Power distribution network distributed photovoltaic bearing capacity assessment method based on Copula function and robust weighted Kmeans method

Abstract

The invention provides a photovoltaic bearing capacity assessment method based on a Copula function and a robust weighted Kmeans method, and on the basis of considering photovoltaic uncertainty and source load correlation, photovoltaic bearing capacity assessment and improvement of the photovoltaic bearing capacity are studied. Firstly, a Copula function is combined with an improved Latin hypercube sampling source load correlation method, a source load correlation matrix is generated, and a robust weighted Kmeans method is adopted to cluster the source load correlation matrix to generate a typical scene. Secondly, considering the influence of economic and technical indexes on a photovoltaic bearing capacity result, establishing a photovoltaic bearing capacity evaluation model taking the minimum total investment and operation cost as a target, and researching the influence of energy storage configuration on a photovoltaic bearing capacity improvement effect through an energy storage configuration mode; the improvement effect of the photovoltaic bearing capacity under different energy storage configuration schemes is analyzed, a YALMIP matrix and a CPLEX solver are adopted to carry out optimization solving on the model, and the effectiveness of the method provided by the invention is verified through IEEE33 node example simulation.

Description

Technical Field

[0001] The present invention belongs to the field of distribution network optimization configuration and planning, and in particular relates to a distribution network distributed photovoltaic carrying capacity evaluation method based on a Copula function and a robust weighted Kmeans method. Background Art

[0002] The access of a large number of distributed photovoltaics has become a new structural form of the distribution network. However, a high proportion of photovoltaic access to the distribution network will change the distribution of the distribution network flow, resulting in problems such as voltage exceeding the limit due to changes in grid parameters, increased network losses, and abandoned light. In addition, the correlation between photovoltaic power sources and loads is often ignored in research, and the uncertainty of photovoltaics affects the planning and operation of the distribution network. Therefore, it is urgent to consider the source-load correlation and photovoltaic uncertainty when evaluating the photovoltaic carrying capacity. In addition, from the perspective of photovoltaic absorption rate and economic planning, it is also of practical significance to study methods to improve the carrying capacity of distributed photovoltaics.

[0003] Considering that configuring energy storage is an effective way to alleviate voltage over-limit, reduce abandoned light, and improve photovoltaic carrying capacity, compared with traditional methods of improving photovoltaic carrying capacity, configuring energy storage can not only greatly improve photovoltaic carrying capacity, but also the calculation process is relatively simple. Moreover, reasonable configuration of energy storage can reduce the annual total cost and improve the economy of system operation. Summary of the invention

[0004] The purpose of the present invention is to propose a method for evaluating the carrying capacity of distributed photovoltaic power generation in a distribution network. The overall objective function takes into account the one-time investment cost of photovoltaic power generation, the one-time investment cost of energy storage, the operation and maintenance cost of photovoltaic power generation, the operation and maintenance cost of energy storage, the cost of abandoned light, the network loss cost, and the energy storage income. A second-order cone relaxation optimal power flow distributed photovoltaic carrying capacity evaluation planning model is constructed. The annual total photovoltaic cost and safety constraint indicators are calculated in a scenario considering correlation and uncertainty. The annual total cost is used as the objective function to optimize the relevant quantities such as the maximum capacity that can be connected to photovoltaic power generation.

[0005] In a first aspect, the present invention provides a method for evaluating the distributed photovoltaic carrying capacity of a distribution network using a Copula function and a robust weighted Kmeans method, which comprises:

[0006] Obtain historical data of photovoltaic power and load power to determine the probability density functions of photovoltaic random variables and load random variables, and integrate and calculate the marginal distribution functions of photovoltaic random variables and load random variables respectively;

[0007] Based on the marginal distribution functions of the photovoltaic random variable and the load random variable, determining the joint distribution function of the photovoltaic random variable and the load random variable according to the selected Copula function;

[0008] Based on the joint distribution function, random sampling is performed to generate random photovoltaic power values ​​and random load power values, and further generate a photovoltaic matrix and a load matrix with source-load correlation;

[0009] Based on the photovoltaic matrix and the load matrix, a robust weighted Kmeans method is used for clustering to generate a plurality of photovoltaic load scenarios that take into account correlation; the photovoltaic load scenarios include photovoltaic power curves and load power curves that change over time;

[0010] Based on the multiple photovoltaic load scenarios, according to a pre-built evaluation model, an optimization solution is performed with annual total cost as the objective function and under safety constraints.

[0011] Preferably, the determining of the probability density functions of the photovoltaic random variable and the load random variable is performed based on a kernel density estimation method.

[0012] Preferably, the Copula function is selected and determined using the AIC (BIC) information criterion method, and the unknown parameters of the Copula function are solved using the maximum likelihood estimation method.

[0013] Preferably, the random sampling is performed using an improved Latin hypercube method; the step of generating a photovoltaic matrix and a load matrix with source-load correlation comprises:

[0014] The number of samples of photovoltaic random variables and load random variables is set, and a multi-dimensional parameter space is established for the photovoltaic random variables and load random variables;

[0015] Divide the parameter space of each dimension into N small intervals evenly;

[0016] A point is randomly selected in each small interval, and then the points selected in each dimension are combined into a vector containing photovoltaic power and load power to form the final sample set;

[0017] A photovoltaic matrix and a load matrix with source-load correlation are generated according to the sample set.

[0018] As a preferred method, in the sampling process, according to the regular characteristics of the peak at noon and the peak at night of the load curve, a time constraint condition is added; the time constraint condition is: when 5<t<10, the loads meeting P are screened out. t >P t-1 Load sample data; when 12<t<15, filter out the load sample data that meets P t <P t-1 Load sample data; when 15<t<20, filter out the load sample data that meets P t >P t-1 The load sample data is used to consider the load relationship between time t and t-1, where t represents the moment, P trepresents the load power at time t, P t-1 Represents the load power at time t-1.

[0019] Preferably, the step of generating a plurality of photovoltaic load scenarios taking into account correlation comprises:

[0020] Select the initial cluster center and set the hyperparameters;

[0021] According to the Euclidean distance between each data point and the cluster center, the weight coefficient is calculated according to the Euclidean distance;

[0022] Assign each data point to the nearest cluster according to the weight coefficient;

[0023] Recalculate the cluster centers. If the cluster centers converge, multiple photovoltaic load scenarios are obtained.

