Method for screening dynamic reactive power compensation device configuration based on improved entropy weight superior and inferior solution distance method

Abstract

The invention relates to a method for screening dynamic reactive power compensation device configuration based on an improved entropy weight superior and inferior solution distance method, which comprises the following steps of: screening a to-be-configured scheme of a dynamic reactive power compensation device by using an improved entropy weight superior and inferior solution distance method TOPSIS method, and determining a final configuration scheme of the dynamic reactive power compensation device by using a perturbation method. Compared with each optimal compensation scheme under the compensation of a single-type dynamic reactive power compensation device, the optimal compensation schemes in the invention are generally larger in close degree, the economic cost is considered while the reactive power compensation effect is ensured, and the reactive power compensation scheme is better in comprehensiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of reactive power planning for power grids, and in particular to a reactive power planning method that takes transient voltage stability into consideration and is adaptable to large-scale wind and solar power access. Background Art

[0002] As the main body of renewable energy power generation, wind power and photovoltaic power are increasingly connected to the grid, which poses a severe challenge to the transient voltage stability of the power grid. Reactive power planning, as a forward-looking approach to improving the voltage stability of the power grid, has been widely studied by engineers, but most of the current research focuses only on reducing system losses and voltage deviations, mainly solving the problem of static voltage stability, and there are few literatures on transient voltage stability. Therefore, it is urgent to establish a scientific and effective reactive power planning method to improve the robustness of the power grid after a fault.

[0003] At present, the reactive power planning methods considering transient voltage stability of power grid generally have the following three problems:

[0004] 1) The planned dynamic reactive power sources are of a single type, which cannot take into account the economic efficiency while ensuring the system can obtain the best reactive power compensation effect;

[0005] 2) The research background is mostly limited to traditional power grids, without considering the access of new energy sources;

[0006] 3) For power grids containing large-scale renewable energy, existing methods do not take into account the impact of renewable energy uncertainty on planning results.

[0007] Therefore, when planning reactive power for a power grid containing large-scale renewable energy, it is necessary to consider the objective factor of the uncertainty of renewable energy and rationally select dynamic reactive power compensation devices to maximize economic cost savings while ensuring the safe and stable operation of the bus voltage of the entire network. Summary of the invention

[0008] Based on the shortcomings of the reactive power planning methods in the prior art, the present invention provides a reactive power planning method that takes transient voltage stability into consideration and is suitable for large-scale wind and solar access. The method takes into account the uncertainty of large-scale wind power and large-scale photovoltaic output, and the resulting planning method is applicable to all working conditions; and while ensuring the compensation effect, it saves economic costs to a great extent.

[0009] The technical solution adopted by the present invention is:

[0010] A reactive power planning method considering transient voltage stability and adapting to large-scale wind and solar access includes the following steps:

[0011] Step 1: Based on the scenario probability theory, typical scenarios representing the uncertainty of large-scale wind power and large-scale photovoltaic power are constructed in turn;

[0012] Step 2: Propose a transient voltage safety margin index based on voltage binary table to guide reactive power planning and construct a benchmark scenario for reactive power planning;

[0013] Step 3: First, the improved entropy weighted superiority-inferior solution distance method (TOPSIS method) is used to screen the configuration schemes of the dynamic reactive power compensation device, and then the perturbation method is used to determine the final configuration scheme of the dynamic reactive power compensation device.

[0014] In step 1, a typical scenario of large-scale wind power is constructed as follows:

[0015]

[0016] In formula (1), f(v) is the probability density function of wind speed; v is the wind speed; k is the shape parameter of wind speed distribution; c is the scale parameter; and e is a natural constant.

[0017]

[0018] In formula (2), P w is the output power of large-scale wind power; P wr is the rated capacity of large-scale wind power; v ci 、v r 、v co They are the cut-in wind speed, rated wind speed and cut-out wind speed of the wind turbine respectively;

[0019]

[0020] In formula (3), P WT.1 is the probability of large-scale wind power operating under typical wind power scenario 1; f(v) is the probability density function of wind speed.

[0021]

[0022] In formula (4), It is the output power of large-scale wind power in typical wind power scenario 1.

[0023]

[0024] In formula (5), P WT.2 is the probability of large-scale wind power operating under typical wind power scenario 2;

[0025]

[0026] In formula (6), The output power of large-scale wind power when operating in typical wind power scenario 2;

[0027]

[0028] In formula (7), P WT.3 is the probability of large-scale wind power operating under typical wind power scenario 3;

[0029]

[0030] In formula (8), It is the output power of large-scale wind power in typical wind power scenario 3.

[0031] Typical wind power scenario 1 is a planned wind power shutdown scenario; typical wind power scenario 2 is a wind power under-output scenario; typical wind power scenario 3 is a wind power rated output scenario.

[0032] In step 1, the typical scenario of large-scale photovoltaic is constructed as follows:

[0033]

[0034] In formula (9), f(G) is the probability density function of the solar irradiance; G is the solar irradiance; ψ is the standard deviation, and its calculation method is shown in formula (11); γ is the average value, and its calculation method is shown in formula (10);

[0035]

[0036] In formula (10), H is the mean of the logarithmic random variable; E is the variance of the logarithmic random variable.

[0037]

[0038] In formula (12), P p is the output power of large-scale photovoltaics; G is the solar irradiance; R c is the irradiance of a particular sun; P pr is the rated output power of large-scale photovoltaics; G std is the solar irradiance under standard conditions.

[0039] In step 1, typical scenarios of large-scale photovoltaics include: typical photovoltaic scenario 1, typical photovoltaic scenario 2, and typical photovoltaic scenario 3; typical photovoltaic scenario 1 is a planned photovoltaic shutdown scenario; typical photovoltaic scenario 2 is a photovoltaic under-output scenario; and typical photovoltaic scenario 3 is a photovoltaic rated output scenario.

[0040] The probability of large-scale photovoltaic operation in each typical photovoltaic scenario and its active output are shown in formulas (13) to (18):

[0041]

[0042] In formula (13), P PV.1is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 1; R c is the irradiance of a specific sun; f(G) is the probability density function of the sun's irradiance.

[0043]

[0044] In formula (14), is the output power of large-scale photovoltaic operation in typical photovoltaic scenario 1; P pr is the rated output power of large-scale photovoltaics; G std is the solar irradiance under standard environment; ψ is the standard deviation; P p is the output power of large-scale photovoltaics; γ is the average value; e is a natural constant.

