A method for improving distribution network reliability based on hybrid energy storage and dynamic network reconstruction

Abstract

The present invention discloses a distribution network reliability improvement method based on hybrid energy storage and dynamic network reconstruction, comprising the following steps: 1) generating a wind farm wind power scenario; 2) based on a variational mode decomposition algorithm, decomposing the generated wind farm power into high-frequency, medium-frequency and low-frequency components, and then distributing the high-frequency and medium-frequency components to supercapacitors and batteries respectively, using the low-frequency components as direct grid-connected power, and calculating the energy storage capacity and power of supercapacitors and batteries; 3) calculating grid reliability evaluation indicators; 4) establishing a two-stage optimization configuration model; 5) solving the two-stage optimization configuration model to optimize the grid reliability evaluation indicators. The present invention can effectively improve the operational reliability of the distribution network under wind power penetration.

Description

Technical Field

[0001] The present invention relates to the field of power system planning, and in particular to a method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction. Background Art

[0002] Renewable energy, especially wind energy, is a typical intermittent energy source with the characteristics of uneven temporal and spatial distribution and violent energy fluctuations. The access of a large number of distributed wind power to the distribution network brings further challenges to the planning and operation of the distribution network. For example, the fluctuation of wind power leads to real-time power imbalance in the distribution network, causing system frequency and voltage fluctuations, voltage waveform distortion, etc. The access of distributed wind power transforms the traditional distribution network from a single power supply network to a multi-power supply network, changes the size and direction of the power flow in the power grid, and changes the power grid voltage, increasing the risk of power grid voltage collapse. The energy storage system has the advantages of fast response, two-way regulation, and high flexibility, making the energy storage system an important control method in the future distribution network. The use of energy storage can solve a series of problems caused by wind power grid connection. However, due to the constraints of current energy storage technology, it is difficult for a single energy storage to achieve a balance between performance and cost. Therefore, the concept of hybrid energy storage system is proposed. The hybrid energy storage system combines the advantages of different energy storage, realizes technical and economic complementarity, and becomes the first choice for improving the reliability of distribution network operation.

[0003] The dynamic behavior of the power grid is usually affected by the system topology and operating status. Therefore, there are two main methods for identifying weak points in the reliability of the power grid. The first one is mainly based on complex network theory, which uses the topological structure of the network to construct indicators such as degree or betweenness. The second type is based on the operating status of the power system. However, existing research has failed to effectively combine the above two methods. At the same time, research on the identification of weak links in reliability is concentrated on the transmission network, and there is relatively little research on the distribution network, and most of its research ideas are derived from the transmission network. Summary of the invention

[0004] The object of the present invention is to provide a method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, comprising the following steps:

[0005] 1) Generate wind farm wind power scenario;

[0006] 2) Based on the variational mode decomposition algorithm, the generated wind farm power is decomposed into high-frequency, medium-frequency and low-frequency components, and then the high-frequency and medium-frequency components are allocated to supercapacitors and batteries respectively, and the low-frequency components are used as direct grid-connected power, and the energy storage capacity and power of supercapacitors and batteries are calculated;

[0007] 3) Calculate the grid reliability assessment index;

[0008] 4) Establish a two-stage optimization configuration model;

[0009] 5) Solve the two-stage optimization configuration model to optimize the grid reliability assessment indicators.

[0010] Furthermore, the steps of generating a wind farm wind power scenario include:

[0011] a1) Use the Gaussian mixture model to describe the probability distribution of wind speed in the wind farm f(x), that is:

[0012]

[0013]

[0014]

[0015] Among them, K represents the total number of Gaussian components, p k (x|θ k ) is given by the parameter θ k The k-th Gaussian component expression is determined by k ∈[0,1] is the weight corresponding to the kth Gaussian component, μ is the mathematical expectation of the random variable x, σ 2 is the variance of the random variable x, θ k =(μ,σ 2 ) represents the set of expected and variance distribution functions; x is the observed sample;

[0016] a2) Based on the probability distribution of wind speed f(x), construct the correlation coefficient matrix C of the relevant random vector Y in the wind farm y ,Right now:

[0017]

[0018] Among them, ρ ij are any two variables y in the random vector Y i and j The correlation coefficient between .

[0019] a3) By using formula (5), the correlated random vector Y is converted into a correlated standard normal random vector H = (h1, h2, ..., h c );

[0020]

[0021] Among them, Φ i (h i ) is the variable h i The standard normal distribution function of is the variable y i The cumulative distribution function F i The inverse function of (.);

[0022] Among them, the correlation coefficient matrix C of the related standard normal random vector H h As shown below:

[0023]

[0024] Among them, ρ i ' j are any two variables h in the vector H i and h j The correlation coefficient between .

