find the best solution becomes another way. For example, an optimal switch configuration strategy for the distribution system is developed by taking into account the failure rate of remote automatic switches and the load growth rate, and the strategy is described as a mixed integer linear programming model to ensure the optimality of the solution; or a distribution network automatic switch planning method is proposed in the form of a mixed integer linear programming model considering the randomness of failure occurrence; or a reliability analysis method containing distribution automation terminals is proposed to achieve the separation of reliability calculation and optimization solution, and a commercial solver is used to solve the automation terminal optimization configuration model.

[0056] Although the above methods have studied the switch planning problem from different aspects, they have not considered the influence of uncertain parameters and related investment risks in the switch planning problem based on mixed integer linear models. To this end, the present invention proposes a distribution network switch planning method considering conditional risk value. Among them, the investment risk is quantified by the conditional value-at-risk (CVaR), which means that under a certain confidence level, the risk loss is greater than the average risk loss value of a given risk value critical value, which can overcome the shortcomings of risk value that cannot meet consistency, the insufficiency of tail loss measurement, the inability to correctly identify risks, and the lack of subadditivity.

[0057] The technical solution provided by the present invention will be further explained below in combination with embodiments and drawings.

[0058] Example 1

[0059] like Figure 1 As shown in the figure, in this embodiment, a method for generating a fault scenario set is first proposed, and the fast predecessor elimination technology is used to reduce the fault scenarios to improve the solution efficiency. Secondly, a distribution network switch planning model considering risk value is proposed to maximize the expected profit and minimize the investment risk. Finally, the RBTS 4 distribution system is used as an example for simulation analysis, which shows the necessity of considering the conditional risk value in the switch planning problem, and analyzes the influence of different key parameters on the switch planning scheme.

[0060] Step 1: Generate fault scenario

[0061] This embodiment uses the sequential Monte Carlo simulation method to generate fault scenarios, and each scenario represents the normal / faulty state of all components in the planning period. Assuming that all components are in normal state at the initial stage, the fault scenario generation steps are as follows:

[0062] Step (1) Select a component of a distribution network system, including electrical components such as transformers, circuit breakers, disconnectors, capacitors, current transformers, etc. Generate a random number through a standard uniform distribution, and combine it with an appropriate probability density function to sample the fault time of the component. The probability density function can be obtained through statistical data of component failures;

[0063] Specifically, sampling is implemented in the following way: given M random variables X1, X2, ..., X M , the probability distribution function Y of the random variable X K for:

[0064] Y K =G K (X K ), K∈[1,2,...,M];

[0065] Assuming the sampling scale is N, the probability space of each random variable N is divided into N equal parts, that is, N sub-intervals are obtained:

[0066]

[0067] In each subinterval, a value is randomly selected as the sampling value.

[0068] It can be seen that for any sample set obtained by sampling, its probability density function is simple and easy to obtain. For any sample, its probability of occurrence can be obtained by statistical calculation of the sample set.

[0069] Step (2) randomly samples the repair time of the component in the faulty state in step (1). Similar to step (1), a random number is generated by a standard uniform distribution and combined with an appropriate probability density function to sample the repair time of the component. The sampling method is the same as (1).

[0070] Step (3) repeats step (1) and step (2) until the normal / fault state sampling of the component covers the entire planning period.

[0071] Step (4) Repeat steps (1) to (3) until all components in the system are traversed.

[0072] Step (5) finds the first faulty component and determines its repair time; after the first component is repaired, finds the next faulty component, and so on until the entire planning cycle is covered, and combines them into a fault scenario. It should be pointed out that since multiple faults rarely occur in actual production, the scenario generation method in this paper only considers the case of single faults.

[0073] Step (6) Repeat steps (1) to (5) until a sufficient number of fault scenarios are generated.

[0074] Optionally, a fault scenario may also be generated by a Latin hypercube sampling method.

[0075] Step 2: Failure scenario reduction

[0076] In the sequential Monte Carlo simulation method, the accuracy of the study increases with the increase in the number of generated scenarios, but it also greatly increases the complexity of the model, and the scenario reduction technology must be used to select a subset of scenarios that can effectively represent the original set. This embodiment uses the fast previous generation elimination technology to reduce the fault scenario, and the specific steps are as follows:

[0077] Step (1) Set Ω D is the original scene set, Ω S is an empty set, and the Euclidean distance between each scene pair is calculated.

[0078]

[0079] In the formula, v w,s is the Euclidean distance between scene w and scene s, v w and v s are the total system costs for scenario w and scenario s respectively.