[0045]

[0046] In formula (15), P PV.2 is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 2;

[0047]

[0048] In formula (16), The output power of large-scale photovoltaics when operating in typical photovoltaic scenario 2;

[0049]

[0050] In formula (17), P PV.3 is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 3;

[0051]

[0052] In formula (18), It is the output power of large-scale photovoltaic when operating in typical photovoltaic scenario 3.

[0053] In step 2, the transient voltage safety margin index based on the voltage binary table includes the regional voltage qualification rate index shown in formula (19) and the regional voltage stability margin index shown in formula (20);

[0054] Voltage binary meter (V cr ,T cr ) is an evaluation standard to describe the degree of bus deviation. If the bus voltage V i Below the preset threshold V cr The maximum duration T b Not exceeding the specified time T cr When , the bus transient voltage is considered stable, otherwise the transient voltage is unstable.

[0055]

[0056] In formula (19), P a.e is the regional ae voltage qualification rate index after considering M pre-selected fault sets and all N buses in the whole system; M is the total number of pre-selected faults; N is the total number of buses in the whole network; N a.e.b.d is the number of buses with qualified voltage in area ae when type d fault occurs at bus b. When the transient voltage safety margin index η of bus i is i <1, bus i can be considered as a bus with qualified voltage; N a.e is the number of all buses in the region ae; δ d is the weight coefficient of fault d, which is numerically equal to the probability of its occurrence; d is the fault type; b is the number of the bus in area ae.

[0057] Since each typical fault is independent of each other, there are is the probability of fault d occurring at bus b, assuming that the probability of fault d occurring at each bus in the system is equal, that is,

[0058]

[0059] In formula (20), η a.e is the voltage stability margin index of area ae; M is the total number of pre-selected faults; N is the total number of buses in the entire network; N a.e is the number of all buses in the region ae; δ d is the weight coefficient of fault d; is the probability of busbar b having a fault d; η i.d.b B is the transient voltage safety margin index of bus i in area ae when bus b fails d; a.e is the weight of region ae, defined by formula (21).

[0060]

[0061] In formula (21), S a.e is the total apparent power in area ae; L is the total number of areas in the system; L-1 is the remaining area in the system except area ae; e and f are the area numbers. |Z′ a.e-a.f | is the equivalent impedance modulus between area ae and area af, and the calculation method of the equivalent impedance is shown in formula (22).

[0062] Z′ a.e-a.f =Z a.e-a.e +Z a.f-a.f -2Z a.e-a.f (twenty two);

[0063] In formula (22), Z′ a.e-a.f is the equivalent impedance between area ae and area af; Z a.e-a.e is the self-impedance of area ae; Z a.f-a.f is the self-impedance of region af; Z a.e-a.f is the mutual impedance between area ae and area af. When calculating, area ae and area af are regarded as nodes, and the node clusters in the two areas are merged into them.

[0064] In step 2, the construction of the benchmark scenario for reactive power planning is specifically shown in formulas (23) to (24):

[0065]

[0066] In formula (23), B l.h is the system transient voltage safety margin index under typical wind power scenario l and typical photovoltaic scenario h; P u is the probability that the system operates in operation mode u; S is the total number of system operation modes; M is the total number of preselected faults; N is the total number of buses in the entire network; N a.e is the number of all buses in area ae; L is the total number of areas in the system; P WT.l is the probability of the system operating under the typical wind power scenario l; P PV.h is the probability of the system operating under the typical PV scenario h;

[0067] δ u.l.h.d is the weight coefficient of fault d in the system under operating mode u, typical wind power scenario l and typical photovoltaic scenario h; P δu.l.h.d.b To consider the probability of busbar b failing d after the system is in operation mode u, typical wind power scenario l, and typical photovoltaic scenario h;

[0068] η i.u.l.h.d.b It is the transient voltage safety margin index of bus i in the system when bus b fails d after considering the system in operation mode u, typical wind power scenario l and typical photovoltaic scenario h;

[0069] B a.e is the weight of area ae; u is the number of the power system operation mode; i is the number of the bus in the system.

[0070] According to the system transient voltage safety margin index of each scenario, the benchmark scenario for reactive power planning is determined by formula (24):

[0071] J = max{B l.h |l∈{1,2,...,T},h∈{1,2,...,F}} (24);

[0072] In formula (24), J is the determined benchmark scenario for reactive power planning; T is the total number of typical wind power scenarios; F is the total number of typical photovoltaic scenarios; max{B l.h} for each B l.h The maximum value of; l is the number of the typical wind power scene; h is the number of the typical photovoltaic scene.

[0073] In step 2, based on the constructed benchmark scenario of reactive power planning, the most serious fault under the benchmark scenario is selected by formulas (25) to (26):

[0074]

[0075] In formula (25), D d.b is the voltage instability risk index of the system when bus b fails d; d is the weight coefficient of fault d; is the probability of busbar b having fault d; B a.e is the weight of region ae; η i.d.b is the transient voltage safety margin index of bus i in area ae when bus b fails d; L is the total number of areas in the system; N a.e is the number of all buses in area ae.

[0076] ξ=max{D d.b |d∈{1,2,...,M},b∈{1,2,...,N}} (26);

[0077] In formula (26), ξ is the most serious fault in the benchmark scenario of reactive power planning; M is the total number of pre-selected faults; N is the total number of buses in the entire network; max{D d.b} for each D d.b The largest one among .

[0078] In step 3, before the evaluation by the improved entropy weighted superior and inferior solution distance method (TOPSIS method), differentiated compensation is implemented for different candidate buses, with priority given to C SI.bus The busbars with the highest priority in the sequence are equipped with static synchronous compensators (STATCOM), and then C SI.bus The busbar with the lowest ranking is equipped with a static var compensator (SVC), and the optimization model shown in formulas (29) to (31) is established. The MOEA / D algorithm is used to solve its Pareto optimal solution set.

[0079] The improved entropy weighted superior and inferior solution distance method (TOPSIS method) is a decision evaluation method that combines the entropy weight method, the coefficient of variation method and the TOPSIS method, which can provide decision-making basis for decision makers. Among them, the entropy weight method is a method for judging the degree of dispersion of indicators. According to the definition of information entropy, for a certain indicator, the entropy value can be used to judge the degree of dispersion of a certain indicator. The smaller the information entropy value, the greater the degree of dispersion of the indicator, and the greater the influence of the indicator on the comprehensive evaluation (i.e., weight). If the values ​​of a certain indicator are all equal, the indicator will not work in the comprehensive evaluation. Therefore, the entropy weight method can be used to calculate the weight of each indicator to provide a basis for the comprehensive evaluation of multiple indicators. The coefficient of variation method is a method that directly uses the information contained in each indicator to obtain the indicator weight by calculation. The coefficient of variation, also known as the "standard deviation rate", is a statistic that measures the degree of variation of each indicator. When comparing the degree of variation of two or more indicators, if the measurement unit is the same as the mean, the standard deviation can be used directly for comparison. If the unit and (or) the mean are different, the ratio of the standard deviation to the mean (relative value) should be used to compare the degree of variation.

