A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network considering the access of new energy

Abstract

The present invention discloses a hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account the access of new energy, including: analyzing the correlation between characteristic indicators and evaluation indicators that affect the reactive comprehensive support capability of distribution network, and screening out key indicators that affect the reactive comprehensive support capability of distribution network from characteristic indicators; constructing a reactive comprehensive support capability evaluation model of distribution network according to key indicators, and adjusting the weights of key indicators in the reactive comprehensive support capability evaluation model of distribution network by using dynamic entropy weight method and assigning points; collecting historical data of key indicators of the access of new energy to distribution network, calculating the score under each data section by using the constructed reactive comprehensive support capability evaluation model of distribution network, and clustering by using semi-supervised clustering method, and obtaining the hierarchical evaluation result of reactive comprehensive support capability of distribution network. The present invention realizes the dynamic classification of reactive comprehensive support capability of distribution network, and more scientifically and intuitively reflects the comprehensive reactive level of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of hierarchical evaluation of reactive comprehensive supporting capability of distribution network, and specifically relates to an online hierarchical evaluation method of reactive comprehensive supporting capability of distribution network taking into account access of new energy. Background Art

[0002] The stable operation of the distribution network depends on sufficient reactive power support, which is related to voltage stability and efficient energy transmission. With the massive access of new energy sources such as wind power and photovoltaics, the reactive power management of the distribution network becomes more complicated, because the output of these energy sources is highly dependent on natural conditions and usually cannot provide stable reactive power support. Modern wind power and photovoltaic systems are usually connected to the grid through power electronic devices, which have certain reactive power regulation capabilities. However, these capabilities are limited by equipment design and operating strategies and may not be sufficient to cope with rapidly changing load demands or power generation conditions.

[0003] Due to the uncertainty and intermittency of new energy, as well as the gradual reduction of traditional power generation methods, the distribution network needs a more flexible and intelligent method to evaluate and manage reactive power. Through the hierarchical dynamic evaluation of the reactive comprehensive support capability of the distribution network, the grid operator is allowed to adjust the reactive support strategy according to the current and past comprehensive status of the grid, so that the distribution network can better adapt to the access of new energy, improve its responsiveness to changing conditions and overall operating efficiency. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention proposes a hierarchical dynamic evaluation method for reactive comprehensive support capability of a distribution network taking into account the access of new energy sources.

[0005] To achieve the above object, the present invention adopts the following technical solution: a hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account the access of new energy, specifically comprising the following steps:

[0006] Step S1, performing correlation analysis on characteristic indicators and evaluation indicators that affect the reactive comprehensive support capability of the distribution network, and screening out key indicators that affect the reactive comprehensive support capability of the distribution network from the characteristic indicators;

[0007] Step S2: construct a distribution network reactive power comprehensive support capability evaluation model based on key indicators, and use a dynamic entropy weight method to adjust the weights of key indicators in the distribution network reactive power comprehensive support capability evaluation model and assign scores;

[0008] Step S3: collect historical data of key indicators of new energy access to the distribution network, use the constructed distribution network reactive comprehensive support capability evaluation model to calculate the score of each data section, and use the semi-supervised clustering method to perform clustering to obtain the distribution network reactive comprehensive support capability grading evaluation results.

[0009] Furthermore, step S1 includes the following sub-steps:

[0010] Step S11, calculating the Spearman rank correlation coefficient of each characteristic index affecting the comprehensive reactive support capability of the power grid relative to each evaluation index;

[0011] Step S12, averaging all Spearman rank correlation coefficients of each characteristic index to obtain a comprehensive correlation coefficient of each characteristic index;

[0012] Step S13: sort the absolute values ​​of the comprehensive correlation coefficients of all characteristic indicators in descending order, and take the characteristic indicators whose absolute values ​​of the comprehensive correlation coefficients are higher than the threshold as key indicators.

[0013] Furthermore, it is characterized in that the key indicators are: new energy short-circuit ratio, effective reactive power reserve and reactive output cost.

[0014] Furthermore, it is characterized in that the calculation process of the new energy short-circuit ratio is:

[0015] A. Calculate the expected and standard deviation of active power at each site based on the wind power and photovoltaic output power models;

[0016] B. Use the impedance matrix to obtain the interaction factors between each station, and combine the expected and standard deviation of the active power of each station to calculate the comprehensive capacity of each station.

[0017]

[0018] in, represents the expected active power of the yth station, k P represents the safety factor reflecting probabilistic fluctuations, The standard deviation of the active power of the yth station, WPIF yz represents the interaction factor between the y-th station and the z-th station, P z represents the active power of the zth station;

[0019] C. The ratio of the short-circuit capacity of each station to the comprehensive capacity is taken as the new energy short-circuit ratio.

[0020] Furthermore, the calculation process of the effective reactive power reserve is:

[0021] A. For each reactive source node in the distribution network, according to the probability density function of wind power output, photovoltaic power output and load, randomly extract wind speed, solar irradiance and load values ​​to generate specific scenarios and form a scenario set;

[0022] B. Use the Kmeans algorithm to cluster the scene set, extract the scene parameters of each cluster center, and use the improved continuous power flow method to calculate the PV curve under each cluster center scene;

[0023] C. Obtain the nose point reactive power and the normal operating point reactive power according to the nose point voltage and the normal operating point voltage on each PV curve, and calculate the effective reactive power reserve of each reactive source node;

[0024] D. Sum the effective reactive power reserves of all reactive source nodes to obtain the effective reactive power reserve of the distribution network in each clustering scenario;

[0025] E. Average the effective reactive power reserve of the distribution network under all clustering scenarios to obtain the expected effective reactive power reserve of the distribution network.