[0025] a4) Based on Cholesky decomposition, the correlation coefficient matrix C h Decompose into upper triangular matrix B T And the lower triangular matrix B, we get:

[0026] C h =BB T (7)

[0027] a5) Convert the correlated standard normal random vector H into an independent standard normal random vector L, that is:

[0028] L=B -1 H (8)

[0029] a6) Using the Nataf inverse transform, the standard normal random vector L is converted into a sample of related wind speed random variables to generate a wind farm wind power scenario;

[0030] The expression of wind power in a wind farm is as follows:

[0031]

[0032] Among them, v is the real-time wind speed of the wind farm, v in and v out are the cut-in and cut-out wind speeds of the wind turbines in the wind farm, v N is the rated wind speed of the wind turbine in the wind farm, P N is the rated power of the wind turbine in the wind farm; P W is the output power of the wind farm.

[0033] Furthermore, in the Gaussian mixture model, the parameter ω k and parameter θ k Solved by expectation maximization algorithm;

[0034] Solving for parameter ω k and parameter θ k The steps include:

[0035] b1) Initialization parameter ω k and θ k, calculate the jth observation sample x j The corresponding k hidden variables γ jk ,Right now:

[0036]

[0037] b2) Based on latent variable γ jk , update the parameter ω k and θ k ,get:

[0038]

[0039]

[0040]

[0041] Where N is the number of observed samples; μ k , is the mathematical expectation and variance;

[0042] b3) Repeat steps 1) to 2) until the parameter ω k and parameter θ k convergence.

[0043] Furthermore, the correlation coefficient ρ of the related random variables Y ij The correlation coefficient ρ with the related standard normal random variable H ij The following relationship is satisfied:

[0044]

[0045] in, is the random variable y i and j The joint probability density function, φ2(h i ,h j ,ρ ij ) is the random variable h i and h j The joint probability density function of i , σ j is the random variable y i and j The corresponding standard deviation. i , μ j is the random variable y i and j The corresponding mathematical expectation. ij is the relationship coefficient.

[0046] Further, the step of decomposing the generated wind farm power into high-frequency, medium-frequency and low-frequency components based on the variational mode decomposition algorithm includes:

[0047] c1) Decompose the wind farm wind power into K modal functions with limited bandwidth and obtain:

[0048]

[0049] Among them, u k (t) is the wind farm output power P W (t) the kth modal component after decomposition, represents the gradient operation, ‖·‖2 represents L 2 -norm; ω k is the weight corresponding to the kth Gaussian component; t is the time; P W (t) is the wind farm output power at time t; δ(t) is the pulse function;

[0050] c2) For K mode functions with limited bandwidth, the components with frequencies less than ω2 are regarded as low-frequency components, and the frequency range is [ω2, ω th ) as the intermediate frequency component, and the frequency range is [ω th ,ω K ) as the high-frequency component;

[0051] c3) The high frequency and medium frequency are respectively distributed to the supercapacitor and the battery, and the low frequency component is used as the direct grid-connected power, that is:

[0052] P con (t)=u1(t) (16)

[0053] P BESS (t)=u2(t)+u3(t)+...+u th (t) (17)

[0054] P SC (t) = u th+1 (t)+u th+2 (t)...+u K (t) (18)

[0055] Among them, P con (t) is the grid-connected power at time t, P BESS (t) and P SC (t) are the power leveling task reference values ​​of the battery and supercapacitor, u th (t) represents the wind power at frequency ω th The modal components at .

[0056] Furthermore, the power of supercapacitors and batteries is as follows:

[0057]

[0058]

[0059] Among them, η BESS,c and η BESS,d Indicates the charging and discharging efficiency of the battery, η SC,c and η SC,d Indicates the charging and discharging efficiency of the supercapacitor. B ' ESS (t) and P S ' C (t) represents the actual charging and discharging power of the battery and supercapacitor.

[0060] Further, the steps of calculating the energy storage capacity of the supercapacitor and the battery include:

[0061] d1) Calculate the capacity of the supercapacitor and battery at time n relative to the initial time, that is:

[0062]

[0063]

[0064] Where Δt is the sampling period of wind power; E B (n) and E SC (n) represents the capacity of the battery energy storage system and supercapacitor at time n relative to the initial time; E B0 and E SC0 Indicates the initial capacity of the battery energy storage system and supercapacitor;

[0065] d2) Construct the state of charge constraints of supercapacitors and batteries at any time, that is:

[0066]

[0067]

[0068] In the formula, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery SOC constraint, SOC SC,max and SOC SC,min represents the upper and lower limits of the supercapacitor SOC constraint, E B,rated and E SC,rated Indicates the rated capacity of batteries and supercapacitors.