[0080] The TOPSIS method, also known as the superior and inferior solution distance method, is a ranking method based on the degree of proximity between a finite number of evaluation objects and an idealized goal, which can evaluate the relative superiority of existing objects. The basic principle of the TOPSIS method is to rank the evaluation objects by detecting the distance between them and the optimal solution and the worst solution. If the evaluation object is closest to the optimal solution and farthest from the worst solution, it is the best; otherwise, it is not the best. The values ​​of each indicator of the optimal solution reach the optimal value of each evaluation indicator, and the values ​​of each indicator of the worst solution reach the worst value of each evaluation indicator.

[0081] Compared with the traditional entropy weight TOPSIS method, the improved entropy weight TOPSIS method mainly differs in the following aspects:

[0082] (1) In order to avoid the uncertainty caused by the single weighting method in the traditional entropy weight TOPSIS method, the improved entropy weight TOPSIS method uses the entropy weight method and the coefficient of variation method to weight the combination of each evaluation index to fully reflect the objective characteristics of each index data. The specific formula is as follows:

[0083]

[0084] in, for a' xy The average value; s is the total number of evaluation schemes; a' xy is the new value of the indicator.

[0085] At the same time, a' xy The standard deviation S y The calculation formula is as follows:

[0086]

[0087] Among them, a' xy is the new value of the indicator; S y for a' xy The standard deviation of ; s is the total number of evaluation schemes; for a' xy The average value of .

[0088] Coefficient of variation V y The calculation formula is as follows:

[0089]

[0090] in, for a' xy The average value of S y for a' xy The standard deviation of .

[0091] Therefore, each index is based on the coefficient of variation V y The weight W 1y for;

[0092]

[0093] Among them, V y is the coefficient of variation; w is the total number of evaluation indicators.

[0094] Calculate information entropy E y The formula is as follows;

[0095]

[0096] Where s is the total number of evaluation schemes; a' xy is the new value of the indicator.

[0097] Each indicator is based on information entropy E y The weight W 2y for:

[0098]

[0099] Among them, w is the total number of evaluation indicators.

[0100] Therefore, the final weight of each indicator is W y for:

[0101]

[0102] Where, W 1y The coefficient of variation V for each indicator y The weight of 2y Each indicator is based on information entropy E y Based on this, the weight matrix R is shown in the following formula:

[0103] R=(W y ×a' xy ) s×w =(r xy ) s×w

[0104] In the formula, r xy is the index value in the weighted matrix, a' xy is the new value of the indicator.

[0105] (2) In order to prevent the traditional entropy weight TOPSIS method from generating solutions that are close to both the optimal solution and the worst solution, the improved entropy weight TOPSIS method introduces a virtual worst solution. The specific formula is as follows:

[0106]

[0107] In the formula, F x is the closeness of each scheme to be evaluated to the worst ideal scheme; For each scheme to be evaluated and the optimal scheme The Euclidean distance between Each scheme to be evaluated and the worst scheme The Euclidean distance between .

[0108]

[0109] In formulas (29) to (31), f1(x) and f2(x) are two sub-objective functions to be optimized, which represent the dynamic reactive power compensation effect and the economic cost of dynamic reactive power compensation respectively; ω1 and ω2 are the optimization weights of the sub-objective functions, satisfying ω1+ω2=1; 1-P a.e is the voltage instability rate index of area ae; η a.e is the voltage stability margin index of area ae; L is the total number of areas in the system; C SI.bus is the candidate bus set based on SI.i; is a logical variable, with values ​​of 0, 1, and 2, which respectively represent that no dynamic reactive power compensation device is installed at bus i, STATCOM is installed at bus i, and SVC is installed at bus i; The operating life of the dynamic reactive power compensation device; is the reactive power compensation unit price of the dynamic reactive power compensation device; Q i is the reactive power compensation capacity of the corresponding dynamic reactive power compensation device; The installation cost of the dynamic reactive power compensation device;

[0110] when hour,

[0111] when hour, represents the operating life of STATCOM, represents the reactive power compensation unit price of STATCOM, Q i is the reactive power compensation capacity of STATCOM, The installation cost of STATCOM;

[0112] when hour, Indicates the operating life of the SVC, Indicates the reactive power compensation unit price of SVC, Q i is the reactive power compensation capacity of SVC, is the installation cost of SVC; min f means the two sub-objective functions are minimized.

[0113] The present invention provides a reactive power planning method that takes transient voltage stability into consideration and is suitable for large-scale wind and solar access. The technical effects are as follows:

[0114] 1) The reactive power planning of the present invention takes into account the uncertainty of large-scale wind power and large-scale photovoltaic power generation, and the planning scheme is applicable to all working conditions;

[0115] 2) The transient voltage safety margin index based on the voltage binary table proposed in the present invention can accurately evaluate the transient voltage stability of the region and the system and guide reactive power planning.

[0116] 3) The reactive power planning method of the present invention can save economic costs to a great extent while ensuring the compensation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0117] Figure 1 This is the schematic diagram for equivalent impedance calculation.

[0118] Figure 2 Flow chart of the method of the present invention.

[0119] Figure 3 This is the geographical wiring diagram of the power grid in a certain region of a province.

[0120] Figure 4 This is the voltage waveform of the busbar of the entire network under the most serious fault.

[0121] Figure 5 It is the sensitivity index diagram of the candidate bus.

[0122] Figure 6 The reactive power compensation effect diagram of each optimal solution.

[0123] Figure 7 This is the bus voltage waveform of the entire network under the scheme to be configured. DETAILED DESCRIPTION

[0124] A reactive power planning method considering transient voltage stability and adapting to large-scale wind and solar access includes the following steps:

[0125] Step 1: Based on the scenario probability theory, typical scenarios representing the uncertainty of large-scale wind power and large-scale photovoltaic power are constructed in turn;

[0126] Step 2: Propose a transient voltage safety margin index based on voltage binary table to guide reactive power planning and construct a benchmark scenario for reactive power planning;

[0127] Step 3: First, the improved entropy weighted superiority-inferior solution distance method (TOPSIS method) is used to screen the configuration schemes of the dynamic reactive power compensation device, and then the perturbation method is used to determine the final configuration scheme of the dynamic reactive power compensation device.