[0024] Preferably, there are four photovoltaic load scenarios.

[0025] Preferably, the optimization solution with annual total cost as objective function and under safety constraints is performed by matlab software using YALMIP matrix and CPLEX solver to calculate the maximum photovoltaic accessible capacity and the photovoltaic accessible capacity of each node, the actual photovoltaic output under different photovoltaic load scenarios, and the power purchased from the superior power grid.

[0026] Preferably, the evaluation model comprises:

[0027] Without considering the effect of energy storage configuration on the photovoltaic load capacity improvement scenario, the system investment and operation comprehensive cost f cost Including network loss cost f loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l ;

[0028] f cost =f loss +f pvt +f pvw +f l (1)

[0029] Considering the effect of energy storage configuration on improving photovoltaic carrying capacity, the comprehensive cost of system operation is cost Including network loss loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l , energy storage investment cost f1, energy storage system maintenance cost f2, and energy storage benefit f3;

[0030] f cost =f loss +f pvt +f pvw +f l +f1+f2-f3 (2)

[0031] The network loss cost is calculated as follows:

[0032]

[0033] Where T is the total optimization time period (24 hours); ξ is the magnitude correction coefficient of the system operating cost, and ω is the weight value coefficient of the system operating cost, both of which are real numbers greater than or equal to 0; N B The set of all nodes in the distribution network; The price of unit active energy purchased by the distribution network from the upper power grid at time t; is the active network loss inside the network at time t;

[0034] The annual value coefficient of photovoltaic power generation is calculated as follows:

[0035]

[0036] In the formula, A Inv is the annual value coefficient of photovoltaic power generation, γ is the discount rate; Y pv It is the service life of the photovoltaic unit.

[0037]

[0038] Where N pv The number of photovoltaic connections; is the capacity of the i-th photovoltaic; pv Investment cost per unit of photovoltaic capacity.

[0039] Photovoltaic annual maintenance cost:

[0040]

[0041] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m pv is the photovoltaic maintenance cost per unit of electricity; N PV is the number of photovoltaic connections; T is a calculation cycle, which is 24; is the power of the mth photovoltaic at the nth moment;

[0042] The cost of abandoned light is calculated as follows:

[0043]

[0044] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; N PV P is the number of photovoltaic connections; j is the amount of abandoned light from the j-th photovoltaic power plant in 24 hours, and k is the abandoned light cost per unit power.

[0045] The investment cost of the energy storage system is calculated as follows:

[0046]

[0047] Where γ is the discount rate; Y Ess The service life of the energy storage unit; P i Ess Respectively represent the capacity and power of the i-th energy storage installation, A Inc Indicates the annual value coefficient of energy storage; Ist Essc Represents the energy storage investment cost per unit capacity; Ist Essp Represents the energy storage investment cost per unit power; N Ess Indicates the number of energy storage devices connected.

[0048] The maintenance cost of the energy storage system is calculated as follows:

[0049]

[0050] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m Ess N is the energy storage maintenance cost per unit of electricity; Ess is the number of energy storage connections; T is a calculation cycle, which is 24; is the charging power of the mth energy storage at the nth moment; is the discharge power of the mth energy storage at the nth moment;

[0051] The benefit value of the energy storage system is calculated as follows:

[0052]

[0053] In the formula, Δy t is the price difference between charging and discharging electricity at time t; is the charging and discharging power of the nth energy storage; is the capacity utilization rate of the nth energy storage system during the charging and discharging process; S C Subsidy for electricity;

[0054] The objective function of photovoltaic carrying capacity assessment is:

[0055] F=minfcost (12)

[0056] In a second aspect, the present invention provides a distributed photovoltaic carrying capacity assessment device for a distribution network, comprising:

[0057] A calculation module is used to obtain historical data of photovoltaic power and load power to determine the probability density functions of photovoltaic random variables and load random variables, and to integrate and calculate the marginal distribution functions of photovoltaic random variables and load random variables respectively;

[0058] A determination module, configured to determine a joint distribution function of the photovoltaic random variable and the load random variable based on the marginal distribution functions of the photovoltaic random variable and the load random variable and according to a selected Copula function;

[0059] A first generating module is used for performing random sampling based on the joint distribution function to generate a photovoltaic power random value and a load power random value, and further generating a photovoltaic matrix and a load matrix with source-load correlation;

[0060] A second generation module is used to generate a plurality of photovoltaic load scenarios taking into account correlation by clustering based on the photovoltaic matrix and the load matrix using a robust weighted Kmeans method; the photovoltaic load scenario includes a photovoltaic power curve and a load power curve that change with time;

[0061] A solution module is used to optimize and solve the multiple photovoltaic load scenarios based on a pre-built evaluation model, taking the annual total cost as the objective function and under safety constraints.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The present invention studies the evaluation of distributed photovoltaic carrying capacity of distribution network and the improvement of photovoltaic carrying capacity from the perspective of source-load correlation and photovoltaic uncertainty. In terms of photovoltaic carrying capacity evaluation, the kernel density estimation method is used to construct a photovoltaic output probability distribution model, and the Copula function method is used to describe the source-load correlation to generate photovoltaic power source and load correlation data, and the improved Latin hypercube method is used to sample the generated data, and then the robust weighted kmeans clustering method is used to cluster the generated data to generate typical scenarios. Based on the generation of the previous correlation scenario, a second-order cone relaxation optimal power flow distributed photovoltaic carrying capacity evaluation planning model is constructed. The annual photovoltaic total cost and safety constraint indicators are calculated in the scenario considering correlation and uncertainty, and the maximum photovoltaic capacity that can be connected is optimized and solved with the annual total cost as the objective function. With the sudden increase in the number and capacity of distributed photovoltaic access, problems such as voltage exceeding the limit, excessive operating costs, and low photovoltaic utilization rate in the distribution network will become increasingly prominent. The present invention adopts the photovoltaic carrying capacity evaluation based on the Copula function of the second-order cone relaxation optimal power flow and the robust weighted Kmeans method, which can better evaluate the photovoltaic carrying capacity, and can improve the photovoltaic carrying capacity by reasonably configuring the energy storage, reduce the operation and maintenance costs, and improve the photovoltaic utilization rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of the analysis of the distributed photovoltaic carrying capacity of the distribution network in an embodiment of the present invention;

[0065] Figure 2 It is a topological diagram of the IEEE-33 node distribution network structure in an embodiment of the present invention;

[0066] Figure 3 A photovoltaic output curve diagram for the whole year drawn according to historical photovoltaic data in an embodiment of the present invention;

[0067] Figure 4 The photovoltaic output frequency histogram and kernel density estimation probability distribution diagram at the 13th moment in an embodiment of the present invention;

[0068] Figure 5 A comparison diagram of load characteristic curves obtained when taking into account time regularity and when not taking into account time regularity during sampling in an embodiment of the present invention;

[0069] Figure 6 It is a time-of-use electricity price map;

[0070] Figure 7 It is a graph showing the actual photovoltaic output of all nodes in different scenarios in an embodiment of the present invention;

[0071] Figure 8 It is a load characteristic curve diagram of all nodes in different scenarios in the embodiment of the present invention.