[0080] Step (2) calculates the weighted distance of each scene,

[0081]

[0082] In the formula, π sis the probability of scene s, n is the original scene set Ω D The number of scenes in

[0083] And select the scene with the smallest weighted distance and add it to Ω S In, and from itΩ D delete.

[0084] Step (3) When Ω S When the number of scenes in the program does not reach the preset value, repeat steps 1) and 2) until the number of scenes reaches the preset value.

[0085] Optionally, k-means clustering may be used to reduce the number of fault scenarios.

[0086] Step 3: Set the objective function

[0087] The distribution network switch planning model proposed in this embodiment takes into account the investment risk caused by the uncertainty of fault events and uses CVaR to quantify the risk. Since power grid companies usually hope to obtain higher profits with lower risks, the objective function of the model is to maximize the weighted sum of expected profits and CVaR, as shown in the following formula:

[0088] maxf=(1-β)×P rofit +β×C VaR,α β∈[0,1];

[0089] In the formula, β is the risk preference coefficient, P rofit is the expected profit after installing the switch, C VaR,α is the conditional value at risk under the confidence level of α. β is used to balance expected profit and CVaR and can be set according to the risk preference of the power grid company. An aggressive power grid company may set β to 0 to obtain the highest expected profit without considering the risk; a conservative power grid company may set β to 1 to minimize the investment risk.

[0090] The expected profit of the distribution network is calculated as follows: the expected profit in the fault scenario w is calculated by subtracting the cost associated with installing switchgear from the reduced system interruption cost,

[0091]

[0092] In the formula, C w0-RCS is the system interruption cost when no switch is installed in fault scenario w, C w-RCS is the system interruption cost when installing a switch in failure scenario w, C RCS is the present value of the total cost of installing the switch.

[0093] C w0-RCS and C w-RCS The calculation method is as follows:

[0094]

[0095] In the formula, C w is the set of faults in fault scenario w, J is the set of load points, and K is the set of user types; L f,j,k is the load size of type k user at load point j in feeder f, cdf f,j,k is the unit power outage loss cost of category k user at load point j in feeder f, and The feeder f of the k-type user at load point j is before and after the switchgear is installed. c in i c The duration of the power outage caused by the segmentation fault.

[0096] C RCS The calculation method is:

[0097] C RCS =CI RCS +IC RCS +MC RCS ;

[0098]

[0099] In the formula, CI RCS is the total investment cost of the switchgear, IC RCS is the total installation cost of the switch, MC RCS is the total maintenance cost of the switch, F is the set of feeders, S is the set of candidate locations, X RCS,f,e is a binary variable of whether to install a switchgear at the candidate position e of feeder f, CI RCS,f,e is the unit investment cost of the switchgear, IC RCS,f,e is the installation cost per switch, DR is the discount rate, MC is RCS,f,e is the maintenance cost per switch.

[0100] CVaR is a supplement to Value-at-risk (VaR), and its meaning is the conditional expectation of the part that exceeds VaR.

[0101]

[0102] Where x and y are decision variables and random variables respectively; f(x, y) is the risk loss function; ρ(y) is the probability density function of y; ξ is the VaR value under the confidence level α; [f(x, y)-α] + represents max[f(x,y)-α,0]. Based on the above formula, a virtual auxiliary variable η is introduced w , the risk measurement cost can be expressed as:

[0103]

[0104] Step 4: Set constraints

[0105] Load point outage time calculation: Load point outage time depends on many factors, such as network topology, switch action time and fault repair time. Figure 2 As shown, this embodiment takes a typical distribution network structure as an example to illustrate the impact of installing switchgear on the duration of power outage at a load point.

[0106] Assumption i c After the fault, the circuit breaker at the head end of the feeder is disconnected, and all load points of the feeder are powered off. Depending on whether there is a switchgear between the load point and the fault, there are three ways to restore power to the load:

[0107] 1) If there is an automatic switch between the load point and the fault area, the power outage time of the load point is the action time of the automatic switch;

[0108] 2) If there is a manual switch between the load point and the fault area, the load point power outage time is the manual switch action time;

[0109] 3) If there is no switchgear between the load point and the fault area, the power outage time at the load point is the time required to repair the fault.

[0110] The above three power supply recovery methods are expressed by the following formula:

[0111] If there is an automatic switch between the load point and the fault area, then

[0112]

[0113] In the formula, is the load point outage time, T RCS is the action time of the automatic switch. In this embodiment, the circuit breaker can be remotely controlled.