[0128] The specific steps are as follows:

[0129] 1. Construct typical scenarios of large-scale wind power and large-scale photovoltaic power according to formulas (1) to (18):

[0130]

[0131] In formula (1), f(v) is the probability density function of wind speed; v is the wind speed; k is the shape parameter of wind speed distribution; c is the scale parameter; and e is a natural constant.

[0132]

[0133] In formula (2), P w is the output power of large-scale wind power; P wr is the rated capacity of large-scale wind power; v ci 、v r 、v co They are the cut-in wind speed, rated wind speed and cut-out wind speed of the wind turbine respectively;

[0134]

[0135] In formula (3), P WT.1 is the probability of large-scale wind power operating under typical wind power scenario 1; v ci 、v r 、v co are the cut-in wind speed, rated wind speed and cut-out wind speed of the wind turbine respectively; f(v) is the probability density function of the wind speed.

[0136]

[0137] In formula (4), It is the output power of large-scale wind power in typical wind power scenario 1.

[0138]

[0139] In formula (5), P WT.2 is the probability of large-scale wind power operating under typical wind power scenario 2; v ci 、v r are the cut-in wind speed and rated wind speed of the wind turbine respectively; f(v) is the probability density function of the wind speed.

[0140]

[0141] In formula (6), is the output power of large-scale wind power when operating in typical wind power scenario 2; P w is the output power of large-scale wind power; P wr is the rated capacity of large-scale wind power; c is the scale parameter; k is the shape parameter of wind speed distribution; e is a natural constant.

[0142]

[0143] In formula (7), P WT.3 is the probability of large-scale wind power operating under typical wind power scenario 3; v r 、v co are the rated wind speed and cut-out wind speed of the wind turbine respectively; f(v) is the probability density function of the wind speed.

[0144]

[0145] In formula (8), is the output power of large-scale wind power in typical wind power scenario 3; P wr It is the rated capacity of large-scale wind power.

[0146]

[0147] In formula (9), f(G) is the probability density function of the solar irradiance; G is the solar irradiance; ψ is the standard deviation, and its calculation method is shown in formula (10); γ ​​is the average value, and its calculation method is shown in formula (11); e is a natural constant.

[0148]

[0149] In formula (10), H is the mean of the logarithmic random variable; E is the variance of the logarithmic random variable.

[0150]

[0151] In formula (11), H is the mean of the logarithmic random variable; E is the variance of the logarithmic random variable.

[0152]

[0153] In formula (12), P p is the output power of large-scale photovoltaics; G is the solar irradiance; R c is the irradiance of a particular sun; P pr is the rated output power of large-scale photovoltaics; G std is the solar irradiance under standard conditions.

[0154]

[0155] In formula (13), P PV.1 is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 1; R c is the irradiance of a specific sun; f(G) is the probability density function of the sun's irradiance.

[0156]

[0157] In formula (14), is the output power of large-scale photovoltaic operation in typical photovoltaic scenario 1; R c is the irradiance of a particular sun; P pr is the rated output power of large-scale photovoltaics; G std is the solar irradiance under standard environment; ψ is the standard deviation; P p is the output power of large-scale photovoltaics; γ is the average value; e is a natural constant.

[0158]

[0159] In formula (15), P PV.2 is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 2; R c is the irradiance of a particular sun; G std is the solar irradiance under standard environment; f(G) is the probability density function of the solar irradiance.

[0160]

[0161] In formula (16), is the output power of large-scale photovoltaic operation in typical photovoltaic scenario 2; R c is the irradiance of a particular sun; P pr is the rated output power of large-scale photovoltaics; G std is the solar irradiance under standard environment; ψ is the standard deviation; P p is the output power of large-scale photovoltaics; γ is the average value; e is a natural constant.

[0162]

[0163] In formula (17), P PV.3 is the probability of large-scale photovoltaic operation under typical photovoltaic scenario 3; R c is the irradiance of a particular sun; G std is the solar irradiance under standard environment; f(G) is the probability density function of the solar irradiance.

[0164]

[0165] In formula (18), is the output power of large-scale photovoltaic operation in typical photovoltaic scenario 3; P pr is the rated output power of large-scale photovoltaics.

[0166] 2. Define the regional voltage qualification rate index and the regional voltage stability margin index, as shown in formulas (19) to (20).

[0167]

[0168] In formula (19), P a.e is the regional ae voltage qualification rate index after considering M pre-selected fault sets and all N buses in the whole system; M is the total number of pre-selected faults; N is the total number of buses in the whole network; N a.e.b.d is the number of buses with qualified voltage in area ae when type d fault occurs at bus b. When the transient voltage safety margin index η of bus i is i <1, bus i can be considered as a bus with qualified voltage; N a.e is the number of all buses in the region ae; δ d is the weight coefficient of fault d, which is numerically equal to its probability of occurrence. Since each typical fault is independent of each other, we have is the probability of fault d occurring at bus b. The present invention assumes that the probability of fault d occurring at each bus in the system is equal, that is,

[0169]

[0170] In formula (20), η a.e is the voltage stability margin index of area ae; M is the total number of pre-selected faults; N is the total number of buses in the entire network; N a.e is the number of all buses in the region ae; δ d is the weight coefficient of fault d; is the probability of busbar b having a fault d; η i.d.b B is the transient voltage safety margin index of bus i in area ae when bus b fails d;a.e is the weight of region ae, defined by formula (21).

[0171]

[0172] In formula (21), S a.e is the total apparent power in area ae; L is the total number of areas in the system; L-1 is the remaining area in the system except area ae; |Z′ a.e-a.f | is the equivalent impedance modulus between area ae and area af, and the calculation principle diagram and calculation method of equivalent impedance are as follows: Figure 1 As shown in formula (22).

[0173] Z′ a.e-a.f =Z a.e-a.e +Z a.f-a.f -2Z a.e-a.f (twenty two)

[0174] Figure 1 Middle,U a.e and U a.f are the ground voltages of regions ae and af, respectively. When calculating, the regions are regarded as nodes and the node clusters in the regions are merged into them; U a.e-a.f is the voltage of area ae and area af; I a.e ,I a.f are the currents injected into regions ae and af respectively; Z a.e-a.e is the self-impedance of area ae; Z a.f-a.f is the self-impedance of region af; Z a.e-a.f is the mutual impedance between area ae and area af.

[0175] In formula (22), Z' a.e-a.f is the equivalent impedance between area ae and area af; Z a.e-a.e is the self-impedance of area ae; Z a.f-a.f is the self-impedance of region af; Z a.e-a.f is the mutual impedance between area ae and area af.