[0072] Fig. 9 A photovoltaic capacity diagram that can be accessed by different nodes in an embodiment of the present invention;

[0073] Fig.10 A power diagram of electricity purchased from a superior power grid in different scenarios in an embodiment of the present invention;

[0074] Fig.11 A graph showing the optimization result of each node accessing the distributed photovoltaic capacity in different scenarios in an embodiment of the present invention;

[0075] Fig.12 The optimization result of accessing distributed energy storage capacity of different nodes in the embodiment of the present invention;

[0076] Fig.13 It is a graph of annual total cost under different scenarios in the embodiment of the present invention;

[0077] Fig.14 It is a graph of annual total cost under different energy storage capacities in an embodiment of the present invention;

[0078] Fig.15 Schematic diagram of power purchase from the superior power grid during 19:00-21:00 in different scenarios in the embodiments of the present invention. DETAILED DESCRIPTION

[0079] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more. In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0080] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0081] Example 1

[0082] Combination Figure 1 The present invention provides a photovoltaic carrying capacity assessment strategy based on the Copula function of the second-order cone relaxation optimal power flow and the robust weighted Kmeans method, including constructing a second-order cone relaxation optimal power flow distributed photovoltaic carrying capacity assessment planning model, calculating the annual photovoltaic total cost and safety constraint indicators under the scenario of considering correlation and uncertainty, optimizing the maximum photovoltaic access capacity with the annual total cost as the objective function, and comparing and analyzing the photovoltaic carrying capacity assessment results under different scenarios. The specific steps are:

[0083] Step 1: Construction of the second-order cone relaxation optimal power flow photovoltaic carrying capacity evaluation model:

[0084] (1) Objective function:

[0085] In a specific embodiment, without considering the effect of energy storage configuration on the photovoltaic carrying capacity, the system investment and operation comprehensive cost f cost Including network loss cost f loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l ;

[0086] f cost =f loss +f pvt +f pvw +f l (1)

[0087] Considering the effect of energy storage configuration on improving photovoltaic carrying capacity, the comprehensive cost of system operation is cost Including network loss loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l , energy storage investment cost f1, energy storage system maintenance cost f2, and energy storage benefit f3;

[0088] f cost =f loss +f pvt +f pvw +f l +f1+f2-f3 (2)

[0089] The network loss cost is calculated as follows:

[0090]

[0091] Where T is the total optimization time period (24 hours); ξ is the magnitude correction coefficient of the system operating cost, and ω is the weight value coefficient of the system operating cost, both of which are real numbers greater than or equal to 0; N B The set of all nodes in the distribution network; The price of unit active energy purchased by the distribution network from the upper power grid at time t; is the active network loss inside the network at time t;

[0092] The annual value coefficient of photovoltaic power generation is calculated as follows:

[0093]

[0094] In the formula, A Inv is the annual value coefficient of photovoltaic power generation, γ is the discount rate; Y pv It is the service life of the photovoltaic unit.

[0095]

[0096] Where N pv The number of photovoltaic connections; is the capacity of the i-th photovoltaic; pv Investment cost per unit of photovoltaic capacity.

[0097] Photovoltaic annual maintenance cost:

[0098]

[0099] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m pv is the photovoltaic maintenance cost per unit of electricity; N PV is the number of photovoltaic connections; T is a calculation cycle, which is 24; is the power of the mth photovoltaic at the nth moment;

[0100] The cost of abandoned light is calculated as follows:

[0101]

[0102] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; N PV P is the number of photovoltaic connections; j is the amount of abandoned light from the j-th photovoltaic power plant in 24 hours, and k is the abandoned light cost per unit power.

[0103] The investment cost of the energy storage system is calculated as follows:

[0104]

[0105]

[0106] Where γ is the discount rate; Y Ess The service life of the energy storage unit; P i Ess Respectively represent the capacity and power of the i-th energy storage installation, A Inc Indicates the annual value coefficient of energy storage; Ist Essc Represents the energy storage investment cost per unit capacity; Ist Essp Represents the energy storage investment cost per unit power; N Ess Indicates the number of energy storage devices connected.

[0107] The maintenance cost of the energy storage system is calculated as follows:

[0108]

[0109] Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m Ess N is the energy storage maintenance cost per unit of electricity; Ess is the number of energy storage connections; T is a calculation cycle, which is 24; is the charging power of the mth energy storage at the nth moment; is the discharge power of the mth energy storage at the nth moment;

[0110] The benefit value of the energy storage system is calculated as follows:

[0111]

[0112] In the formula, Δy t is the price difference between charging and discharging electricity at time t; is the charging and discharging power of the nth energy storage; is the capacity utilization rate of the nth energy storage system during the charging and discharging process; S C Subsidy for electricity;

[0113] The objective function of photovoltaic carrying capacity assessment is:

[0114] F=minf cost (12)

[0115] (2) Constraints

[0116] In a specific embodiment, in order to achieve the purpose of optimizing the operation of the distribution network, we need to use safety constraints to regulate the controllable units in the distribution network. The optimal power flow constraints will be established from aspects such as the power flow of the distribution network.