[0026] Furthermore, the calculation process of the reactive power output cost is:

[0027]

[0028] Among them, C represents the reactive power output cost, C Q,d represents the reactive power shortage compensation cost of node i, C Q,loss represents the contribution cost of the reactive power source output of node i to the distribution network loss, C Q,PV represents the photovoltaic reactive power production cost, represents the reactive power production cost of wind power, C gci (Q ci ) represents the reactive power production cost of the reactive power compensation equipment at node i, C gqi (Q i ) represents the reactive power production cost of the generator connected to node i.

[0029] Furthermore, the specific process of constructing the reactive comprehensive support capability evaluation model of the distribution network in step S2 is as follows:

[0030] V s =ω1(REGSCR y ) * +ω2(E[Q res ]) * +ω3C *

[0031] Among them, V s Represents the score of the distribution network reactive power comprehensive support capability evaluation model, (REGSCR y ) * Represents the quantitative value of the short-circuit ratio of new energy, ω1 represents (REGSCR y ) * The weight of (E[Q res ]) *Represents the quantitative value of effective reactive power reserve, ω2 represents (E[Q res ]) * The weight, C * represents the quantitative value of reactive power output cost, ω3 represents C * The weight of .

[0032] Furthermore, the specific process of adjusting the weights of key indicators in the distribution network reactive comprehensive support capability evaluation model using the dynamic entropy weight method in step S2 is as follows:

[0033] A. Obtain a cross-sectional data matrix consisting of key indicators within a certain period of time in the distribution network;

[0034] B. Standardize the elements in the cross-sectional data matrix to obtain a standardized matrix;

[0035] C. Calculate the characteristic weight of each key indicator in the standardized matrix under all data sections to obtain the entropy value of the key indicator Where m represents the number of data sections, j * Indicates the key indicator index, Represents the key indicator j * The characteristic proportions in all data sections;

[0036] D. Calculate the comprehensive weight of the key indicator in m data sections according to the entropy value of the key indicator

[0037] E. The comprehensive weight of the key indicators and the calculation weights of the key indicators Perform weight smoothing to obtain the dynamic entropy weight of the key indicator Among them, τ is the smoothing factor.

[0038] Furthermore, the process of assigning points to key indicators in step S2 is as follows:

[0039] The process of assigning points to the quantitative value of the short-circuit ratio of new energy and the quantitative value of the effective reactive power reserve is:

[0040]

[0041] Among them, S a The score representing the quantitative value of the short-circuit ratio of renewable energy or the quantitative value of the effective reactive power reserve, x a Represents the quantitative value of the short-circuit ratio of new energy or the quantitative value of effective reactive power reserve, x max Indicates the upper limit of the score for the corresponding key indicator, x min Indicates the lower limit of the score for the corresponding key indicator;

[0042] The process of assigning points to the quantitative value of reactive power output cost is as follows:

[0043]

[0044] Among them, S cost The score representing the quantitative value of reactive power output cost, x cost It is the quantitative value of reactive power output cost indicator.

[0045] Furthermore, in step S3, when clustering is performed using the semi-supervised clustering method, the must-connect constraint and the no-connect constraint are determined;

[0046] The necessary connection constraints are:

[0047] Data sections where the scores of the three key indicators of the reactive power comprehensive support capability of the distribution network are all higher than 60 points;

[0048] Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only two of them had scores higher than 60 points;

[0049] Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only one data section had a score higher than 60 points;

[0050] Data sections where the scores of the three key indicators of the reactive power comprehensive support capability of the distribution network are all below 60 points;

[0051] The non-connection constraint is: the data section corresponding to the above must-connection constraint and the data section corresponding to other must-connection constraints.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention proposes a mining method for factors affecting the reactive comprehensive support capacity of the distribution network based on historical data, and screens out key indicators by correlation analysis of characteristic indicators and evaluation indicators that affect the reactive comprehensive support capacity of the distribution network, which helps to prepare and optimize reactive resource allocation in advance, thereby improving the responsiveness and efficiency of the distribution network; at the same time, the present invention constructs a reactive comprehensive support capacity evaluation model for the distribution network through key indicators, which can comprehensively consider various factors and provide a comprehensive reactive comprehensive support capacity evaluation, and uses the dynamic entropy weight method to adjust the weights of key indicators in the reactive comprehensive support capacity evaluation model of the distribution network, and ensure that the evaluation model always reflects the most current network status and importance distribution, thereby improving the accuracy and adaptability of the evaluation. The present invention can cope with different levels of distribution network status and needs through a dynamic hierarchical evaluation strategy, with higher flexibility, which helps to improve the stability and reliability of the distribution network and optimize resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of a hierarchical dynamic evaluation method of reactive comprehensive support capability of a distribution network taking into account access to new energy sources according to the present invention;

[0054] Figure 2It is a node PV curve diagram obtained by the present invention using the improved continuous power flow method;

[0055] Figure 3 It is a diagram illustrating the power output partitioning of a photovoltaic inverter according to the present invention;

[0056] Figure 4 It is a curve diagram showing the variation of reactive power production cost of the generator according to the present invention with reactive power output. DETAILED DESCRIPTION

[0057] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0058] like Figure 1 The flowchart of the method for hierarchical dynamic evaluation of reactive comprehensive support capability of distribution network considering the access of new energy sources in the present invention is as follows:

[0059] Step S1: The reactive comprehensive support capability of the distribution network can be reflected in the voltage stability regulation of the distribution network by the reactive source, the reduction of network losses and the increase of the economic benefits of the distribution network. To systematically evaluate the reactive comprehensive support capability of the distribution network, it is necessary to analyze the correlation between the characteristic indicators and evaluation indicators that affect the reactive comprehensive support capability of the distribution network, and select the key indicators that affect the reactive comprehensive support capability of the distribution network from the characteristic indicators, which is helpful to prepare and optimize the reactive resource allocation in advance, thereby improving the responsiveness and efficiency of the distribution network. It specifically includes the following sub-steps:

[0060] Step S11: For the selection of indicators representing the overall capacity of the distribution network, in order to take into account the safety situation and economic benefits of the distribution network, an evaluation indicator data set is established, and the voltage stability margin index and the network loss index are selected as the constituent elements; for the selection of features representing the reactive comprehensive support capacity of the distribution network, select from the indicators that affect the reactive output capacity of the distribution network to establish a feature indicator data set, including reactive power reserve, short-circuit ratio, reactive power demand, voltage deviation, power factor, reactive compensation device capacity, voltage stability index, reactive power factor, system impedance, reactive production cost, generator reactive output capacity, etc. Calculate the Spearman rank correlation coefficient of each feature indicator that affects the reactive comprehensive support capacity of the power grid relative to each evaluation indicator d c is the rank difference between the two characteristic indices and the evaluation index, and e is the data point;

[0061] Step S12, averaging all Spearman rank correlation coefficients of each characteristic index to obtain a comprehensive correlation coefficient of each characteristic index;

[0062] Step S13, sort the absolute values ​​of the comprehensive correlation coefficients of all characteristic indicators in descending order, a high correlation indicates that the characteristic indicator has a strong connection with the voltage stability margin and the network loss indicator, and the characteristic indicator whose absolute value of the comprehensive correlation coefficient is higher than the threshold is taken as the key indicator. In the present invention, the threshold is set to 0.2, and the key indicators are found to be: short-circuit ratio, reactive power reserve and reactive production cost. These three key indicators are used as the differentiated embodiment of the reactive comprehensive support capability of the distribution network, and the three key indicators are used to construct the reactive comprehensive support capability evaluation system of the distribution network.

[0063] The short-circuit ratio is a quantitative indicator that describes the strength of voltage support, indicating the response capability of the distribution network to the injection and absorption of active and reactive power. However, there are certain deficiencies in simply using the short-circuit ratio indicator to evaluate the voltage support capability and stability of new energy sites such as wind farms and photovoltaic power stations. The short-circuit ratio is often calculated based on the ratio of the traditional short-circuit capacity to the rated capacity, without fully considering the randomness of wind and solar power generation, the response characteristics of reactive power, and the complex interactions between various new energy sites. In order to more accurately reflect the voltage support capability and system stability of new energy sites, the present invention adopts the new energy short-circuit ratio indicator, which can more accurately evaluate the impact of new energy access on the stability of the power grid.

[0064] The calculation process of the new energy short-circuit ratio in the present invention is:

[0065] A. Constructing a wind and solar uncertainty probability model: The output power of a wind turbine is closely related to the wind speed. In order to construct a wind power output probability model, the wind speed needs to be fitted first. Weibull distribution, as a two-parameter distribution theory, can fit the wind speed well. Its probability density function is:

[0066]

[0067] Where k is the shape parameter of the Weibull distribution, c is the scale parameter of the Weibull distribution, and v is the wind speed;

[0068] The above two Weibull distribution parameters can be obtained from the average wind speed μ and standard deviation σ of historical data:

[0069]

[0070]

[0071] Among them, Γ is the Gamma function;

[0072] From the probability density function of wind speed, we can approximate the functional relationship between wind turbine output power and wind speed:

[0073]

[0074] Among them, v in 、v r and v out are cut-in wind speed, rated wind speed and cut-out wind speed respectively, P r is the rated power of the wind turbine;

[0075] Photovoltaic power generation output is closely related to solar irradiance. Solar irradiance is directly affected by weather conditions and installation location. Its size obeys Beta distribution. The probability density function of solar irradiance can be constructed as:

[0076]

[0077] Among them, S s and S max are the actual solar irradiance and the maximum solar irradiance, α and β are the shape parameters of the Beta distribution, and Γ is the Gamma function;

[0078] The shape parameters of the Beta distribution can be obtained from the average solar irradiance μ′ and variance σ′ over a given period of time:

[0079]

[0080] The probability density function of photovoltaic power generation output and solar irradiance is:

[0081]

[0082] Among them, P s is the photovoltaic array output power, P s =S s A v η′,A v is the total area of ​​the solar cell array, and η′ is the light energy utilization rate.

[0083] B. Calculate the expected active power of each station y based on the wind power and photovoltaic output power model and standard deviation

[0084]

[0085] Among them, P y is the active power of node y.

[0086] C. Use the impedance matrix to obtain the interaction factor WPIF between each station yz , combined with the expected and standard deviation of active power at each station, calculate the comprehensive capacity of each station

[0087]

[0088] in, represents the expected active power of the yth station, k P represents the safety factor reflecting probabilistic fluctuations, The standard deviation of the active power of the yth station, WPIF yz represents the interaction factor between the y-th station and the z-th station, P z represents the active power of the zth station, Z eq,yy , Z eq,yz is the element in the impedance matrix related to stations y and z, U y and U z is the voltage at stations y and z;

[0089] D. Set the short-circuit capacity S of each station ac,y The ratio to the comprehensive capacity is taken as the new energy short-circuit ratio:

[0090]

[0091] Determining the amount of reactive power reserve available in the system is key to ensuring voltage stability and meeting sudden demands. By real-time monitoring and calculating reactive power reserves, the power grid can respond more flexibly to load changes and avoid voltage problems. The calculation process of the effective reactive power reserve in the present invention is:

[0092] The system power distribution load changes over time. In order to systematically reflect the load changes, it is necessary to perform probabilistic modeling on the load and use normal distribution to fit the load changes:

[0093]

[0094]

[0095] Among them, μ P , σ P and μ Q , σ Q are the real and imaginary parameters of the load respectively.