[0176] 3. Construct a benchmark scenario for reactive power planning by referring to formulas (23) to (24), and select the most serious fault under the benchmark scenario by formulas (25) to (26).

[0177]

[0178] In formula (23), B l.h is the system transient voltage safety margin index under typical wind power scenario l and typical photovoltaic scenario h; P uis the probability that the system operates in operation mode u; S is the total number of system operation modes; M is the total number of preselected faults; N is the total number of buses in the entire network; N a.e is the number of all buses in area ae; L is the total number of areas in the system; P WT.l is the probability of the system operating under the typical wind power scenario l; P PV.h is the probability of the system operating under the typical photovoltaic scenario h; δ u.l.h.d is the weight coefficient of fault d under the system operation mode u, typical wind power scenario l and typical photovoltaic scenario h; is the probability of busbar b failing d after considering the system in operating mode u, typical wind power scenario l and typical photovoltaic scenario h; η i.u.l.h.d.b B is the transient voltage safety margin index of bus i in the system when bus b fails d after the system is in operation mode u, typical wind power scenario l and typical photovoltaic scenario h; a.e is the weight of region ae.

[0179] According to the system transient voltage safety margin index of each scenario, the benchmark scenario for reactive power planning is determined by formula (24).

[0180] J = max{B l.h |l∈{1,2,...,T},h∈{1,2,...,F}} (24)

[0181] In formula (24), J is the determined benchmark scenario for reactive power planning; T is the total number of typical wind power scenarios; F is the total number of typical photovoltaic scenarios; max{B l.h} for each B l.h The maximum value of .

[0182]

[0183] In formula (25), D d.b is the voltage instability risk index of the system when bus b fails d; d is the weight coefficient of fault d; is the probability of busbar b having fault d; B a.e is the weight of region ae; η i.d.b is the transient voltage safety margin index of bus i in area ae when bus b fails d; L is the total number of areas in the system; N a.e is the number of all buses in area ae.

[0184] ξ=max{D d.b |d∈{1,2,...,M},b∈{1,2,...,N}} (26)

[0185] In formula (26), ξ is the most serious fault in the benchmark scenario of reactive power planning; M is the total number of pre-selected faults; N is the total number of buses in the entire network; max{D d.b} for each D d.b The largest one among .

[0186] 4. Construct a candidate bus set to be compensated, as shown in formulas (27) to (28).

[0187]

[0188] In formula (27), SI.i is the sensitivity index of bus i based on the transient voltage safety margin of the bus; η g0.d is the transient voltage safety margin index of the unstable bus g when a system fault d occurs before the dynamic reactive power compensation device is installed at the bus i; η g.d △Q is the transient voltage safety margin index of bus g when a system fault d occurs after a dynamic reactive power compensation device is installed at bus i; c.i is the capacity of the dynamic reactive power compensation device installed at busbar i; L is the total number of areas in the system; N a.e is the number of all buses in area ae; N a.e.b.d is the number of buses with qualified voltage in area ae when type d fault occurs at bus b.

[0189] C SI.bus ={i|sort{SI.i},i∈{1,2,...,N}} (28)

[0190] In formula (28), C SI.bus is a set of candidate buses based on SI.i; sort{SI.i} is a set of buses arranged in descending order according to their SI.i values; N is the total number of buses in the entire network.

[0191] 5. Implement differentiated compensation for different candidate buses, giving priority to C SI.bus The busbars with the highest priority in the ranking are equipped with STATCOM, and the busbars with the highest priority in the ranking are equipped with STATCOM. SI.bus The buses with the lowest ranking are equipped with SVCs, and the optimization model shown in formulas (29) to (31) is established. The MOEA / D algorithm is used to solve the Pareto optimal solution set.

[0192]

[0193] min f={ω1f1(x),ω2f2(x)} (31)

[0194] In formulas (29) to (31), f1(x) and f2(x) are two sub-objective functions to be optimized, which represent the dynamic reactive power compensation effect and the economic cost of dynamic reactive power compensation respectively; ω1 and ω2 are the optimization weights of the sub-objective functions, satisfying ω1+ω2=1; 1-P a.e is the voltage instability rate index of area ae; η a.e is the voltage stability margin index of area ae; L is the total number of areas in the system; C SI.bus is the candidate bus set based on SI.i; is a logical variable, with values ​​of 0, 1, and 2, which respectively represent that no dynamic reactive power compensation device is installed at bus i, STATCOM is installed at bus i, and SVC is installed at bus i; The operating life of the dynamic reactive power compensation device; is the reactive power compensation unit price of the dynamic reactive power compensation device; Q i is the reactive power compensation capacity of the corresponding dynamic reactive power compensation device; is the installation cost of the dynamic reactive power compensation device; hour, when hour, represents the operating life of STATCOM, represents the reactive power compensation unit price of STATCOM, is the installation cost of STATCOM; hour, Indicates the operating life of the SVC, Indicates the reactive power compensation unit price of SVC, Q i is the reactive power compensation capacity of SVC, is the installation cost of SVC; min f means the two sub-objective functions are minimized.

[0195] The solution steps of the MOEA / D algorithm are as follows:

[0196] 1) Input initial data.

[0197] Set the parameters of the MOEA / D algorithm: population size N p , the neighborhood size D of the weight vector, the probability of parent selection in the neighborhood C R Equivalent control parameters, number of iterations g enmax wait.

[0198] 2) Initialization.

[0199] a) Set the Pareto optimal solution set to an empty set and decompose the multi-objective problem into N P There are single-objective optimization subproblems, and each optimization subproblem can be described by formula (32).

[0200]

[0201] In formula (32), ε is the number of the optimization target, ε = 1, 2; γ is the optimization sub-problem, γ = 1, 2, ..., N P ; g te is the decomposed single-objective optimization subproblem, and the weight vector Corresponding to them respectively, in

[0202] b) Calculate the Euclidean distance between any weight vectors, for each weight vector β γ Find the D weight vectors closest to it and form a neighborhood U(γ)=(γ1,γ2,…,γ D ).

[0203] c) Randomly generate the initial population in the feasible domain Calculate the objective function value for each population

[0204] d) Initialize z = (z1, z2), where

[0205] 3) Evolution.

[0206] a) Replication: for each population γ=1,2,…,Ν P , randomly select two numbers κ and τ from the neighborhood U(γ), and let α κ , α τ This generates a new solution ψ′.

[0207] b) Improvement: A repair and improvement heuristic method is used to generate an improved solution ψ from ψ′.