[0069] The charging and discharging power during the energy storage operation process does not exceed its rated power, that is,

[0070] P B,rated =max(|P′ BESS (t)|) (25)

[0071] P SC,rated =max(|P′ SC (t)|) (26)

[0072] Among them, P B,rated and P SC,rated is the rated power of the battery and supercapacitor.

[0073] d3) Calculate the rated capacity of the battery and supercapacitor, that is:

[0074]

[0075]

[0076] In the formula, E B,rated 、E SC,rated is the rated capacity of the battery and supercapacitor.

[0077] Furthermore, the grid reliability evaluation index includes the voltage stability index VSI, the maximum value of the grid branch dielectric constant Be, the unbalanced power P of the entire grid, and the un ;

[0078] The voltage stability indicators are as follows:

[0079]

[0080] Among them, N n is a collection of grid nodes;

[0081] Voltage stability index VSI at node i i As shown below:

[0082]

[0083] P i -jQ i =V i * ×I i (31)

[0084] |V i | 4 -b i |V i | 2 +c i =0 (32)

[0085]

[0086]

[0087]

[0088]

[0089] in, represents the Thevenin equivalent voltage at grid node i, V i represents the voltage at grid node i, and are the Thevenin equivalent resistance and reactance at grid node i, respectively, P i and Q i represents the active and reactive power fed into by node i, Represents the node voltage V i The conjugate voltage of I i represents the current at the grid node i; b i and c i is the coefficient;

[0090] The maximum value of the dielectric constant Be of the power grid branch is as follows:

[0091]

[0092] When the unit current flows into the path (i, j) between the generator and the load node, the current generated on the branch (m, n) and B e (m,n) is as follows:

[0093]

[0094] I mn (g,l)=y mn (U m -U n )=y mn [(Z mg -Z ml )-(Z ng -Z nl )] (39)

[0095] U k =Z kg -Z kl (40)

[0096]

[0097] Where U i is the voltage at node i, I i is the current injected into node i, Z ij is the impedance between nodes i and j; Z kg is the impedance between nodes k and g in the shortest path, Z kl is the impedance between nodes k and l in the shortest path; U k , U m, U n is the voltage of node k, node m, and node n; k is any node on the shortest path between the generator node and the load node; V G is the set of generator nodes in the power grid, V L is the set of load nodes in the power grid, W g and W l are the weights of the generator node g and the load node l respectively; I mn (g, l) represents the current generated in branch (m, n) when unit current flows from generator node g into load node l; y mn represents the admittance of branch (m,n); Z NN Represents impedance; Z mg is the impedance between nodes m and g; Z ml is the impedance between nodes m and l; Z ng is the impedance between nodes n and g; Z nl is the impedance between nodes n and l; I N Indicates current; U N Indicates voltage;

[0098] The overall unbalanced power of the power grid P un As shown below:

[0099]

[0100]

[0101] Where P un,j is the unbalanced power of the jth region; A j is the number of nodes in the jth region, P G,i (t) represents the active output of the ith generator at time t, P D,i (t) represents the load power at bus i at time t. W,i (t) represents the output power of the i-th wind farm at time t; N A The number of regions after the distribution network is divided.

[0102] Further, the two-stage optimization configuration model includes a first-stage optimization configuration model and a second-stage optimization configuration model;

[0103] Among them, the objective function minf of the first stage optimization configuration model c As shown below:

[0104]

[0105]

[0106]

[0107] Among them, r is the funding discount rate, T L is the entire life cycle of energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors respectively; C op is the daily operating cost of hybrid energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors, and are the unit power costs of batteries and supercapacitors, respectively; and are the unit capacity costs of batteries and supercapacitors respectively; P B,rated and P SC,rated is the rated power of the battery and supercapacitor; E B,rated and E SC,rated Indicates the rated capacity of the battery and supercapacitor; T indicates the time period; and is the unit operating cost of supercapacitors and batteries; P′ BESS (t) and P′ SC (t) represents the actual charging and discharging power of the battery and supercapacitor. Δt is the sampling period;

[0108] The objective function minf of the second-stage optimization configuration model v As shown below:

[0109] minf v =ρ1VSI+ρ2B e +ρ3P un (47)

[0110] Among them, ρ1, ρ2 and ρ3 are the weight coefficients of the reliability index, and they satisfy ρ3=2ρ1=2ρ2. e , P un It is an indicator of voltage stability, the maximum dielectric constant of a power grid branch, and the unbalanced power of the entire power grid.