[0114] If there is a manual switch between the load point and the fault area,

[0115]

[0116] Where, T MCS is the action time of the manual switch, X RCS,f,e A binary variable indicating whether to install an automatic switch at the candidate position e of feeder f, v f A binary variable indicating whether there is a connection between feeder f and other feeders.

[0117] If there is no switch between the load point and the fault area, then

[0118]

[0119] In the formula, is the time required to repair the fault, X MS,f,e A binary variable indicating whether a manual switch is installed at the candidate position e of feeder f.

[0120] Switch installation constraints: Automatic switches and manual switches cannot be installed at the same switch candidate location.

[0121]

[0122] Switch budget constraint: Installing switchgear requires huge investment costs, so its investment budget should be taken into account in the switch planning model.

[0123]

[0124] Where N is the maximum planned number of switchgear that the power grid company can afford.

[0125] Solve the objective function based on the constraints to obtain the number, type and installation location of the distribution network switches. In the switch planning scheme, any fault scenario determines a switch installation location; for the installable location, there is a binary option of installing a switch\not installing a switch; for the installable location of the switch, there are two types of switches to choose from: manual switch and automatic switch. Installing a switch\not installing a switch and the type of installed switch can all be obtained based on the constraints combined with solving the objective function. Among them, commercial solvers such as CPLEX can be used to quickly and accurately solve.

[0126] Specifically, this embodiment provides a calculation analysis. The calculation analysis uses the RBTS4 distribution network as an example to verify the effectiveness of the model. The system has been widely used in reliability assessment and switch planning problems. Its network topology is as follows: Figure 3 As shown. The system supplies power to 4,700 residential users, 70 commercial users, and 9 small users through 38 load points on 7 feeders. The failure rate, maintenance time, and unit power outage loss cost of each component in the system can be obtained from the IEEE (Institute of Electrical and Electronics Engineers) journal. The investment and installation cost of each automatic switchgear is set to 33,000 yuan, and the annual maintenance cost is 2% of the installation cost. Over the 15-year planning period, the annual load growth rate and discount rate are 3% and 8%, respectively. The confidence level of the conditional risk value is set to 0.95, and 10,000 fault scenarios are generated using the Monte Carlo simulation method, and then the scenarios are reduced to 5,000 using the rapid predecessor elimination technology.

[0127] Furthermore, this embodiment sets three scenarios to analyze the calculation results, and gives the switch planning results in Tables 1 to 3.

[0128] Scenario 1: In this scenario, the switch planning study is conducted from the perspective of an aggressive grid company, and the risk preference coefficient is set to 0 to maximize the expected profit without considering the investment risk. As can be seen from Table 1, 21 automatic switches are installed in the system, 2 switches are installed in feeders F1, F3, F4, and F7 to supply power to a large number of commercial and residential users, and multiple automatic switches are installed in feeders F2, F5, and F6 to supply power to small users.

[0129] Table 1 Scenario 1 Switch Installation Location

[0130] Feeder Location Feeder Location 1 5,9 5 1,2,4,5 2 1,2,3,4 6 1,2,3,4,5 3 5,9 7 5,9 4 4,9

[0131] Scenario 2: In this scenario, the switch planning study is conducted from the perspective of a conservative power grid company, and the risk preference coefficient is set to 1 to maximize CVaR. As can be seen from Table 2, a total of 10 automatic switches are installed in the system to minimize investment risks. Two automatic switches are installed in feeders F2, F5, and F6 to supply power to small users, and the remaining four automatic switches are evenly distributed in feeders F1, F3, F4, and F7, all installed at the end of the feeder connecting users with high unit power outage loss costs.

[0132] Table 2 Switch installation position for scenario 2

[0133] Feeder Location Feeder Location 1 9 5 2,5 2 3,4 6 2,5 3 9 7 9 4 9

[0134] Scenario 3: In this scenario, the switch planning is studied from the perspective of a rational power grid company, and the risk preference coefficient is set to 0.5 to balance the expected profit and investment risk. As can be seen from Table 3, a total of 11 automatic switches are installed in the system. Although the number is similar to that in Scenario 2, the planned positions are different. One more automatic switch is installed on feeder F4, and the positions of the switches in feeders F5 and F6 have also changed.