[0096] A. For each reactive source node in the distribution network, according to the probability density function of wind power output, photovoltaic power output and load, randomly extract wind speed, solar irradiance and load values ​​to generate specific scenarios and form a scenario set;

[0097] B. Use Kmeans algorithm to cluster the scene set, extract the scene parameters of each cluster center, and use the improved continuous power flow method to calculate the PV curve of each cluster center scene; specifically,

[0098] (1) Define two sets X * and Y *, where X * is the distribution network state variable vector, which consists of the voltage amplitude and phase angle of the PQ node and the phase angle of the PV node, and is expressed as:

[0099] X * =[θ1,θ2,…,θ n-1 ,V1,V2,…,V m ] T

[0100] Y * is the distribution network parameter variable vector, which is composed of the active and reactive power of the PQ nodes and the active power of the PV nodes, and is expressed as:

[0101] Y * =[P 1s ,Q 1s ,P 2s ,Q 2s ,…] T

[0102] (2) Assume that the load of the distribution network increases in a certain way, then Y at a certain time * The matrix can be represented as:

[0103] Y * 1=Y * +λY d

[0104] Among them, Y d =[P 1d ,Q 1d ,P 2d ,Q 2d ,…] T , λ is the variation parameter of the distribution network;

[0105] Y * 1 Bring in X * The mathematical model of the system at the steady-state operating point is obtained by the expression:

[0106] f(X * ,λ)=0

[0107] (3) Using conventional power flow calculation to obtain X * The initial parameter value is X * =X * 0, λ=λ0=0; put the above function in By performing Taylor series expansion, we can find the point (X * ,λ) is expressed as:

[0108]

[0109] Where J is the Jacobian matrix of the conventional power flow equation;

[0110] The solution method for the prediction point is:

[0111]

[0112] Among them, h is the step length. The selection of the step length is related to the curvature of the curve. When the curvature of the curve is large, the step length is small, and when the curvature of the curve is small, the step length can be appropriately increased.

[0113] The exact solution of the prediction point is:

[0114] Passing prediction point Make a point The intersection of the vertical hyperplane and the function curve is the exact solution. Using the above method to draw the PV curve of a specific scenario, such as Figure 2 shown.

[0115] C. During the actual operation of the distribution network, due to the influence of load conditions and external environment, the distribution network usually collapses before the reactive source outputs to the upper limit of the reactive output of the reactive source. Therefore, choosing to use effective reactive reserves can better reflect the actual reactive support of the distribution network. The definition of effective reactive reserves is the difference between the reactive power generated by each reactive source at the operating point of the distribution network system and the reactive power generated by each reactive source when voltage collapse occurs. The steps to calculate the effective reactive reserve using the PV curve are: determine the position of the nose point of the PV curve. The nose point of the PV curve is the inflection point of the curve, that is, after this point, the voltage drops rapidly as the power increases. The nose point reactive power Q is obtained based on the nose point voltage on each PV curve and the normal operating point voltage u，c and the reactive power Q at the normal operating point u，n , calculate the effective reactive power reserve Q of each reactive power source node u，res =Q u，c -Q u，n ;

[0116] D. Sum the effective reactive power reserves of all reactive source nodes to obtain the effective reactive power reserve of the distribution network in each clustering scenario;

[0117] E. Average the effective reactive power reserve of the distribution network under all clustering scenarios to obtain the expected effective reactive power reserve of the distribution network.

[0118] Reactive power output cost can optimize the economic use of resources and ensure sufficient reactive power support on a cost-effective basis, thereby improving the economic operation efficiency of the power grid. The calculation process of reactive power output cost is:

[0119] (1) Calculation of reactive power shortage compensation cost: The reactive power margin of a distribution network node is a quantitative indicator that reflects the degree of demand for reactive power compensation at that node. The reactive power margin can be calculated from the expected reactive power shortage value E of node i: Q,d reflect:

[0120]

[0121] Among them, Q Qij is the reactive load removed due to reactive power shortage at node i in state j, N C is the sum of the operating states, p j is the active power value in state j,

[0122] Assume that the conversion coefficient of reactive power deficiency compensation cost of node i is η1, then the reactive power deficiency compensation cost C of the node is Q,d =η1E Q,d .

[0123] (2) Calculate the contribution cost of reactive power output to system network loss: The impact of reactive power injection of different reactive sources on system network loss is different. The present invention analyzes the sensitivity of reactive power output of reactive sources by calculating the change in system network loss before and after reactive power input of reactive sources, so as to evaluate its contribution to the reduction of system network loss. The reduction in system loss before and after the reactive source at node i participates in system regulation can be expressed as:

[0124] ΔP loss (Q gi )=P loss (Q gi ′)-P loss (Q gi )

[0125] Among them, Q gi ′ represents the reactive power output of the reactive source at node i before regulation, P loss (Q gi ′) is the network loss before reactive power source participates in regulation, Q gi It represents the reactive power output of the reactive source at node i after the reactive source participates in the regulation, ΔP loss (Q gi ) indicates that the reactive power output of reactive source node i is Q gi The amount of network loss reduction.