[0111] Furthermore, the constraints of the two-stage optimization configuration model include power flow constraints, voltage and phase constraints of bus nodes, generator output constraints, load power, wind power, generator power, hybrid energy storage system power and power purchase from the upper grid, supercapacitor and battery charging and discharging power constraints, supercapacitor and battery capacity constraints, supercapacitor and battery SOC constraints, initial capacity and terminal capacity balance constraints of hybrid energy storage during dispatch, wind power direct grid-connected power constraints, and distribution network branch constraints;

[0112] The power flow constraints are as follows:

[0113]

[0114]

[0115] Among them, P W,i (t) and Q W,i (t) is the wind power at bus i at time t, P G,i (t) and Q G,i (t) is the generator power at bus i at time t, P D,i (t) and Q D,i (t) is the load power at bus i at time t, P HESS,i (t) represents the charging and discharging power of the hybrid energy storage at bus i at time t, V i (t) and θ i (t) is the voltage amplitude and phase at bus i at time t, C ij and B ij is the conductance and susceptance between busbars i and j, N b (i) is the set of other buses connected to bus i. j (t) and θ j (t) is the voltage amplitude and phase at bus j at time t;

[0116] The voltage and phase constraints of the bus node are as follows:

[0117]

[0118] in, and are the upper and lower limits of the voltage amplitude at bus node i, and are the upper and lower limits of the voltage phase at bus i.

[0119] The generator output constraints are as follows:

[0120]

[0121] in, and are the upper and lower limits of the active output of the generator at bus i, and Indicates the upper and lower limits of the reactive output of the generator at bus i.

[0122] The constraints on the relationship between load power, wind power, generator power, hybrid energy storage system power and power purchased from the upper grid are as follows:

[0123] P D (t) = P grid (t)+P W (t)+PG (t)+P HESS (t) (52)

[0124] Among them, P D (t) represents the total load power in the distribution network at time t, P grid (t) represents the power purchased by the distribution network from the upper grid at time t, P W (t) represents the wind power in the distribution network at time t, P G (t) represents the active output of the generator in the distribution network at time t, P HESS (t) represents the output power of the hybrid energy storage system in the distribution network at time t;

[0125] The charging and discharging power constraints of supercapacitors and batteries are as follows:

[0126]

[0127] Among them, P' SC (t) and P' BESS (t) is the charge and discharge power of the supercapacitor and battery at time t, P SC,rated and P B,rated is the rated power of the supercapacitor and battery.

[0128] The supercapacitor and battery capacity constraints are as follows:

[0129]

[0130] Among them, E SC (t) and E B (t) is the capacity of the supercapacitor and battery at time t, E SC,rated and E B,rated is the rated capacity of the supercapacitor and battery;

[0131] The supercapacitor and battery SOC constraints are as follows:

[0132]

[0133] Among them, SOC SC (t) and SOC B (t) is the state of charge of the supercapacitor and battery at time t, SOC SC,max and SOC SC,min is the upper and lower limits of the SOC of the supercapacitor, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery's SOC;

[0134] The initial capacity and terminal capacity balance constraints of hybrid energy storage during dispatch are as follows:

[0135] E SC (0) = E SC (T), E B (0) = E B (T) (56)

[0136] Among them, E SC (0) and E B (0) represents the capacity of the energy storage system at the initial moment, E SC (T) and E B (T) represents the capacity of the energy storage system at the end time;

[0137] The power constraints for direct grid connection of wind power are as follows:

[0138]

[0139] Among them, F Δt is the wind power fluctuation rate within Δt, and is the maximum and minimum wind power in the Δt period, P N is the rated output power of the fan, F lim Indicates the fluctuation rate limit that the wind power grid-connected power needs to meet.

[0140] The distribution network branch constraints are as follows:

[0141]

[0142] Among them, the binary variable U (m,n) (t) represents the switch state of branch (m,n) at time t, U (m,n) (t) = 0 means the branch switch is disconnected, U (m,n) (t) = 1 means the branch switch is closed; N b is the set of busbar nodes in the distribution network, L i is the number of branches of loop i in the distribution network, N Li is the set of branches in loop i in the distribution network.

[0143] The technical effect of the present invention is unquestionable. The present invention proposes a two-stage energy storage hybrid configuration method based on dynamic network partitioning, which can reduce wind power fluctuations while improving the system's operational reliability. The present invention determines the energy storage capacity and power with the goal of hybrid energy storage investment cost economy, and determines the system's topological structure and number of partitions at each moment with the goal of minimizing the system voltage stability index, edge betweenness, and unbalanced power values ​​in each region, which can effectively improve the operational reliability of the distribution network under wind power penetration. BRIEF DESCRIPTION OF THE DRAWINGS

[0144] Figure 1It is the flow chart of EM algorithm;

[0145] Figure 2 This is the schematic diagram of Nataf transformation;

[0146] Figure 3 is the Thevenin equivalent system diagram of node i;

[0147] Figure 4 This is the schematic diagram of the dynamic reconstruction partitioning method. DETAILED DESCRIPTION