[0208] c) Update z: If z ε >f ε (ψ), then let z ε =f ε (ψ), ε=1,2.

[0209] d) Update the neighboring population: For any χ∈U(γ), if g te (ψ|β χ ,z)≤g te (α χ |β χ ,z), then let α χ =ψ, f(α χ )=f(ψ), update the Pareto solution set.

[0210] 4) Termination conditions.

[0211] When the number of iterations reaches the set value g enmax When , stop the iteration, otherwise let g en =g en +1, return to step 3).

[0212] 5) Output.

[0213] Output the obtained Pareto optimal solution set and end the program.

[0214] In addition, in order to avoid the MOEA / D algorithm from over-evolving towards targets with large differences, the values ​​of each objective function are normalized according to formula (33).

[0215]

[0216] In formula (33), f′ ε is the normalized value of the objective function; f ε is the true value of the objective function; is the upper limit of the true value of the objective function; is the lower bound of the true value of the objective function.

[0217] 6. In the improved entropy weighted superiority and inferiority solution distance method (TOPSIS method), a virtual worst solution is introduced to evaluate s evaluation solutions and w evaluation indicators. The evaluation steps refer to formulas (34) to (47).

[0218] 1) Use the extreme value method to standardize each indicator.

[0219]

[0220] In formula (34), x is the current evaluation scheme; y is the current evaluation index; a xy is the initial value of the indicator; a' xy is the new value of the indicator; Corresponding to the evaluation index y under a xy The maximum and minimum values ​​of .

[0221] 2) According to a' xy The coefficient of variation V is calculated from the mean and standard deviation of y .

[0222]

[0223] In formula (35), for a' xy The average value; s is the total number of evaluation schemes; a' xy is the new value of the indicator.

[0224]

[0225] In formula (36), S y for a' xy The standard deviation of ; s is the total number of evaluation schemes; a' xy is the new value of the indicator; for a' xy The average value of .

[0226]

[0227] In formula (37), V y is the coefficient of variation; for a' xy The average value of S y for a' xy The standard deviation of .

[0228] 3) Calculate each index based on the coefficient of variation V y The weight value W under 1y .

[0229]

[0230] In formula (38), W 1y The coefficient of variation V for each indicator y The weight of V y is the coefficient of variation; w is the total number of evaluation indicators.

[0231] 4) Calculate the information entropy E of each indicator y .

[0232]

[0233] In formula (39), E y is the information entropy; s is the total number of evaluation schemes; a' xy is the new value of the indicator.

[0234] 5) Calculate each index based on information entropy E y The weight value W under 2y .

[0235]

[0236] In formula (40), W 2y Each indicator is based on information entropy E y The weight of; w is the total number of evaluation indicators; E y is the information entropy.

[0237] 6) Calculate the final weight W y , and assign weights to each indicator.

[0238]

[0239] In formula (41), W y is the final weight of each indicator; W 1y The coefficient of variation V for each indicator y The weight of 2y Each indicator is based on information entropy E y The weight of .

[0240] R=(W y ×a' xy ) s×w =(r xy ) s×w (42)

[0241] In formula (42), R is the weighting matrix; W y is the final weight of each indicator; a' xy is the new value of the indicator; s is the total number of evaluation schemes; w is the total number of evaluation indicators; r xy is the index value in the weighted matrix.

[0242] 7) Determine the best solution and the worst solution

[0243]

[0244] In formula (43), is the optimal solution; r xy is the index value in the weighted matrix; x is the current evaluation scheme; y is the current evaluation index; w is the total number of evaluation indicators; The index r xy The maximum value in .

[0245]

[0246] In formula (44), is the worst solution; r xy is the index value in the weighted matrix; x is the current evaluation scheme; y is the current evaluation index; w is the total number of evaluation indicators; The index r xy The minimum value in .

[0247] 8) Calculate each evaluation scheme and the optimal scheme and the worst solution The Euclidean distance between

[0248]

[0249] In formula (45), For each scheme to be evaluated and the optimal scheme The Euclidean distance between them; y is the current evaluation index; w is the total number of evaluation indicators; is the optimal solution; r xy is the index value in the weighted matrix.

[0250]

[0251] In formula (46), Each scheme to be evaluated and the worst scheme The Euclidean distance between them; y is the current evaluation index; w is the total number of evaluation indicators; is the worst solution; r xy is the index value in the weighted matrix.

[0252] 9) Construct a virtual worst solution and calculate the closeness F of each evaluation solution to the worst ideal solution x .

[0253]

[0254] In formula (47), F x is the closeness of each scheme to be evaluated to the worst ideal scheme; For each scheme to be evaluated and the optimal scheme The Euclidean distance between Each scheme to be evaluated and the worst scheme The Euclidean distance between .

[0255] The scheme with the greatest degree of similarity is selected as the scheme to be configured for the dynamic reactive power compensation device.

[0256] 7. Based on Figure 2 Configure the dynamic reactive power compensation device. The specific configuration process is as follows:

[0257] 1) Calculate the operation probability and output size of large-scale wind power and large-scale photovoltaic power in their respective typical scenarios;

[0258] 2) Perform power flow calculation on the example through PSD-BPA, collect the power flow information of the system, and calculate the weight of each area;

[0259] 3) Perform N-1 fault scan on the system to determine the benchmark scenario for reactive power planning;

[0260] 4) For the most serious fault under the reactive power planning benchmark scenario, dynamic reactive power compensation devices of the same type and capacity are connected to each busbar of the system in turn, and time domain simulation analysis is performed on them;

[0261] 5) Calculate the SI.i value of each bus and construct a candidate bus set based on SI.i;

[0262] 6) Install STATCOM on the bus with the highest ranking in the candidate bus set, and install SVC on the remaining buses in the order of ranking;

[0263] 7) Solve the Pareto optimal solution set according to the MOEA / D algorithm;

[0264] 8) Using the improved entropy weight TOPSIS method to evaluate the various solutions, screen out the solutions to be configured for the dynamic reactive power compensation device;

[0265] 9) Based on the solution to be configured, re-scan the faults of each scenario;

[0266] 10) If the bus voltages of all the scenarios are not unstable after the pre-selected faults, the scheme to be configured is output as the final configuration scheme. Otherwise, for the fault that causes the bus voltage to be unstable and the scenario corresponding to the fault, the compensation capacity of the dynamic reactive power compensation device is increased on the basis of the scheme to be configured, and the location and capacity of the increase are searched by the perturbation method. When the transient voltage of each bus can be maintained stable, the original scheme to be configured is updated and the process returns to step 9).