[0126] The calculation formula for the sensitivity of the network loss change before and after the reactive source participates in reactive power regulation to the reactive power output of the reactive source is:

[0127]

[0128] From the above formula, we can know that the sensitivity of the change in network loss before and after the reactive source participates in reactive power regulation to the reactive power output of the reactive source is the inverse of the sensitivity of network loss to reactive power, which has nothing to do with the reactive power output of the reactive source before regulation and the network loss at that time. Therefore, the value can be obtained by first calculating the sensitivity of network loss to reactive power. The expression of system network loss can be deduced from the sensitivity of network loss to reactive power output of reactive source:

[0129]

[0130] Among them, U i , U j are the voltage values ​​of nodes i and j respectively, G ij is the conductance value of nodes i and j, θ ij is the phase angle difference between nodes i and j;

[0131] The node reactive power, node voltage amplitude, and node voltage phase angle are expressed in vector form as follows:

[0132]

[0133] The sensitivity of system network loss to reactive power output of reactive source can be expressed as:

[0134]

[0135] After conversion to matrix form, the expression is:

[0136]

[0137] Among them, the sensitivity vector of system network loss to reactive power output of reactive source is:

[0138]

[0139] Where: S Q It is the sub-matrix in the inverse matrix of the sensitivity matrix form of the system network loss to the reactive power output of the reactive source;

[0140] From the above formula, we can get the sensitivity of the network loss change before and after the reactive power source of node i is put into use to the reactive power output of the reactive source:

[0141] Assume that the cost conversion coefficient of reactive power source output contribution to system network loss is η2, then the cost contribution of reactive power source output of node i to system network loss is C Q,loss =η2k i .

[0142] (3) Photovoltaic reactive power production cost: The photovoltaic reactive power production cost mainly comes from the reactive power output cost of the photovoltaic inverter. The cost calculation is mainly based on the inverter power output partition model. The power output characteristics of the inverter power output partition model include four areas, such as Figure 3As shown, where:

[0143] OBC area: PV inverters absorb reactive power and generate active power, but reactive power output does not affect active power output. The reactive power in this area is used to reduce the voltage of the system when reactive power is in excess. At this time, the reactive power production cost of the inverter needs to be paid, but no opportunity cost needs to be paid;

[0144] OAB area: PV inverters generate active power and absorb reactive power. This area is where the reactive power in the power system is seriously oversupplied. In order to absorb more reactive power, PV inverters compress the active power output, which not only increases the operating cost, but also loses the active power income. Therefore, the reactive power cost of this part should include these two aspects;

[0145] OCD: PV inverters generate both active power and reactive power, but reactive power output does not affect active power output. The reactive power in this area is used to increase the voltage of the system when reactive power is scarce, and to ensure that the voltage amplitude of the grid is reasonable when the load is low. At this time, the reactive power production cost of the inverter needs to be paid, but the opportunity cost does not need to be paid;

[0146] ODE: PV inverters generate active power and reactive power. This area is in the extreme scenario where the reactive power of the power system is extremely scarce. In order to generate more reactive power, PV inverters compress the active power output. In addition to increasing the operating cost, it also includes the opportunity cost of compressing the active power output. Therefore, the reactive power cost of this part should include these two aspects.

[0147] Combined with the above process analysis, the photovoltaic production cost should be:

[0148]

[0149] Among them, C Q,PV The reactive power cost for the PV inverter is a OAB , a OBD , a ODE are the reactive power cost coefficients of the corresponding areas, Q PV,i is the reactive power value generated by the photovoltaic inverter, b is the cost coefficient of compressed active power, ΔP PV,i,loss The amount of reactive power reduced in order to increase reactive power output.

[0150] (4) Wind power reactive power production cost: There are four types of wind turbines, and the most widely used type is DFIG (doubly-fed induction generator). The reactive power cost of DFIG is mainly due to the loss caused by its increased reactive power, which is mainly composed of mechanical structure loss and power electronic device loss. Since the grid-side converter works in a power factor of 1 mode, there is no reactive power involved, so the operation loss of the grid-side converter can be ignored. Therefore, the loss in the power electronic device only considers the machine-side converter.

[0151] The loss expression of the machine-side converter is:

[0152] P L,con = l 0,PW +l v,PW |S PW |+l p,PW |S PW | 2

[0153]

[0154] Among them, S PW Indicates the apparent power of DFIG, U PW ,I PW Represent the voltage and current of DFIG, P L,con is the total active power loss on the machine side, which has a quadratic function relationship with the apparent power of the doubly fed wind turbine. 0,PW , l v,PW , l p,PW is the coefficient in this functional relationship;

[0155] Mechanical loss is mainly copper loss and iron loss of winding. According to the mathematical model of DFIG:

[0156] P r =sP s

[0157] P PW =P s -P r

[0158] Among them, P s , P r , P PW are respectively the active power on the stator side, the active power on the rotor side, and the active power output of DFIG; s is the slip rate, which is converted to:

[0159]

[0160] Q s =Q PW

[0161] Among them, Q s Indicates the reactive power on the stator side; Q PW Provide reactive power for DFIG;

[0162] Convert the rotor voltage and current to the stator side:

[0163]

[0164] Among them, U s is the actual stator voltage, I s is the actual stator current, Z'r is the converted rotor impedance, Z m is the mutual inductance impedance, Z' m is the converted mutual inductance impedance, Z s is the actual stator impedance;

[0165] So the stator current is:

[0166]

[0167] Copper loss P L,Cu and iron loss P L,Fe The expression is:

[0168] P L,Cu =3R s |I s | 2 +3R' r |I' r | 2

[0169] P L,Fe =3R m |I s +I' r | 2

[0170] Among them, R s is the resistance value on the stator side, R' r is the rotor side resistance converted to the stator side;

[0171] The additional operating cost of DFIG due to the additional reactive power generation, i.e. the reactive power loss of the unit ΔP L,PW for:

[0172]

[0173] Where: It is the loss when active and reactive power are output simultaneously; The reactive power loss is the difference between the two.

[0174] Therefore, the calculation formula for the reactive power production cost of the double-fed wind turbine is:

[0175]

[0176] in, The cost of DFIG outputting reactive power; FIT PW is the on-grid electricity price of wind power; PW Wind power conversion factor.