[0148] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0149] Embodiment 1:

[0150] See also Figures 1 to 4 , a distribution network reliability improvement method based on hybrid energy storage and dynamic network reconstruction, comprising the following steps:

[0151] 1) Generate wind farm wind power scenario;

[0152] 2) Based on the variational mode decomposition algorithm, the generated wind farm power is decomposed into high-frequency, medium-frequency and low-frequency components, and then the high-frequency and medium-frequency components are allocated to supercapacitors and batteries respectively, and the low-frequency components are used as direct grid-connected power, and the energy storage capacity and power of supercapacitors and batteries are calculated;

[0153] 3) Calculate the grid reliability assessment index;

[0154] 4) Establish a two-stage optimization configuration model;

[0155] 5) Solve the two-stage optimization configuration model to optimize the grid reliability assessment indicators.

[0156] Embodiment 2:

[0157] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as that of Example 1, and further, the step of generating a wind farm wind power scenario includes:

[0158] a1) Use the Gaussian mixture model to describe the probability distribution of wind speed in the wind farm f(x), that is:

[0159]

[0160]

[0161]

[0162] Among them, K represents the total number of Gaussian components, p k (x|θ k ) is given by the parameter θ k The k-th Gaussian component expression is determined by k ∈[0,1] is the weight corresponding to the kth Gaussian component, μ is the mathematical expectation of the random variable x, σ 2 is the variance of the random variable x, θ k =(μ,σ 2 ) represents the set of expected and variance distribution functions; x is the observed sample;

[0163] a2) Based on the probability distribution of wind speed f(x), construct the correlation coefficient matrix C of the relevant random vector Y in the wind farm y ,Right now:

[0164]

[0165] Among them, ρ ij are any two variables y in the random vector Y i and j The correlation coefficient between .

[0166] a3) By using formula (5), the correlated random vector Y is converted into a correlated standard normal random vector H = (h1, h2, ..., h c );

[0167]

[0168] Among them, Φ i (h i ) is the variable h i The standard normal distribution function of is the variable y i The cumulative distribution function F i The inverse function of (.);

[0169] Among them, the correlation coefficient matrix C of the related standard normal random vector H h As shown below:

[0170]

[0171] Among them, ρ ij are any two variables h in the vector H i and h j The correlation coefficient between .

[0172] a4) Based on Cholesky decomposition, the correlation coefficient matrix C h Decompose into upper triangular matrix BT And the lower triangular matrix B, we get:

[0173] C h =BB T (7)

[0174] a5) Convert the correlated standard normal random vector H into an independent standard normal random vector L, that is:

[0175] L=B -1 H (8)

[0176] a6) Using the Nataf inverse transform, the standard normal random vector L is converted into a sample of related wind speed random variables to generate a wind farm wind power scenario;

[0177] The expression of wind power in a wind farm is as follows:

[0178]

[0179] Among them, v is the real-time wind speed of the wind farm, v in and v out are the cut-in and cut-out wind speeds of the wind turbines in the wind farm, v N is the rated wind speed of the wind turbine in the wind farm, P N is the rated power of the wind turbine in the wind farm; P W is the output power of the wind farm.

[0180] Embodiment 3:

[0181] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-2, and further, in the Gaussian mixture model, the parameter ω k and parameter θ k Solved by expectation maximization algorithm;

[0182] Solving for parameter ω k and parameter θ k The steps include:

[0183] b1) Initialization parameter ω k and θ k , calculate the jth observation sample x j The corresponding k hidden variables γ jk ,Right now:

[0184]

[0185] b2) Based on latent variable γ jk , update the parameter ω k and θ k ,get:

[0186]

[0187]

[0188]

[0189] Where N is the number of observed samples; μ k , is the mathematical expectation and variance;

[0190] b3) Repeat steps 1) to 2) until the parameter ω k and parameter θ k convergence.

[0191] Embodiment 4:

[0192] A distribution network reliability improvement method based on hybrid energy storage and dynamic network reconstruction, the technical content is the same as any one of embodiments 1-3, further, the correlation coefficient ρ of the related random variable Y ij The correlation coefficient ρ with the related standard normal random variable H i ' j The following relationship is satisfied:

[0193]

[0194] Among them, f yiyj (y i ,y j ) is the random variable y i and j The joint probability density function, φ2(h i ,h j ,ρ i ' j ) is the random variable h i and h j The joint probability density function of i , σ j is the random variable y i and j The corresponding standard deviation. i , μ j is the random variable y i and j The corresponding mathematical expectation. ij is the relationship coefficient.