[0267] The technical effects of the present invention will be further illustrated by examples below:

[0268] The parameters mentioned in the present invention are explained as follows:

[0269] (1): In the example used, the doubly-fed wind turbines in the wind farm and the photovoltaic cells in the photovoltaic power station are of the same model. The rated voltage of a single wind turbine is 690 V and the rated capacity is 1.5 MW. The parameters of its Weibull distribution model are as follows: c = 8.5; k = 2; v ci =3.5m / s; v r =11.5m / s; v co =25m / s. The rated voltage of a single photovoltaic array is 400V, the rated output power is 1.05MW, and its environmental parameters are as follows: ψ=0.6; γ=6; G std =800W / m 2 ; R c =120W / m 2 The probability of large-scale wind power and large-scale photovoltaic power operating in various typical scenarios and their active output calculated by formulas (1) to (18) are shown in Tables 1 and 2 respectively.

[0270] Table 1 Typical wind power scenario parameters

[0271]

[0272] Table 2 Typical photovoltaic scene parameters

[0273]

[0274] (2): Figure 3 The grid shown is divided into area 1 and area 2, corresponding to N a.1 =48, B a.1 =0.66, N a.2 =33, B a.2 =0.34.

[0275] (3): The pre-selected fault set consisting of three-phase short-circuit fault and single-phase grounding fault acts on the system at 0.1s. After the fault occurs, the five frequency protection devices are activated to cut off the fault line. Assume that the probability of each fault occurring at each location in the system under each operating condition is equal, that is, M = 2, δ1 = δ u.l.h.1 =0.07, δ2 = δ u.l.h.2 =0.93,

[0276] (4): Reactive power planning is based on the normal operation of the power system, considering three typical operation scenarios of large-scale wind power and large-scale photovoltaic power, namely S = 1, T = 3, F = 3, P u =1, P WT.1 =0.1561, P WT.2 =0.6837, P WT.3 =0.1602, P PV.1 =0.0216, P PV.2 =0.8514, P PV.1 =0.1270.

[0277] (5): According to the system transient voltage safety margin indicators under each scenario shown in Table 3, scenario 8 is determined as the benchmark scenario for reactive power planning.

[0278] Table 3 System transient voltage safety margin indicators under different scenarios

[0279]

[0280] (6): For the candidate busbars with STATCOM installed, it is required that the guarantee system can obtain the best compensation effect during the transient process. At this time, the optimization weights of the sub-objective function are ω1:ω2=2:1, that is, ω1=0.67, ω2=0.33; while for the candidate busbars with SVC installed, the emphasis is on economic cost, and ω1:ω2=1:2, that is, ω1=0.33, ω2=0.67.

[0281] (7): The economic parameters and investment costs of the dynamic reactive power compensation device are shown in Table 4.

[0282] Table 4 Economic parameters and investment costs of dynamic reactive power compensation device

[0283]

[0284] (8): In the MOEA / D algorithm, N p =100; D = 20; C R =0.5; g enmax =50.

[0285] (9): The perturbation position of the perturbation method is taken as 2, and the perturbation amount is taken as 10Mvar.

[0286] Based on PSD-BPA and MATLAB, Figure 3 The actual power grid shown is used to verify the feasibility of the method of the present invention. The power grid is a receiving-end power grid with a large-scale wind and solar power and an active load of 4545MW. The double-fed wind farms with a total installed capacity of 700MW and 600MW are connected to the system via node 70; the photovoltaic power stations with a total installed capacity of 400MW, 200MW, and 400MW are connected to the grid from nodes 27, 19, and 23 respectively. The results of the system N-1 fault scan show that the three-phase short-circuit fault at the head end of line 70-69 is the most serious fault in the baseline scenario, and it has the greatest impact on the system bus voltage. Figure 4 The bus voltage waveform of the entire network under this fault is given, and the bus voltage of the entire network is unstable at this time.

[0287] When laying out the points, a 400Mvar SVC is used and connected to each node in turn to explore its compensation for the bus voltage. Figure 5 Given C SI.bus Table 5 shows the SI.i value of each busbar in Figure 5 The voltage qualification rate index and voltage stability margin index of each area under the compensation of some nodes ranked higher in the ranking are shown in Figure 2.

[0288] Table 5 Transient voltage safety margin index of the region when compensation is applied to some candidate buses

[0289]

[0290] It can be seen from Table 5 that the dynamic reactive power compensation device is connected to the candidate busbar with the highest ranking, which is beneficial to improving the transient voltage stability of each area. It can be seen that the layout idea of ​​the present invention is scientific and reasonable.

[0291] In view of the computational complexity and representativeness of the examples used, this paper only considers the first 6 candidate buses in the candidate bus set and uses the MOEA / D algorithm to calculate the number of candidate buses installed in Figure 5 The optimal capacity of the dynamic reactive power compensation device at the top 6 candidate buses is solved. Table 6 gives the Pareto optimal solution set solved under the benchmark scenario, where the reactive power compensation effect of each optimal solution is as follows: Figure 6 Table 7 shows the evaluation results of each optimal solution after evaluation by the improved entropy weight TOPSIS method.

[0292] Table 6 Optimal configuration scheme of dynamic reactive power compensation device under the benchmark scenario

[0293]

[0294]

[0295] Table 7 Evaluation results of the optimal solutions under the benchmark scenario

[0296]

[0297] Among the evaluation results of the optimal schemes shown in Table 7, Scheme 5 has the best reactive power compensation effect, but the economic cost of investment is also the highest. Although Scheme 2 achieves the best economic performance, the reactive power compensation effect is not as good as the other schemes. After weighing, Scheme 3 with the highest degree of closeness is selected as the scheme to be configured for the dynamic reactive power compensation device. Figure 7 The voltage waveform of the busbar of the entire network under Scheme 3 is given.

[0298] Figure 7 It is shown that under the compensation of scheme 3, the bus voltages of the system can remain stable during the transient process. To further highlight the advantages of the present invention, it is now compared with the reactive power compensation scheme using only SVC or STATCOM. Table 8 shows the evaluation results of each optimal scheme under compensation of different types of dynamic reactive power compensation devices after evaluation by the improved entropy weight TOPSIS method.

[0299] Table 8 Evaluation results of the optimal compensation schemes under the benchmark scenario

[0300]

[0301] The results in Table 8 show that compared with the optimal solutions under the compensation of a single type of dynamic reactive power compensation device, the optimal compensation solutions in the present invention are generally more similar. While ensuring the reactive power compensation effect, it also takes into account the economic cost, and is a more comprehensive reactive power compensation solution.