[0177] (5) Reactive cost of reactive power compensation equipment: Since reactive power compensation equipment such as SCB, SVC, and SVG only output reactive power, they do not have the opportunity cost of generators. In addition, the operating costs of switching capacitors, adjusting SVCs, and SVGs are basically negligible compared to the investment costs. Therefore, the reactive power production costs of such reactive power equipment are mainly calculated based on the investment costs of purchasing and installing these reactive power equipment, and the depreciation of the equipment is mainly considered in the calculation. The reactive power cost of reactive power equipment is expressed as:

[0178]

[0179] Among them, C gci (Q ci ) is the reactive power production cost of traditional reactive power equipment, ε represents the investment per unit capacity of reactive power equipment, and its value mainly depends on the type of reactive power equipment, Q ci represents the reactive power output by the equipment, y is the expected payback period of the equipment, and f u Indicates the average utilization of the device.

[0180] (6) Generator reactive power production cost: Synchronous generators generate reactive power costs when providing reactive power. Unlike the cost of active output, their unit power cost is much smaller. From the operating characteristics of the generator, it can be seen that its output of active and reactive power is simultaneous, and it is an ideal dynamically and continuously adjustable reactive power source. However, in order to ensure the output of active power and act as a reactive power source, the generator needs to increase its total capacity, which will bring investment costs, operation and maintenance costs, loss costs and opportunity costs. The curve of the reactive power production cost of the generator changing with the reactive output is shown in the figure below: Figure 4 As shown. Among them:

[0181] a) In [0,Q BS ], the small amount of reactive power output by the generator is to meet its own reactive power demand.

[0182] b) In [Q BS ,Q A ], the additional reactive power supplied to the grid will increase the loss cost.

[0183] c) In [Q A ,Q B ], during the period when the reactive power of the power grid is extremely scarce, the generator reduces the active output in order to provide more reactive support, which increases the opportunity cost. At the same time, there is a loss cost, so the slope of the curve gradually increases.

[0184] d) In [Q min ,0], the generator operates in the leading phase mode, absorbing excess reactive power to reduce the system voltage, causing the unit to generate temperature rise, which affects the service life of the generator and also incurs loss costs.

[0185] Therefore, the reactive power production cost of the generator can be approximately expressed as:

[0186]

[0187] C gqi (Q i )=C q (Q i )+C loss (Q i )+C opp (Q i )

[0188] Among them, C gqi (Q i ) represent the reactive power cost of the generator connected to node i, C q (Q i ) is the investment cost, C loss (Q i ) is the loss cost, C opp (Q i ) is the operation and maintenance cost.

[0189] In summary, the mathematical expression of reactive power output cost is:

[0190]

[0191] Among them, C represents the reactive power output cost, C Q,d represents the reactive power shortage compensation cost of node i, C Q,loss represents the contribution cost of the reactive power source output of node i to the distribution network loss, C Q,PV represents the photovoltaic reactive power production cost, represents the reactive power production cost of wind power, C gci (Q ci ) represents the reactive power production cost of the reactive power compensation equipment at node i, C gqi (Q i ) represents the reactive power production cost of the generator connected to node i.

[0192] Step S2: construct a distribution network reactive comprehensive support capability evaluation model based on key indicators, and use the dynamic entropy weight method to adjust the weights of key indicators in the distribution network reactive comprehensive support capability evaluation model and assign scores to ensure that the evaluation model always reflects the most current network status and importance distribution, thereby improving the accuracy and adaptability of the evaluation. By assigning scores to each participating factor of the evaluation model, each indicator can be numerically quantified to obtain the evaluation value of the reactive comprehensive support capability, providing a basis for subsequent classification.

[0193] The specific process of constructing the evaluation model of reactive comprehensive support capacity of distribution network is as follows:

[0194] Vs =ω1(REGSCR y ) * +ω2(E[Q res ]) * +ω3C *

[0195] Among them, V s Represents the score of the distribution network reactive power comprehensive support capability evaluation model, (REGSCR y ) * Represents the quantitative value of the short-circuit ratio of new energy, ω1 represents (REGSCR y ) * The weight of (E[Q res ]) * Represents the quantitative value of effective reactive power reserve, ω2 represents (E[Q res ]) * The weight, C * represents the quantitative value of reactive power output cost, ω3 represents C * The weight of .

[0196] The specific process of using the dynamic entropy weight method to adjust the key indicator weights in the distribution network reactive comprehensive support capacity evaluation model is as follows:

[0197] A. Obtain a cross-sectional data matrix consisting of key indicators within a certain period of time in the distribution network;

[0198] B. Standardize the elements in the cross-section data matrix to obtain a standardized matrix:

[0199] The standardization process of the new energy short-circuit ratio and effective reactive power reserve elements in the cross-section data matrix is ​​as follows:

[0200]

[0201] The standardization process of reactive power cost elements in the cross-section data matrix is ​​as follows:

[0202]

[0203] in, They are the corresponding elements in the cross-sectional data matrix A and the standardized matrix B respectively.

[0204] C. Calculate the characteristic weight of each key indicator in the standardized matrix under all data sections to obtain the entropy value of the key indicator Where m represents the number of data sections, j * Indicates the key indicator index, Represents the key indicator j * The characteristic proportions under all data sections,

[0205] D. Calculate the comprehensive weight of the key indicator in m data sections according to the entropy value of the key indicator

[0206] E. The comprehensive weight of the key indicators and the calculation weights of the key indicators Perform weight smoothing to obtain the dynamic entropy weight of the key indicator Among them, τ is the smoothing factor.