[0195] Embodiment 5:

[0196] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-4, and further, the step of decomposing the generated wind farm power into high-frequency, medium-frequency and low-frequency components based on a variational mode decomposition algorithm comprises:

[0197] c1) Decompose the wind farm wind power into K modal functions with limited bandwidth and obtain:

[0198]

[0199] Among them, u k (t) is the wind farm output power P W (t) the kth modal component after decomposition, represents the gradient operation, ‖·‖2 represents L 2 -norm; ω k is the weight corresponding to the kth Gaussian component; t is the time; P W (t) is the wind farm output power at time t; δ(t) is the pulse function;

[0200] c2) For K mode functions with limited bandwidth, the components with frequencies less than ω2 are regarded as low-frequency components, and the frequency range is [ω2, ω th ) as the intermediate frequency component, and the frequency range is [ω th ,ω K ) as the high-frequency component;

[0201] c3) The high frequency and medium frequency are respectively distributed to the supercapacitor and the battery, and the low frequency component is used as the direct grid-connected power, that is:

[0202] P con (t)=u1(t) (16)

[0203] P BESS (t)=u2(t)+u3(t)+...+u th (t) (17)

[0204] P SC (t) = u th+1 (t)+u th+2 (t)...+u K (t) (18)

[0205] Among them, P con (t) is the grid-connected power at time t, P BESS (t) and P SC (t) are the power leveling task reference values ​​of the battery and supercapacitor, u th (t) represents the wind power at frequency ω th The modal components at .

[0206] Embodiment 6:

[0207] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-5, and further, the power of the supercapacitor and the battery is as follows:

[0208]

[0209]

[0210] Among them, η BESS,c and η BESS,d Indicates the charging and discharging efficiency of the battery, η SC,c and η SC,d Indicates the charging and discharging efficiency of the supercapacitor. B ' ESS (t) and P S ' C (t) represents the actual charging and discharging power of the battery and supercapacitor.

[0211] Embodiment 7:

[0212] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-6, and further, the step of calculating the energy storage capacity of a supercapacitor and a battery comprises:

[0213] d1) Calculate the capacity of the supercapacitor and battery at time n relative to the initial time, that is:

[0214]

[0215]

[0216] Where Δt is the sampling period of wind power; E B (n) and E SC (n) represents the capacity of the battery energy storage system and supercapacitor at time n relative to the initial time; E B0 and E SC0 Indicates the initial capacity of the battery energy storage system and supercapacitor;

[0217] d2) Construct the state of charge constraints of supercapacitors and batteries at any time, that is:

[0218]

[0219]

[0220] In the formula, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery SOC constraint, SOC SC,max and SOC SC,minrepresents the upper and lower limits of the supercapacitor SOC constraint, E B,rated and E SC,rated Indicates the rated capacity of batteries and supercapacitors.

[0221] The charging and discharging power during the energy storage operation process does not exceed its rated power, that is,

[0222] P B,rated =max(|P′ BESS (t)|) (25)

[0223] P SC,rated =max(|P′ SC (t)|) (26)

[0224] Among them, P B,rated and P SC,rated is the rated power of the battery and supercapacitor.

[0225] d3) Calculate the rated capacity of the battery and supercapacitor, that is:

[0226]

[0227]

[0228] In the formula, E B,rated 、E SC,rated is the rated capacity of the battery and supercapacitor.

[0229] Embodiment 8:

[0230] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of embodiments 1-7, and further, the grid reliability evaluation index includes a voltage stability index VSI, a maximum value of a grid branch dielectric constant Be, and an overall unbalanced power P of the grid. un ;

[0231] The voltage stability indicators are as follows:

[0232]

[0233] Among them, N n is a collection of grid nodes;

[0234] Voltage stability index VSI at node i i As shown below:

[0235]

[0236] P i -jQ i =V i* ×I i (31)

[0237] |V i | 4 -b i |V i | 2 +c i =0 (32)

[0238]

[0239]

[0240]

[0241]

[0242]

[0243] in, represents the Thevenin equivalent voltage at grid node i, V i represents the voltage at grid node i, and are the Thevenin equivalent resistance and reactance at grid node i, respectively, P i and Q i represents the active and reactive power fed into by node i, Represents the node voltage V i The conjugate voltage of I i represents the current at the grid node i; b i and c i is the coefficient;

[0244] The maximum value of the dielectric constant Be of the power grid branch is as follows:

[0245]

[0246] When the unit current flows into the path (i, j) between the generator and the load node, the current generated on the branch (m, n) and B e (m,n) is as follows:

[0247]

[0248] I mn (g,l)=y mn (U m -U n )=y mn [(Z mg -Z ml )-(Z ng -Znl )] (39)

[0249] U k =Z kg -Z kl (40)

[0250]