[0302] Based on Scheme 3 in Table 7, the fault scan is performed again on the example. The scan results show that in the benchmark scenario, although the proposed configuration scheme can guarantee the transient voltage stability of the system after the pre-selected fault, in some scenarios, the bus voltage in the system still has voltage instability. Table 9 shows the final configuration scheme of the dynamic reactive power compensation device determined by the perturbation method. The transient voltage safety margin index of the system after the final configuration scheme is applied to all scenarios is shown in Table 10.

[0303] Table 9 Final configuration scheme of dynamic reactive power compensation device in the system

[0304]

[0305] Table 10 Transient voltage safety margin indicators of the system in each scenario under the final configuration scheme

[0306]

[0307] It can be seen from Table 10 that the final configuration scheme can effectively guarantee the transient voltage stability of the system after being applied to each scenario, which confirms the universality of the capacity determination strategy of the present invention.

Claims

1. A method for selecting configuration of dynamic reactive power compensation device based on improved entropy weight superior and inferior solution distance method, characterized in that: The following steps are involved: S1: Establish an improved entropy weighted superior and inferior solution distance method, as follows: The entropy weight method and the coefficient of variation method are used to weight the combination of each evaluation index, which is specifically reflected in the following formula: in, for a' xy The average value; s is the total number of evaluation schemes; a' xy is the new value of the indicator; At the same time, a' xy The standard deviation S y The calculation formula is as follows: Among them, a' xy is the new value of the indicator; S y for a' xy The standard deviation of ; s is the total number of evaluation schemes; for a' xy The average value of Coefficient of variation V y The calculation formula is as follows: in, for a' xy The average value of S y for a' xy The standard deviation of Therefore, each index is based on the coefficient of variation V y The weight W 1y for; Among them, V y is the coefficient of variation; w is the total number of evaluation indicators; Calculate information entropy E y The formula is as follows; Where s is the total number of evaluation schemes; a' xy is the new value of the indicator; Each indicator is based on information entropy E y The weight W 2y for: Among them, w is the total number of evaluation indicators; Therefore, the final weight of each indicator is W y for: Where, W 1y The coefficient of variation V for each indicator y The weight of 2y Each indicator is based on information entropy E y The weight of Based on this, the weighted matrix R is shown in the following formula: R=(W y ×a' xy ) s×w =(r xy ) s×w In the formula, r xy is the index value in the weighted matrix, a' xy is the new value of the indicator; Introducing the virtual worst solution, the specific formula is as follows: In the formula, F x is the closeness of each scheme to be evaluated to the worst ideal scheme; For each scheme to be evaluated and the optimal scheme The Euclidean distance between Each scheme to be evaluated and the worst scheme The Euclidean distance between S2: Before the evaluation by the improved entropy weighted superior and inferior solution distance method, differentiated compensation is implemented for different candidate buses, with priority given to C SI.bus The busbars with the highest priority in the sequence are equipped with static synchronous compensators STATCOM, and then C SI.bus Static VAR compensators (SVCs) are installed on the buses ranked later in the ranking, and the optimization model shown in formulas (29) to (31) is established, and the Pareto optimal solution set is solved by the MOEA / D algorithm; min f={ω1f1(x),ω2f2(x)} (31); In formulas (29) to (31), f1(x) and f2(x) are two sub-objective functions to be optimized, which respectively represent the dynamic reactive power compensation effect and the economic cost of dynamic reactive power compensation; ω1 and ω2 are the optimization weights of the sub-objective functions, satisfying ω1+ω2=1; 1-P a.e is the voltage instability rate index of area ae; η a.e is the voltage stability margin index of area ae; L is the total number of areas in the system; C SI.bus is the candidate bus set based on SI.i; is a logical variable, with values ​​of 0, 1, and 2, which respectively represent that no dynamic reactive power compensation device is installed at bus i, STATCOM is installed at bus i, and SVC is installed at bus i; The operating life of the dynamic reactive power compensation device; is the reactive power compensation unit price of the dynamic reactive power compensation device; Q i is the reactive power compensation capacity of the corresponding dynamic reactive power compensation device; The installation cost of the dynamic reactive power compensation device; when hour, when hour, represents the operating life of STATCOM, represents the reactive power compensation unit price of STATCOM, Q i is the reactive power compensation capacity of STATCOM, The installation cost of STATCOM; when hour, Indicates the operating life of the SVC, Indicates the reactive power compensation unit price of SVC, Q i is the reactive power compensation capacity of SVC, is the installation cost of SVC; min f means the two sub-objective functions are minimized.

2. According to claim 1, a reactive power planning method considering transient voltage stability and adapting to large-scale wind and solar access is characterized by: The improved entropy weighted superior-inferior solution distance method introduces a virtual worst solution and calculates the closeness F between each evaluation solution and the worst ideal solution. x ; In formula (47), F x is the closeness of each scheme to be evaluated to the worst ideal scheme; For each scheme to be evaluated and the optimal scheme The Euclidean distance between Each scheme to be evaluated and the worst scheme The Euclidean distance between .

3. A method for configuring a dynamic reactive power compensation device, characterized in that The following steps are involved: Step 1): Calculate the operation probability and output size of large-scale wind power and large-scale photovoltaic in their respective typical scenarios; Step 2): Perform power flow calculation on the example through PSD-BPA, collect the power flow information of the system, and calculate the weight of each area; Step 3): Perform N-1 fault scan on the system to determine the benchmark scenario for reactive power planning; Step 4): For the most serious fault under the reactive power planning benchmark scenario, a dynamic reactive power compensation device of the same type and capacity is connected to each bus of the system in turn, and a time domain simulation analysis is performed on it; Step 5): Calculate the SI.i value of each busbar and construct a candidate busbar set based on SI.i; Step 6): Install STATCOM on the busbars with the highest ranking in the candidate busbar set, and install SVC on the remaining buses in the order of ranking; Step 7): Solve the Pareto optimal solution set according to the MOEA / D algorithm; Step 8): using the improved entropy weight TOPSIS method to evaluate the various solutions, and screen out the solutions to be configured for the dynamic reactive power compensation device; Step 9): Based on the solution to be configured, re-scan the faults of each scenario; Step 10): If the bus voltages of all the scenarios are not unstable after the pre-selected faults, the scheme to be configured is output as the final configuration scheme; otherwise, for the fault that causes the bus voltage instability and the scenario corresponding to the fault, the compensation capacity of the dynamic reactive power compensation device is increased on the basis of the scheme to be configured. When the transient voltage of each bus can be maintained stable, the original scheme to be configured is updated and the process returns to step 9).

Images