[0207] The process of assigning points to key indicators is as follows:

[0208] The three key indicators are divided into benefit indicators and cost indicators. The benefit indicators are effective reactive power reserve and new energy short-circuit ratio indicators, and the cost indicators are reactive power output costs. The larger the benefit indicator value, the higher the score, and the larger the cost indicator value, the lower the score. The process of assigning points to the quantitative value of the new energy short-circuit ratio and the quantitative value of the effective reactive power reserve is:

[0209]

[0210] Among them, S a The score representing the quantitative value of the short-circuit ratio of renewable energy or the quantitative value of the effective reactive power reserve, x a Represents the quantitative value of the short-circuit ratio of new energy or the quantitative value of effective reactive power reserve, x max Indicates the upper limit of the score for the corresponding key indicator, x min Indicates the lower limit of the score for the corresponding key indicator;

[0211] The process of assigning points to the quantitative value of reactive power output cost is as follows:

[0212]

[0213] Among them, S cost The score representing the quantitative value of reactive power output cost, x cost It is the quantitative value of reactive power output cost indicator.

[0214] Step S3: collect historical data of key indicators of new energy access to the distribution network, use the constructed distribution network reactive comprehensive support capacity evaluation model to calculate the score of each data section, and use the semi-supervised clustering method to perform clustering to obtain the classification evaluation results of the reactive comprehensive support capacity of the distribution network. The semi-supervised clustering method can provide favorable real data support for the classification strategy, making the clustering results more credible; specifically, it includes the following sub-steps:

[0215] S31. Collect historical data of key indicators of new energy access to the distribution network and obtain a distribution network sample data set S n ={s1,s2,…,sn}, where n is the number of sample data;

[0216] S32. Perform cluster analysis on the data set samples. The steps of clustering using the semi-supervised clustering method are as follows:

[0217] Define the must-connect constraint and the no-connect constraint. The must-connect constraint means that some data points must be in the same cluster, and the no-connect constraint means that some data points must not be in the same cluster.

[0218] Set the number of iterations g * =1, the threshold is M. Divide the data samples into g clusters, namely C g =[C1,C2,…,C g ]. Randomly select g samples from g clusters as the initial cluster center vectors [μ1,μ2,…,μ g ].

[0219] Initialize the cluster, for (t=1,2,...,g), let Calculate the data sample S i With each cluster center vector μ t The Euclidean distance d(S) of (t=1,2,...,g) i , μ t ), mark and sample S i The cluster center vector μ with the smallest distance t , and update cluster C t (t=1,2,...,g).

[0220] Recalculate cluster C according to the cluster center calculation formula t The cluster center μ t * :

[0221]

[0222] Determine the new cluster center μ t * With each cluster center vector μ t If the distance is greater than the threshold M, then the number of iterations g * =g * +1, otherwise the cluster is output. After the iteration is completed, the cluster partition C is output. g =[C1,C2,…,C g ].

[0223] S83. Take g as 4 and give the classification standard of the reactive comprehensive support capability of the distribution network:

[0224] Level 1: The reactive power support capability is very strong and can effectively cope with a wide range of load changes and system disturbances. The voltage level is very stable, the economic benefits are good, and the reactive power supply and regulation capabilities far exceed the demand.

[0225] Level 2: Reactive power support capability is strong and can cope with common load changes and system disturbances, but there may be slight voltage problems under extreme conditions. The voltage level is stable most of the time, the economic benefits are average, and the reactive power supply and regulation capabilities meet normal needs.

[0226] Level 3: The reactive power support capacity is basically sufficient, but it may not be able to fully maintain voltage stability under load peaks or specific disturbances. The voltage level is stable in most cases, and the economic benefits are poor, but additional reactive power support may be required under high load or large disturbances.

[0227] Level 4: Insufficient reactive power support capacity, frequent voltage instability, and slow system response to disturbances. Large voltage fluctuations, tight reactive power resources, and poor economic benefits require additional reactive power compensation equipment or adjustment strategies.

[0228] The necessary constraints in the present invention are:

[0229] Data sections where the scores of the three key indicators of the reactive power comprehensive support capability of the distribution network are all higher than 60 points;

[0230] Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only two of them had scores higher than 60 points;

[0231] Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only one data section had a score higher than 60 points;

[0232] Data sections where the scores of the three key indicators of the distribution network's comprehensive reactive support capability are all below 60 points.

[0233] The no-connection constraint in the present invention is: the data section corresponding to the above must-connection constraint and the data section corresponding to other must-connection constraints.

[0234] In summary, the method for hierarchical dynamic evaluation of the reactive comprehensive support capacity of the distribution network taking into account the access of new energy sources can cope with different levels of grid conditions and demands through a dynamic grading strategy. The flexibility of this strategy helps to improve the stability and reliability of the grid while optimizing resource utilization.

[0235] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0236] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0237] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network considering access to new energy sources, characterized in that: The specific steps include: Step S1, performing correlation analysis on characteristic indicators and evaluation indicators that affect the reactive comprehensive support capability of the distribution network, and selecting key indicators that affect the reactive comprehensive support capability of the distribution network from the characteristic indicators; the key indicators are: new energy short-circuit ratio, effective reactive power reserve and reactive output cost; The calculation process of the new energy short-circuit ratio includes: using the impedance matrix to obtain the interaction factor between each station, combining the expectation and standard deviation of the active power of each station, and calculating the comprehensive capacity of each station in, represents the expected active power of the yth station, k P represents the safety factor reflecting probabilistic fluctuations, The standard deviation of the active power of the yth station, WPIF yz represents the interaction factor between the y-th station and the z-th station, P z represents the active power of the zth station; Step S2: construct a distribution network reactive power comprehensive support capability evaluation model based on key indicators, and use a dynamic entropy weight method to adjust the weights of key indicators in the distribution network reactive power comprehensive support capability evaluation model and assign scores; The specific process of using the dynamic entropy weight method to adjust the key indicator weights in the distribution network reactive comprehensive support capacity evaluation model is as follows: A. Obtain a cross-sectional data matrix consisting of key indicators within a certain period of time in the distribution network; B. Standardize the elements in the cross-sectional data matrix to obtain a standardized matrix; C. Calculate the characteristic weight of each key indicator in the standardized matrix under all data sections to obtain the entropy value of the key indicator; D. Calculate the comprehensive weight of the key indicator in m data sections according to the entropy value of the key indicator; E. The comprehensive weight of the key indicators and the calculation weights of the key indicators Perform weight smoothing to obtain the dynamic entropy weight of the key indicator Among them, τ is the smoothing factor; Step S3: collect historical data of key indicators of new energy access to the distribution network, use the constructed distribution network reactive comprehensive support capability evaluation model to calculate the score of each data section, and use the semi-supervised clustering method to perform clustering to obtain the distribution network reactive comprehensive support capability grading evaluation results.