[0135] Table 3 Scenario 3 switch installation locations

[0136] Feeder Location Feeder Location 1 9 5 3,5 2 3,4 6 3,5 3 9 7 9 4 4,9

[0137] Table 4 lists the expected profits and CVaR under the three scenarios. Compared with the aggressive grid company, the conservative grid company reduced the investment risk by nearly 33.1%, while the expected profit decreased by 2.3%. It can be seen that the aggressive grid company faces objective expected profits and inevitable investment risks. The expected profit of the rational grid company is 1.8% lower than the highest achievable expected profit, and the CVaR is also 1.5% lower than the highest achievable CVaR. Therefore, a balance between expected profits and investment risks is achieved in scenario three.

[0138] Table 4: Installation positions of switches in scenario 4

[0139] Scenario Number of switches Expected profit (ten thousand yuan) CVaR (10,000 yuan) one 21 343 78 two 10 334 104

[0140] Furthermore, this embodiment analyzes the impact of key parameters of the example on the switch. Figure 4 As shown in the figure, in order to study the impact of the risk preference coefficient, it is gradually increased from 0 to 1. Increasing the risk preference coefficient will reduce the expected profit. At the same time, CVaR increases with the increase of the risk preference coefficient, indicating that the investment risk is also reduced. In addition, as the risk preference coefficient increases, the planned automatic switches also decrease. Therefore, it can be concluded that grid companies that tend to avoid risks rarely carry out switch planning projects. Figure 4 It can also be seen that a small deviation in expected profit may lead to a large change in investment risk. Specifically, the expected profit is only reduced by 1.4% from 3.41 million yuan to 3.39 million yuan, while the investment risk is reduced by 17.8% from 1.03 million yuan to 860,000 yuan. At the same time, the optimal number of automatic switches is reduced from 17 to 11, which reduces the investment cost of the power grid company.

[0141] Furthermore, this embodiment analyzes the impact of confidence level on switch planning. Confidence level plays a vital role in measuring CVaR, and power grid companies with different risk preferences will choose different confidence levels. To analyze the impact of confidence level, the switch planning problem under various confidence levels is simulated. Figure 5 It can be seen that in scenario 1, the optimal number of automatic switches remains unchanged. This is because the aggressive power grid company does not consider the impact of investment risk when planning, so changing the confidence level will not affect its decision. In scenarios 2 and 3, the optimal number of automatic switches decreases as the confidence level increases. When the confidence level is set to 1, the number of automatic switches in scenarios 2 and 3 converges to 8 and 10, respectively. At the same time, it can be seen that the fluctuation of the optimal number of automatic switches will increase with the increase of the risk preference coefficient. Therefore, the setting of confidence has a great impact on the switch planning problem.

[0142] To study the impact of reducing the number of fault scenarios on the accuracy and running time of switch planning results. In this embodiment, the number of fault scenarios is set to 10000, 7500, 5000 and 2500 respectively to simulate scenario 1, and the results are shown in Table 5. It can be seen that compared with the most accurate expected profit, reducing the number of fault scenarios will lead to an increase in the error of the expected profit. At the same time, when the number of fault scenarios is reduced to 2500, the number of optimal switch plans has also changed. However, when the number of fault scenarios decreases, the simulation time is also shortened, reducing by up to 93%. In addition, Table 6 shows the location of automatic switch planning under different numbers of fault scenarios. It can be seen that only when the number of scenarios is 2500, the number and location of switches change, and the head end of feeder 6 is no longer planned. Switches. Therefore, the fault scenario reduction method proposed in this paper is very effective. At the same time, in order to achieve a balance between result accuracy and simulation time, this paper sets the number of fault scenarios to 5000.

[0143] Table 5 The impact of the number of fault scenarios on expected profit and solution time

[0144] Number of scenes Number of switches Expected profit (ten thousand yuan) Profit error (%) Solution time(s) Time error (%) 10000 21 347.6 - 398 - 7500 21 347.2 0.11 199 50 5000 21 343.1 1.33 99 75 2500 20 341.4 1.76 29 92

[0145] Table 6 Impact of the number of fault scenarios on switch planning locations

[0146]

[0147] As can be seen from this embodiment, the technical solution provided by the present invention can ensure that the distribution network system maintains good stability in different fault events and ensure the reliability of the distribution network system through reasonable switch planning. Furthermore, the expected profit and investment risk can be balanced according to the risk preference, and the investment risk can be reduced by setting a reasonable risk preference coefficient and confidence level.