[0251] Where U i is the voltage at node i, I i is the current injected into node i, Z ij is the impedance between nodes i and j; Z kg is the impedance between nodes k and g in the shortest path, Z kl is the impedance between nodes k and l in the shortest path; U k , U m , U n is the voltage of node k, node m, and node n; k is any node on the shortest path between the generator node and the load node; V G is the set of generator nodes in the power grid, V L is the set of load nodes in the power grid, W g and W l are the weights of the generator node g and the load node l respectively; I mn (g, l) represents the current generated in branch (m, n) when unit current flows from generator node g into load node l; y mn represents the admittance of branch (m,n); Z NN Represents impedance; Z mg is the impedance between nodes m and g; Z ml is the impedance between nodes m and l; Z ng is the impedance between nodes n and g; Z nl is the impedance between nodes n and l; I N Indicates current; U N Indicates voltage;

[0252] The overall unbalanced power of the power grid P un As shown below:

[0253]

[0254]

[0255] Where P un,j is the unbalanced power of the jth region; A j is the number of nodes in the jth region, P G,i (t) represents the active output of the ith generator at time t, P D,i(t) represents the load power at bus i at time t. W,i (t) represents the output power of the i-th wind farm at time t; N A The number of regions after the distribution network is divided.

[0256] Embodiment 9:

[0257] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-8, and further, the two-stage optimization configuration model includes a first-stage optimization configuration model and a second-stage optimization configuration model;

[0258] Among them, the objective function minf of the first stage optimization configuration model c As shown below:

[0259]

[0260]

[0261]

[0262] Among them, r is the funding discount rate, T L is the entire life cycle of energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors respectively; C op is the daily operating cost of hybrid energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors, and are the unit power costs of batteries and supercapacitors, respectively; and are the unit capacity costs of batteries and supercapacitors respectively; P B,rated and P SC,rated is the rated power of the battery and supercapacitor; E B,rated and E SC,rated Indicates the rated capacity of the battery and supercapacitor; T indicates the time period; and is the unit operating cost of supercapacitors and batteries; P′ BESS (t) and P′ SC (t) represents the actual charging and discharging power of the battery and supercapacitor. Δt is the sampling period;

[0263] The objective function minf of the second-stage optimization configuration model v As shown below:

[0264] minf v =ρ1VSI+ρ2Be +ρ3P un (47)

[0265] Among them, ρ1, ρ2 and ρ3 are the weight coefficients of the reliability index, and they satisfy ρ3=2ρ1=2ρ2. e , P un It is an indicator of voltage stability, the maximum dielectric constant of a power grid branch, and the unbalanced power of the entire power grid.

[0266] Embodiment 10:

[0267] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, the technical content of which is the same as any one of Embodiments 1-9, and further, the constraints of the two-stage optimization configuration model include power grid flow constraints, voltage and phase constraints of bus nodes, generator output constraints, load power, wind power, generator power, hybrid energy storage system power and power purchase power constraints from the upper power grid, supercapacitor and battery charging and discharging power constraints, supercapacitor and battery capacity constraints, supercapacitor and battery SOC constraints, initial capacity and termination capacity balance constraints of hybrid energy storage during scheduling, wind power direct grid-connected power constraints, and distribution network branch constraints;

[0268] The power flow constraints are as follows:

[0269]

[0270]

[0271] Among them, P W,i (t) and Q W,i (t) is the wind power at bus i at time t, P G,i (t) and Q G,i (t) is the generator power at bus i at time t, P D,i (t) and Q D,i (t) is the load power at bus i at time t, P HESS,i (t) represents the charging and discharging power of the hybrid energy storage at bus i at time t, V i (t) and θ i (t) is the voltage amplitude and phase at bus i at time t, C ij and B ij is the conductance and susceptance between busbars i and j, N b (i) is the set of other buses connected to bus i. j (t) and θ j (t) is the voltage amplitude and phase at bus j at time t;

[0272] The voltage and phase constraints of the bus node are as follows:

[0273]

[0274] in, and are the upper and lower limits of the voltage amplitude at bus node i, and are the upper and lower limits of the voltage phase at bus i.

[0275] The generator output constraints are as follows:

[0276]

[0277] in, and are the upper and lower limits of the active output of the generator at bus i, and Indicates the upper and lower limits of the reactive output of the generator at bus i.

[0278] The constraints on the relationship between load power, wind power, generator power, hybrid energy storage system power and power purchased from the upper grid are as follows:

[0279] P D (t) = P grid (t)+P W (t)+P G (t)+P HESS (t) (52)

[0280] Among them, P D (t) represents the total load power in the distribution network at time t, P grid (t) represents the power purchased by the distribution network from the upper grid at time t, P W (t) represents the wind power in the distribution network at time t, P G (t) represents the active output of the generator in the distribution network at time t, P HESS (t) represents the output power of the hybrid energy storage system in the distribution network at time t;

[0281] The charging and discharging power constraints of supercapacitors and batteries are as follows:

[0282]

[0283] Among them, P' SC (t) and P' BESS (t) is the charge and discharge power of the supercapacitor and battery at time t, P SC,rated and P B,rated is the rated power of the supercapacitor and battery.