[0117] 1) Branch flow constraints

[0118] Select the operating status of one branch at time t to establish the branch power flow model, and the constraints that should be met are:

[0119]

[0120] Where i and j are node numbers; p i,t 、p j,t are the active injected power of nodes i and j respectively; q i,t ,q j,t are the reactive injection power of nodes i and j respectively; P ij,t , Q ij,t are respectively the active and reactive power at the head end of branch ij; r ij +jx ij is the impedance of branch ij; P jk,t , Q jk,t are respectively the first section active and reactive power of branch jk; k:j→k is the set of child nodes with node j as the parent node; α i,t and β ij,t are the square of the voltage at node i and the square of the current at branch ij during period t,

[0121] And It is a nonlinear equality constraint. At this time, SOCR is used to process the formula as follows:

[0122]

[0123] After equivalent transformation, formula (14) can be written in standard second-order cone form, that is,

[0124]

[0125] 2) Power balance constraints

[0126]

[0127] Where N pv is the number of photovoltaic access nodes; N Ess N is the number of energy storage connections; bus is the number of load nodes; P i Output active power to the grid; is the photovoltaic active power of the i-th node at the t-th moment; is the active power stored in node j at time t; Q is the active power of the load of the kth node at the tth moment. i Export reactive power to the grid; is the photovoltaic reactive power of the i-th node at the t-th moment; is the reactive power stored in node j at time t; is the reactive power of the load of the kth node at the tth moment.

[0128] 3) Node voltage constraints

[0129]

[0130] In the formula, u min 、u max are the lower and upper limits allowed for the voltage amplitude of each node respectively.

[0131] 4) Branch current constraints

[0132] The constraint formula of branch current is as follows:

[0133] |I ij |≤I ij,max (18)

[0134] In the formula, I ij,max is the current limit of line ij.

[0135] 5) Distributed photovoltaic output constraints

[0136]

[0137] Where P pv.i and Q pv.i are respectively the current output active and reactive power of the distributed photovoltaic connected to node i; is the upper limit of the active power output of the photovoltaic on node i; is the upper limit of the PV output reactive power at node i; its value is determined by the following formula

[0138]

[0139] In the formula, S max is the photovoltaic inverter capacity; P pv Make contributions to the current situation.

[0140] 6) Energy storage element constraints

[0141] The constraint of energy storage SOC at time t can be expressed as:

[0142]

[0143] In the formula, is the lower limit of energy storage SOC; It is the upper limit of energy storage SOC.

[0144] 7) Energy storage capacity constraints

[0145]

[0146] In the formula, E t is the electric quantity at time t; E t-1 is the power at time t-1; Charging power for energy storage; is the energy storage discharge power; η is the efficiency.

[0147] 8) Energy storage power constraints

[0148]

[0149] In the formula, is the lower limit of the energy storage system; is the upper limit of the energy storage system, The charging power for energy storage, is the energy storage discharge power.

[0150] 9) Energy storage energy balance constraints

[0151] Cap T ≥Cap1 (24)

[0152] In the formula, T is a calculation cycle, which is 24; Cap T Cap1 is the energy storage capacity at the final moment; Cap2 is the energy storage capacity at the initial moment.

[0153] Step 2: Solving the second-order cone relaxation optimal power flow model for photovoltaic carrying capacity assessment:

[0154] (1) Description of correlation and generation of typical scenarios

[0155] 1) Input historical photovoltaic and load data, select the standard Gaussian function as the kernel function, and use the kernel density estimation method to calculate the photovoltaic and load probability density functions. The probability density function formula is:

[0156]

[0157]

[0158] Where N is the number of samples, h is the bandwidth, x is the random variable, and X m is the mth sample, f(x) represents the probability density function of the distributed photovoltaic random variable or load random variable, and P(*) is the Gaussian kernel function;

[0159] 2) AIC (BIC) information criterion method is used to calculate and select the appropriate Copula function. This method mainly calculates the AIC and BIC values ​​and compares and analyzes them to select the Copula function corresponding to the minimum AIC and BIC. This Copula function is the optimal Copula function. The specific calculation formula is:

[0160]

[0161] Where C is the probability density function (PDF) value of the Copula function of all sample points in the first step of sampling; k is the spatial dimension of the random variable; N is the number of samples…

[0162] 3) Use the maximum likelihood estimation method to solve the first unknown parameter θ1, the second unknown parameter θ2, and the third unknown parameter α in the Copula function. Set the marginal distribution functions F(x; θ1) and G(y; θ2) of the continuous photovoltaic random variable and the load random variable, and the marginal density functions are f(x; θ1) and g(y; θ2), respectively. Suppose the selected Copula distribution function is C(u, v; α), and the Copula density function is c(u, v; α). The joint distribution function of (X, Y) composed of photovoltaic power X and load power Y is H(x, y; θ1, θ2, α) = C(F(x; θ1), G(y; θ2); α) to calculate the joint density function of (X, Y) So we can get the log-likelihood function By solving the maximum point of the log-likelihood function, we can then solve the first unknown parameter θ1, the second unknown parameter θ2, and the third unknown parameter α in the Copula function; where X here refers to the photovoltaic power, Y here refers to the load power, and the joint distribution function and the joint density function refer to the joint function of the photovoltaic variable function and the load variable function.

[0163] 4) Fitting random sampling to generate random values ​​of photovoltaic and load;

[0164] 5) The load and photovoltaic are sampled by the improved Latin hypercube method. The specific steps are as follows: first, the number of samples of photovoltaic random variables and load random variables is set, and a multidimensional parameter space is established for photovoltaic power and load power; then, each dimension of the parameter space is evenly divided into N small intervals, a point is randomly selected in each small interval, and then the points extracted in each dimension are combined into a vector to form the final sample set. In the sampling process, according to the regular characteristics of the peak at noon and the peak at night of the load curve, time constraints are added to consider the load relationship between time t and t-1, and a reasonable photovoltaic matrix and load matrix with source-load correlation are generated; the time constraint is: when 5<t<10, the photovoltaic matrix and load matrix that meet P are screened out. t>P t-1 Load sample data; when 12<t<15, filter out the load sample data that meets P t <P t-1 Load sample data; when 15<t<20, filter out the load sample data that meets P t >P t-1 The load sample data is used to consider the load relationship between time t and t-1, where t represents the moment, P t represents the load power at time t, P t-1 Represents the load power at time t-1.

[0165] 6) Based on the photovoltaic matrix and load matrix of source-load correlation generated in step 5), clustering is performed using the robust weighted Kmeans method, the initial cluster center is selected, the hyperparameter a is set, the weight coefficient is calculated based on the Euclidean distance between each data point and the cluster center, each data point is divided into different cluster clusters C according to the weight coefficient, the cluster center μ is recalculated, and it is determined whether the cluster center μ converges. If the cluster center converges, multiple typical photovoltaic load scenarios that change with time t are generated, and the photovoltaic load scenarios include photovoltaic power curves and load power curves. In a specific embodiment, the photovoltaic load scenarios are 4, and the 4 typical photovoltaic and load scenarios are based on the four seasons of spring, summer, autumn and winter, so four typical scenarios are generated, among which the photovoltaic and load scenarios in spring and autumn are relatively similar, so two of the scenarios can represent both spring and autumn.