2. According to claim 1, a hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access is characterized in that: Step S1 includes the following sub-steps: Step S11, calculating the Spearman rank correlation coefficient of each characteristic index affecting the comprehensive reactive support capability of the power grid relative to each evaluation index; Step S12, averaging all Spearman rank correlation coefficients of each characteristic index to obtain a comprehensive correlation coefficient of each characteristic index; Step S13: sort the absolute values ​​of the comprehensive correlation coefficients of all characteristic indicators in descending order, and take the characteristic indicators whose absolute values ​​of the comprehensive correlation coefficients are higher than the threshold as key indicators.

3. According to claim 1, a hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access is characterized in that: The calculation process of the new energy short-circuit ratio also includes: Calculate the expected and standard deviation of active power at each site based on wind and PV output power models; The ratio of the short-circuit capacity of each station to the comprehensive capacity is taken as the new energy short-circuit ratio.

4. According to claim 1, a method for hierarchical dynamic evaluation of the reactive comprehensive support capacity of a distribution network taking into account the access of new energy sources, characterized in that: The calculation process of the effective reactive power reserve is: A. For each reactive source node in the distribution network, according to the probability density function of wind power output, photovoltaic power output and load, randomly extract wind speed, solar irradiance and load values ​​to generate specific scenarios and form a scenario set; B. Use the Kmeans algorithm to cluster the scene set, extract the scene parameters of each cluster center, and use the improved continuous power flow method to calculate the PV curve under each cluster center scene; C. Obtain the nose point reactive power and the normal operating point reactive power according to the nose point voltage and the normal operating point voltage on each PV curve, and calculate the effective reactive power reserve of each reactive source node; D. Sum the effective reactive power reserves of all reactive source nodes to obtain the effective reactive power reserve of the distribution network in each clustering scenario; E. Average the effective reactive power reserve of the distribution network under all clustering scenarios to obtain the expected effective reactive power reserve of the distribution network.

5. According to claim 1, a hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access is characterized in that: The calculation process of the reactive power output cost is: Among them, C represents the reactive power output cost, C Q,d represents the reactive power shortage compensation cost of node i, C Q,loss represents the contribution cost of the reactive power source output of node i to the distribution network loss, C Q,PV represents the photovoltaic reactive power production cost, represents the reactive power production cost of wind power, C gci (Q ci ) represents the reactive power production cost of the reactive power compensation equipment at node i, C gqi (Q i ) represents the reactive power production cost of the generator connected to node i.

6. A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access according to claim 1, characterized in that: The specific process of constructing the reactive comprehensive support capability evaluation model of the distribution network in step S2 is as follows: V s =ω1(REGSCR y ) * +ω2(E[Q res ]) * +ω3C * Among them, V s Represents the score of the distribution network reactive power comprehensive support capability evaluation model, (REGSCR y ) * Represents the quantitative value of the short-circuit ratio of new energy, ω1 represents (REGSCR y ) * The weight of (E[Q res ]) * Represents the quantitative value of effective reactive power reserve, ω2 represents (E[Q res ]) * The weight of C * represents the quantitative value of reactive power output cost, ω3 represents C * The weight of .

7. A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access according to claim 1, characterized in that: Entropy value of the key indicator The calculation process is: Where m represents the number of data sections, j * Indicates the key indicator index, Represents the key indicator j * The characteristic proportions in all data sections; The calculation process of the comprehensive weight of the key indicators in m data sections is:

8. A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access according to claim 1, characterized in that: The process of assigning points to key indicators in step S2 is as follows: The process of assigning points to the quantitative value of the short-circuit ratio of new energy and the quantitative value of the effective reactive power reserve is: Among them, S a The score representing the quantitative value of the short-circuit ratio of renewable energy or the quantitative value of the effective reactive power reserve, x a Represents the quantitative value of the short-circuit ratio of new energy or the quantitative value of effective reactive power reserve, x max Indicates the upper limit of the score for the corresponding key indicator, x min Indicates the lower limit of the score for the corresponding key indicator; The process of assigning points to the quantitative value of reactive power output cost is as follows: Among them, S cost The score representing the quantitative value of reactive power output cost, x cost It is the quantitative value of reactive power output cost indicator.

9. A hierarchical dynamic evaluation method for reactive comprehensive support capability of distribution network taking into account new energy access according to claim 1, characterized in that: In step S3, when clustering is performed using a semi-supervised clustering method, the must-connect constraint and the must-not-connect constraint are determined; The necessary connection constraints are: Data sections where the scores of the three key indicators of the reactive power comprehensive support capability of the distribution network are all higher than 60 points; Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only two of them had scores higher than 60 points; Among the three key indicators of the reactive power comprehensive support capability of the distribution network, only one data section had a score higher than 60 points; Data sections where the scores of the three key indicators of the reactive power comprehensive support capability of the distribution network are all below 60 points; The non-connection constraint is: the data section corresponding to the above must-connection constraint and the data section corresponding to other must-connection constraints.

Images