[0148] Furthermore, each preferred scheme has achieved the following beneficial effects on the basis of achieving the above-mentioned beneficial effects: the use of fast previous generation elimination technology can ensure the accuracy of fault scenario reduction, and simulation can be performed for a sufficient period of time while ensuring the accuracy of planning results; setting switch installation constraints and switch budget constraints can further enhance the above-mentioned effects.

Claims

1. A distribution network switch planning method considering conditional risk value, characterized in that: The following steps are involved: Step S1: Generate multiple fault scenarios to form the original scenario set Ω D ; Step S2: Reduce the original scene set Ω D , generate scene set Ω S ; Step S3: Set the objective function maxf, max f=(1-β)×P rofit +β×C VaR,α β∈[0,1]; In the formula, β is the risk preference coefficient, P rofit is the expected profit after installing the switch, C VaR,α is the conditional value at risk under the confidence level α; In the formula, π w is the occurrence probability of scene w, C w0-RCS is the system interruption cost when no switch is installed in fault scenario w, C w-RCS is the system interruption cost when installing a switch in failure scenario w, C RCS is the present value of the total cost of installing the switch; In the formula, C w is the set of faults in fault scenario w, J is the set of load points, and K is the set of user types; L f,j,k is the load size of type k user at load point j in feeder f, cdf f,j,k is the unit power outage loss cost of category k user at load point j in feeder f, and The feeder f of the k-type user at load point j is before and after the switchgear is installed. c in i c The duration of the power outage caused by the segmentation fault; C RCS =CI RCS +IC RCS +MC RCS ; In the formula, CI RCS is the total investment cost of the switchgear, IC RCS is the total installation cost of the switch, MC RCS is the total maintenance cost of the switch, F is the set of feeders, S is the set of candidate locations, X RCS,f,e is a binary variable of whether to install a switchgear at the candidate position e of feeder f, CI RCS,f,e is the unit investment cost of the switchgear, IC RCS,f,e is the installation cost per switch, DR is the discount rate, MC is RCS,f,e The maintenance cost of the switch unit; Where ξ is the conditional risk value under the confidence level α, η w is a dummy auxiliary variable for scenario w; Step S4: setting constraints, including calculating the load point outage time; If there is an automatic switch between the load point and the fault area, then In the formula, is the load point outage time, T RCS is the action time of the automatic switch; If there is a manual switch between the load point and the fault area, Where, T MCS is the action time of the manual switch, X RCS,f,e A binary variable indicating whether to install an automatic switch at the candidate position e of feeder f, v f A binary variable indicating whether there is a connection between feeder f and other feeders; If there is no switch between the load point and the fault area, then In the formula, The time required to repair the fault; X MS,f,e A binary variable indicating whether a manual switch is installed at the candidate position e of feeder f. Step S5: Solve the objective function to obtain the number, type and installation location of the distribution network switches to be installed.

2. A distribution network switch planning method considering conditional risk value as claimed in claim 1, characterized in that: Step S1 includes the following steps: The sequential Monte Carlo simulation method is used to generate fault scenarios. Step S11: traverse each component, generate a first random number through standard uniform distribution, and combine it with the first probability density function to sample the fault time of the corresponding component; Step S12: traverse the components in the fault state, generate a second random number through standard uniform distribution, and combine it with the second probability density function to sample the repair time of the component; Step S13: repeating steps S11 and S12 until the fault time and the repair time cover the entire planning period; Step S14: Find the first faulty component and determine its repair time, and find the next faulty component after repair, until the entire planning period is covered and combined into a fault scenario; Step S15: Repeat steps S11 and S15 to generate multiple fault scenarios to form an original scenario set Ω D .

3. A distribution network switch planning method considering conditional risk value as claimed in claim 1, characterized in that: Step S2 includes: Step S21: Calculate the weighted distance d of scene w w , In the formula, v w,s is the Euclidean distance between scene w and scene s, v w and v s are the total system costs of scenario w and scenario s respectively, ||·|| represents the Euclidean distance, π s is the occurrence probability of scene s, n is the original scene set Ω D The number of scenes in Step S22: Select a preset number of scenes from small to large according to the weighted distance and add them to the empty scene set Ω S , and from the original scene set Ω D Delete it.

4. A distribution network switch planning method considering conditional risk value as claimed in claim 1, characterized in that: Step S4 also includes: Set switch installation constraints, 5. A distribution network switch planning method considering conditional risk value as claimed in claim 1, characterized in that: Step S4 also includes: Set the switch budget constraint, Where N is the maximum number of switches that the grid company can afford.