[0166] (2) Optimize configuration results

[0167] ① Distributed photovoltaic carrying capacity assessment steps

[0168] 1) Generation of four source-load scenarios considering correlation and uncertainty;

[0169] In a specific embodiment, the source-load scenario is divided into the following four scenarios, and the carrying capacity evaluation results in different scenarios are compared:

[0170] Scenario 1: Considering the uncertainty of photovoltaic power generation, ignoring the correlation between source and load, and not configuring energy storage, the evaluation of photovoltaic power generation capacity is carried out;

[0171] Scenario 2: Considering the uncertainty of photovoltaic power generation and ignoring the correlation between source and load, the evaluation of photovoltaic power carrying capacity in the energy storage scenario;

[0172] Scenario 3: Considering the uncertainty of photovoltaic power generation and the correlation between source and load, the evaluation of photovoltaic power generation capacity without configuring energy storage;

[0173] Scenario 4: Considering the uncertainty of photovoltaic power generation and the correlation between source and load, the photovoltaic carrying capacity is evaluated in the energy storage scenario.

[0174] 2) Calculate the total annual photovoltaic cost f under different scenarios;

[0175] 3) Considering the constraints, we can get the annual total cost f under different scenarios;

[0176] 4) Determine whether the annual total cost f is minimum. If so, obtain the photovoltaic access capacity, actual photovoltaic output in different seasonal scenarios, and power purchased from the superior power grid.

[0177] ② Steps to improve distributed photovoltaic carrying capacity

[0178] 1) Generation of four source-load scenarios considering correlation and uncertainty;

[0179] 2) Calculate the total annual photovoltaic cost f1 under different scenarios based on the constraints;

[0180] 3) Determine whether the constraints are met. If the constraints are met, obtain the annual total cost f1 considering different scenarios;

[0181] 4) Determine whether the annual total cost f1 is minimum. If the constraints are met, obtain the accessible photovoltaic capacity, accessible energy storage capacity, energy storage rated power, actual photovoltaic output in different seasonal scenarios, and power purchased from the superior power grid.

[0182] Step 3, case analysis:

[0183] In order to verify the feasibility and effectiveness of the method proposed in this paper, simulation analysis was carried out on the IEEE33 standard example system of the radial distribution network. MATLAB R2021b was used to write the program, and the YALMIP matrix and Cplex solver were selected for solution.

[0184] The IEEE33 node example system is selected. This example is a typical radial distribution network. The system voltage level (reference voltage) is 10kV, and the system reference capacity is SB = 10000kVA. The data sampling interval is 1h. Node 0 is a balancing node, which is connected to the upper-level distribution network for power transmission. The specific example system is as follows Figure 2 shown.

[0185] (1) Correlation description and typical scenario generation analysis

[0186] 1 Kernel density estimation analysis

[0187] Based on the historical photovoltaic data, the annual photovoltaic output curve is drawn as follows: Figure 3As shown in the figure. Different bandwidths and step sizes will affect the probability distribution results of kernel density estimation. A smaller bandwidth will cause more high-value or low-value areas to appear in the kernel density estimation results, which is suitable for revealing the local characteristics of density distribution, while a larger bandwidth and step size can observe the characteristics of data distribution at a global scale. Through the analysis and calculation of photovoltaic data, this paper selects the step size Step = 0.05 and the bandwidth h = 20 for the best observation of kernel density estimation results.

[0188] The number of samples is set to N = 365. The light intensity at the 13th moment of the day is relatively high, and the photovoltaic output is relatively strong. Therefore, the photovoltaic output data of the whole year at the 13th moment is selected as the sample. The probability density function is calculated based on the sample data and related parameters. The probability distribution of different light intensities at the 13th moment of the whole year for 365 days is obtained as follows: Figure 4 shown.

[0189] Result analysis: Figure 3 It can be seen that the maximum photovoltaic output in the annual photovoltaic data is 1148KW, and the minimum photovoltaic output data is 0KW. The photovoltaic output in summer is generally greater than that in winter. Figure 4 It can be seen that the photovoltaic output at the 13th moment is distributed between 100KW and 1100KW, among which the probability of photovoltaic output between 400KW and 600KW and 800KW to 1000KW is relatively high, and the probability of light intensity between 200KW and 400KW is relatively low. Since the light intensity is low in winter and rainy weather, and high in summer and sunny days, it can be seen from the probability distribution diagram that in the whole year, rainy weather accounts for a small proportion, while sunny days account for a high proportion.

[0190] ② Analysis of the results of Copula function describing source-charge correlation

[0191] The load data generated by the Copula function method with source-load correlation is compared with the load characteristic curves obtained by taking time regularity into account and not taking time regularity into account when sampling. Figure 5 shown.

[0192] Result analysis: Figure 5 It can be seen that when the time regularity is not considered, the load characteristic curve obtained has a large fluctuation and the regularity of the load characteristic curve is poor; when the load time regularity is considered, the load curve obtained has a small fluctuation, the curve is smoother and the load characteristic regularity of the curve is stronger.

[0193] In order to further analyze the changes in load volatility before and after improvement, this paper conducted a comparative analysis of the load variance before and after improvement (selecting the data from the 00:00-12:00 time period, which has a strong regularity and the analysis results are more accurate), maximum load, minimum load, load average and load change index. The analysis results are shown in Table 1.

[0194] Table 1 Analysis of load fluctuation index before and after improvement

[0195]

[0196] Results analysis: From the analysis in Table 1, it can be seen that the load variance during the period of 00:00-12:00 after the improvement is reduced compared with the load variance before the improvement; the maximum load after the improvement is reduced compared with the maximum load before the improvement, and the minimum load is increased compared with the load before the improvement; the average load after the improvement is reduced. After the improvement, the load change index is reduced by 0.095, and the load is more stable. This shows that the load peak-to-valley difference after the improvement is reduced, the load curve is smoother, the load fluctuation is smaller, and the load curve obtained after the improvement is more accurate.

[0197] (2) Analysis of IEEE33 node example results

[0198] Time-of-use electricity prices Figure 6 shown.

[0199] ① Carrying capacity assessment results considering the source-load correlation scenario

[0200] 1) Calculation conditions

[0201] The actual photovoltaic output curve under different scenarios is as follows: Figure 7 As shown in Figure 2, the load characteristic curves under different scenarios are as follows: Figure 8 shown.

[0202] 2) Example results

[0203] According to the selection of photovoltaic access points 3, 8, 12, 15, 18, 25, 28, 32, the photovoltaic access capacity and the required cost of these nodes are calculated. The total annual cost is 10188420 yuan, and the total photovoltaic access capacity is 5.15MWh. The calculation shows that the photovoltaic access capacity of the 8 nodes is as follows: Fig. 9 The power purchased from the upper grid in different scenarios is as follows Fig.10 shown.

[0204] Result analysis: Fig.10 It can be seen that the load in spring and autumn is smaller than that in winter and summer, so the power purchased in spring and autumn is generally smaller than that in winter and summer. From 10:00 to 15:00, it is the peak period of photovoltaic power generation, and photovoltaic power generation can supply the load. Therefore, the power purchased from the superior power grid during this period is relatively small. From 17:00 to 21:00, the load power is relatively large, and there is no photovoltaic output at this time. Therefore, during this period, the power purchased from the superior power grid is relatively large.

[0205] ② Comparison of carrying capacity results under different scenarios

[0206] 1) Comparison of carrying capacity assessment results

[0207] The following four scenarios are used to compare the load-bearing capacity assessment results under different scenarios:

[0208] Scenario 1: Considering the uncertainty of photovoltaic power generation, ignoring the correlation between source and load, and not configuring energy storage, the photovoltaic carrying capacity assessment is carried out;

[0209] Scenario 2: Considering the uncertainty of photovoltaic power generation and ignoring the correlation between source and load, the evaluation of photovoltaic power carrying capacity in the energy storage scenario;

[0210] Scenario 3: Considering the uncertainty of photovoltaic power generation and the correlation between source and load, the evaluation of photovoltaic power carrying capacity without configuring energy storage;

[0211] Scenario 4: Considering the uncertainty of photovoltaic power generation and the correlation between source and load, the photovoltaic carrying capacity is evaluated in the energy storage scenario.

[0212] Fig.11 The distributed photovoltaic capacity connected to different nodes under four scenarios was compared. Fig.12 The energy storage capacity and rated power of scenario 2 and scenario 4 are compared. Fig.13 The total annual costs under four scenarios are compared.

[0213] Result analysis: In scenario 1, the total capacity of distributed photovoltaics that can be connected is 5.23MW, and the total annual cost is 13,598,220 yuan.

[0214] In scenario 2, without considering the source-load related connected energy storage, the maximum capacity of distributed photovoltaics that can be connected is calculated to be 5.61MW, with an annual total cost of 13,583,650 yuan, and a total energy storage access capacity of 0.63MWh, which is 7.3% higher than the carrying capacity of scenario 1, and the total investment and operation cost is reduced by 14,570 yuan. After configuring energy storage, energy storage is charged during the period of high photovoltaic generation, and energy storage is discharged during the period of high load generation, realizing peak-valley arbitrage to obtain income. When the energy storage income and the reduction in the cost of purchasing electricity from the superior power grid are greater than the energy storage configuration and operation and maintenance costs, the total cost will be reduced.

[0215] In scenario 3, considering the source-load correlation and without energy storage, the annual total cost C1 is 10,188,420 yuan. The total capacity Q1 of distributed photovoltaic access is calculated to be 5.15MW, which is 25.1% lower than the total investment cost of scenario 1, and the total photovoltaic access capacity is only reduced by 1.5%. It can be seen that after considering the source-load correlation, while ensuring the safe operation of the system, the total photovoltaic access capacity is slightly reduced, and the total cost is reduced significantly, which improves the economy of system operation.

[0216] In scenario 4, taking into account the source-load related connected energy storage, the annual total cost C2 is 10,162,610 yuan, the total accessible capacity of distributed photovoltaic power generation Q2 is 5.93MW, and the total access capacity of energy storage is 0.60MWh. Compared with scenario 1, the total accessible capacity of photovoltaic power generation increases by 13.4%, and the total cost decreases by 25.3%; compared with scenario 2, the total accessible capacity of photovoltaic power generation increases by 5.7%, and the total cost decreases by 25.2%; compared with scenario 3, the total accessible capacity of photovoltaic power generation increases by 15.1%, and the total cost decreases by 0.25%.

[0217] It can be seen that when photovoltaic uncertainty and source-load correlation are considered and connected to energy storage, the photovoltaic carrying capacity can be effectively improved and the system economy can be improved.

[0218] 2) Analysis of the change in annual total cost with changes in photovoltaic capacity and energy storage capacity under the scenario of source-load correlation

[0219] Table 2 compares the change in total annual cost with and without energy storage as PV capacity increases.

[0220] Table 2 Changes in annual total cost under different photovoltaic capacities

[0221]

[0222] Results analysis: From the analysis in Table 2, it can be seen that when the photovoltaic access capacity increases, the annual total cost gradually increases. When the photovoltaic capacity increases from 6.5MWh to 7.0MWh, the annual total cost without energy storage increases by 1.20%, and the annual total cost after energy storage increases by 1.20%; when the photovoltaic capacity increases from 7.0MWh to 7.5MWh, the annual total cost without energy storage increases by 1.50%, and the annual total cost after energy storage increases by 1.45%; when the photovoltaic capacity increases from 8.0MWh to 8.5MWh, the annual total cost without energy storage increases by 1.60%, and the annual total cost after energy storage increases by 1.53%; it can be seen that with the increase of photovoltaic capacity, the annual total cost growth rate gradually increases, and the economic efficiency of system operation gradually decreases. In addition, when the same photovoltaic capacity is connected, the total cost of configuring energy storage is lower than that of not configuring energy storage. Therefore, configuring energy storage under the same conditions can improve the economic efficiency of system operation.

[0223] Fig.14 The change of total annual cost with the increase of energy storage capacity is analyzed.

[0224] Result analysis: According to Fig.14Analysis shows that as the energy storage configuration capacity increases, the annual total cost shows an increasing trend. When the energy storage capacity increases from 0.6MWh to 1.4MWh, the annual total cost increases by 6,100 yuan; when the energy storage capacity increases from 1.4MWh to 2.2MWh, the annual total cost increases by 20,870 yuan; when the energy storage capacity increases from 2.2MWh to 3.0MWh, the annual total cost increases by 37,360 yuan; when the energy storage capacity increases from 3.0MWh to 3.8MWh, the annual total cost increases by 50,360 yuan. Therefore, as the energy storage configuration capacity increases, the annual total cost growth also increases slightly, and the economy of system operation gradually decreases. Therefore, we should configure the energy storage capacity based on the actual situation and comprehensively consider the economy of system operation.

[0225] 3) Comparison of the total power purchased from the upper grid during 19:00-21:00 in different scenarios

[0226] The load peaks between 19:00 and 21:00, and the photovoltaic output is zero during this period. The sum of the power purchased from the upper grid between 19:00 and 21:00 in different scenarios is as follows: Fig.15 shown.

[0227] Result analysis: Fig.15 It can be seen that the sum of power purchased from the superior power grid during 19:00-21:00 from scenario 1 to scenario 4 is gradually reduced. When the source-load correlation is not considered and energy storage is not connected, the power purchased from the superior power grid is the largest, so the power purchase cost in this scenario is also the largest; when the source-load correlation is not considered and energy storage is connected, the power purchased from the superior power grid is slightly reduced compared with scenario 1; when the source-load correlation is considered and energy storage is connected, the sum of power purchased from the superior power grid during 19:00-21:00 is the smallest, and the land purchase cost is also the smallest. Therefore, when evaluating the photovoltaic carrying capacity, considering the source-load correlation and configuring energy storage can improve the photovoltaic carrying capacity and the economy of system operation. In each scenario, the power purchased from the superior power grid in winter is the largest, and the power purchased from the superior power grid in spring and autumn is relatively small. Because the weather is relatively cold in winter, electricity is needed for heating, so the residential electricity load is large in winter, and the power purchased from the superior power grid is large.

[0228] 4) Impact of different energy storage configuration schemes on PV carrying capacity and annual total cost under scenario 4

[0229] Table 3 analyzes the changes in the accessible photovoltaic capacity, accessible energy storage capacity and annual total cost under different numbers of energy storage connections.

[0230] Table 3 Changes of various indicators under different numbers of energy storage connections

[0231]

[0232]

[0233] From the analysis of Table 3, it can be seen that with the increase in the number of energy storage access nodes, the annual total cost shows a downward trend, and the photovoltaic access capacity increases. Compared with accessing 1 node, when the energy storage is connected to 12 and 15 nodes, the photovoltaic access capacity and the energy storage access capacity are slightly increased, and the annual total cost is also slightly reduced; compared with accessing 12 and 15 nodes, when the energy storage is connected to 12, 15, and 18 nodes, the energy storage access capacity remains unchanged, but the photovoltaic access capacity increases, and the annual total cost is also slightly reduced. Therefore, the photovoltaic carrying capacity in the distributed energy storage scenario is greater than that in the centralized energy storage scenario, and the economic performance is better. In addition, for distributed energy storage, as the number of energy storage access nodes increases, the photovoltaic carrying capacity gradually increases, and the economic efficiency of system operation improves.

[0234] This paper studies the evaluation of distributed photovoltaic carrying capacity of distribution network and the improvement of photovoltaic carrying capacity from the perspective of source-load correlation and photovoltaic uncertainty. In terms of photovoltaic carrying capacity evaluation, the kernel density estimation method is used to construct the photovoltaic output probability distribution model, and the Copula function method is used to describe the source-load correlation to generate photovoltaic power source and load correlation data, and the improved Latin hypercube method is used to sample the generated data, and then the robust weighted kmeans clustering method is used to cluster the generated data to generate typical scenarios. Based on the generation of the previous correlation scenario, a second-order cone relaxation optimal power flow distributed photovoltaic carrying capacity evaluation planning model is constructed. The annual photovoltaic total cost and safety constraint indicators are calculated in the scenario considering correlation and uncertainty, and the maximum photovoltaic capacity that can be connected is optimized with the annual total cost as the objective function.

[0235] In the future, with the sharp increase in the number and capacity of distributed photovoltaic access, problems such as voltage exceeding the limit, excessive operating costs, and low photovoltaic utilization rate in the distribution network will become increasingly prominent. The photovoltaic carrying capacity assessment based on the Copula function of the second-order cone relaxation optimal power flow and the robust weighted Kmeans method can better evaluate the photovoltaic carrying capacity, and through the reasonable configuration of energy storage, the photovoltaic carrying capacity can be improved, the operation and maintenance costs can be reduced, and the photovoltaic utilization rate can be improved.

[0236] The embodiments of the present invention are not exhaustive and do not constitute a limitation on the protection scope of the claims. Based on the inspiration gained from the embodiments of the present invention, those skilled in the art can think of other substantially equivalent alternatives without creative work, all of which are within the protection scope of the present invention.

Claims

1. A method for evaluating the carrying capacity of distributed photovoltaic power generation in a distribution network based on a Copula function and a robust weighted Kmeans method, characterized in that: include: Obtain historical data of photovoltaic power and load power to determine the probability density functions of photovoltaic random variables and load random variables, and integrate and calculate the marginal distribution functions of photovoltaic random variables and load random variables respectively; Based on the marginal distribution functions of the photovoltaic random variable and the load random variable, determining the joint distribution function of the photovoltaic random variable and the load random variable according to the selected Copula function; Based on the joint distribution function, random sampling is performed to generate random photovoltaic power values ​​and random load power values, and further generate a photovoltaic matrix and a load matrix with source-load correlation; Based on the photovoltaic matrix and the load matrix, a robust weighted Kmeans method is used for clustering to generate a plurality of photovoltaic load scenarios that take into account correlation; the photovoltaic load scenarios include photovoltaic power curves and load power curves that change over time; Based on the multiple photovoltaic load scenarios, according to a pre-built evaluation model, an optimization solution is performed with annual total cost as the objective function and under safety constraints.

2. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The determination of the probability density functions of the photovoltaic random variable and the load random variable is performed based on a kernel density estimation method.

3. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The Copula function is selected and determined using the AIC (BIC) information criterion method, and the unknown parameters of the Copula function are solved using the maximum likelihood estimation method.

4. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The random sampling is performed using an improved Latin hypercube method; the step of generating a photovoltaic matrix and a load matrix with source-load correlation comprises: The number of samples of photovoltaic random variables and load random variables is set, and a multi-dimensional parameter space is established for the photovoltaic random variables and load random variables; Divide the parameter space of each dimension into N small intervals evenly; A point is randomly selected in each small interval, and then the points selected in each dimension are combined into a vector containing photovoltaic power and load power to form the final sample set; A photovoltaic matrix and a load matrix with source-load correlation are generated according to the sample set.

5. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 4, characterized in that: In the sampling process, according to the regular characteristics of the peaks at noon and evening of the load curve, time constraints are added; the time constraints are: when 5<t<10, select the loads that meet P t >P t-1 Load sample data; when 12<t<15, filter out the load sample data that meets P t <P t-1 Load sample data; when 15<t<20, filter out the load sample data that meets P t >P t-1 The load sample data is used to consider the load relationship between time t and t-1, where t represents the moment, P t represents the load power at time t, P t-1 Represents the load power at time t-1.

6. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The step of generating a plurality of photovoltaic load scenarios taking into account correlation comprises: Select the initial cluster center and set the hyperparameters; According to the Euclidean distance between each data point and the cluster center, the weight coefficient is calculated according to the Euclidean distance; Assign each data point to the nearest cluster according to the weight coefficient; Recalculate the cluster centers. If the cluster centers converge, multiple photovoltaic load scenarios are obtained.

7. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 6, characterized in that: There are four photovoltaic load scenarios.

8. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The optimization solution with annual total cost as the objective function and under safety constraints is performed through matlab software using YALMIP matrix and CPLEX solver to calculate the maximum photovoltaic access capacity and the photovoltaic access capacity of each node, the actual photovoltaic output under different photovoltaic load scenarios, and the power purchased from the superior power grid.

9. The method for evaluating the distributed photovoltaic carrying capacity of a distribution network according to claim 1, characterized in that: The evaluation model includes: Without considering the effect of energy storage configuration on the photovoltaic load capacity improvement scenario, the system investment and operation comprehensive cost f cost Including network loss cost f loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l ; f cost =f loss +f pvt +f pvw +f l (1) Considering the effect of energy storage configuration on improving photovoltaic carrying capacity, the comprehensive cost of system operation is cost Including network loss loss , Photovoltaic investment cost pvt , Photovoltaic operation and maintenance costs pvw , abandoned light cost f l , energy storage investment cost f1, energy storage system maintenance cost f2, and energy storage benefit f3; <h2 style=";text-align:left;direction:ltr">f<h2 style=";text-align:left;direction:ltr"> cost <h2 style=";text-align:left;direction:ltr"> =f<h2 style=";text-align:left;direction:ltr"> loss <h2 style=";text-align:left;direction:ltr"> +f<h2 style=";text-align:left;direction:ltr"> pvt <h2 style=";text-align:left;direction:ltr"> +f<h2 style=";text-align:left;direction:ltr"> pvw <h2 style=";text-align:left;direction:ltr"> +f<h2 style=";text-align:left;direction:ltr"> l <h2 style=";text-align:left;direction:ltr"> +f1+f2-f3 (2) The network loss cost is calculated as follows: Where, T is the total optimization time period (24 hours); ξ is the magnitude correction coefficient of the system operating cost, ω is the weight value coefficient of the system operating cost; N B The set of all nodes in the distribution network; The price of unit active energy purchased by the distribution network from the upper power grid at time t; is the active network loss inside the network at time t; The annual value coefficient of photovoltaic power generation is calculated as follows: In the formula, A Inv is the annual value coefficient of photovoltaic power generation, γ is the discount rate; Y pv The service life of the photovoltaic unit; Where N pv The number of photovoltaic connections; is the capacity of the i-th photovoltaic; pv Investment cost per unit of PV capacity; Photovoltaic annual maintenance cost: Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m pv is the photovoltaic maintenance cost per unit of electricity; N PV is the number of photovoltaic connections; T is a calculation cycle, which is 24; is the power of the mth photovoltaic at the nth moment; The cost of abandoned light is calculated as follows: Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; N PV is the number of photovoltaic connections; P j is the amount of abandoned light from the jth photovoltaic power plant in 24 hours, and k is the abandoned light cost per unit power; The investment cost of the energy storage system is calculated as follows: Where γ is the discount rate; Y Ess The service life of the energy storage unit; Respectively represent the capacity and power of the i-th energy storage installation, A Inc Indicates the annual value coefficient of energy storage; Ist Essc Represents the energy storage investment cost per unit capacity; Ist Essp Represents the energy storage investment cost per unit power; N Ess Indicates the number of energy storage connections; The maintenance cost of the energy storage system is calculated as follows: Where N S is the number of photovoltaic load scenarios; W S (i) is the probability under the i-th photovoltaic load scenario; m Ess N is the energy storage maintenance cost per unit of electricity; Ess is the number of energy storage connections; T is a calculation cycle, which is 24; is the charging power of the mth energy storage at the nth moment; is the discharge power of the mth energy storage at the nth moment; The benefit value of the energy storage system is calculated as follows: In the formula, Δy t is the price difference between charging and discharging electricity at time t; is the charging and discharging power of the nth energy storage; is the capacity utilization rate of the nth energy storage system during the charging and discharging process; S C Subsidy for electricity; The objective function of photovoltaic carrying capacity assessment is: F=minf cost (12)。 10. A distributed photovoltaic carrying capacity assessment device for a distribution network, characterized in that: include: A calculation module is used to obtain historical data of photovoltaic power and load power to determine the probability density functions of photovoltaic random variables and load random variables, and to integrate and calculate the marginal distribution functions of photovoltaic random variables and load random variables respectively; A determination module, configured to determine a joint distribution function of the photovoltaic random variable and the load random variable based on the marginal distribution functions of the photovoltaic random variable and the load random variable and according to a selected Copula function; A first generating module is used for performing random sampling based on the joint distribution function to generate a photovoltaic power random value and a load power random value, and further generating a photovoltaic matrix and a load matrix with source-load correlation; A second generation module is used to generate a plurality of photovoltaic load scenarios taking into account correlation by clustering based on the photovoltaic matrix and the load matrix using a robust weighted Kmeans method; the photovoltaic load scenario includes a photovoltaic power curve and a load power curve that change with time; A solution module is used to optimize and solve the multiple photovoltaic load scenarios based on a pre-built evaluation model, taking the annual total cost as the objective function and under safety constraints.

Images