[0284] The supercapacitor and battery capacity constraints are as follows:

[0285]

[0286] Among them, E SC (t) and E B (t) is the capacity of the supercapacitor and battery at time t, E SC,rated and E B,rated is the rated capacity of the supercapacitor and battery;

[0287] The supercapacitor and battery SOC constraints are as follows:

[0288]

[0289] Among them, SOC SC (t) and SOC B (t) is the state of charge of the supercapacitor and battery at time t, SOC SC,max and SOC SC,min is the upper and lower limits of the SOC of the supercapacitor, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery's SOC;

[0290] The initial capacity and terminal capacity balance constraints of hybrid energy storage during dispatch are as follows:

[0291] E SC (0) = E SC (T), E B (0) = E B (T) (56)

[0292] Among them, E SC (0) and E B (0) represents the capacity of the energy storage system at the initial moment, E SC (T) and E B (T) represents the capacity of the energy storage system at the end time;

[0293] The power constraints for direct grid connection of wind power are as follows:

[0294]

[0295] Among them, F Δt is the wind power fluctuation rate within Δt, and is the maximum and minimum wind power in the Δt period, P N is the rated output power of the fan, F lim Indicates the fluctuation rate limit that the wind power grid-connected power needs to meet.

[0296] The distribution network branch constraints are as follows:

[0297]

[0298] Among them, the binary variable U (m,n) (t) represents the switch state of branch (m,n) at time t, U (m,n) (t) = 0 means the branch switch is disconnected, U (m,n) (t) = 1 means the branch switch is closed; N b is the set of busbar nodes in the distribution network, L i is the number of branches of loop i in the distribution network, N Li is the set of branches in loop i in the distribution network.

[0299] Embodiment 11:

[0300] A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, comprising the following steps:

[0301] 1) Generate relevant wind farm wind power scenarios, the details are as follows:

[0302] The Gaussian Mixture Model (GMM) is used to describe the probability distribution of wind speed in a wind farm. Its specific form is:

[0303]

[0304]

[0305]

[0306] Among them, K represents the total number of Gaussian components, p k (x|θ k ) is given by the parameter θ k The k-th Gaussian component expression is determined by k ∈[0,1] is the weight corresponding to the kth Gaussian component, μ is the mathematical expectation of the random variable x, σ 2 is the variance of the random variable x, θ=(μ,σ 2 ) represents the set of distribution function expectation and variance.

[0307] Assume X = (x1, x2, ..., x N ) is N observation samples, γ j =(γ j1 ,γ j2 ,...,γ jk,) are k latent variables corresponding to the jth observation sample. Due to the existence of latent variables, it is very difficult to directly obtain the parameters of GMM. Therefore, the expectation-maximization algorithm (EM) is introduced. The specific process is as follows: Figure 1 As shown. The EM algorithm is an iterative algorithm, and each iteration consists of two parts: the E step and the M step. For the E step, first initialize the parameter ω k and θ k , and then calculate the hidden variable according to formula (4)

[0308]

[0309] Based on the latent variables obtained in step E, step M updates the parameters ω of GMM according to formulas (5)-(7) k and θ k

[0310]

[0311]

[0312]

[0313] Repeat steps E and M. When the GMM parameters converge, the model finds a feasible solution.

[0314] In order to solve the problem of wind speed correlation between adjacent wind farms in the distribution network, Nataf transformation is introduced to convert the independent normal random vectors obtained by Latin hypercube sampling into correlated random wind speed vectors, such as Figure 2 Assume that there are c related random vectors Y=(y1,y2,...,y c ), and its correlation coefficient matrix is

[0315]

[0316] Among them, ρ ij are any two variables y in the random vector Y i and j The correlation coefficient between .

[0317] According to formula (9), i.e., the equal probability transformation principle, the related random vector Y is transformed into the related standard normal random vector H = (h1, h2, ..., h c )

[0318]

[0319] Among them, Φ i (h i ) is the variable h i The standard normal distribution function of is the variable y i The cumulative distribution function F i The inverse function of (.).

[0320] For a correlated standard normal random vector H, its correlation coefficient matrix is

[0321]

[0322] Among them, ρ′ ij are any two variables h in the vector H i and h j The correlation coefficient between .

[0323] Correlation coefficient ρ of the related random variables Y ij The correlation coefficient ρ′ of the standard normal random variable H is ij The following relationship is satisfied between

[0324]

[0325] Among them, f yiyj (y i ,y j ) is the random variable y i and j The joint probability density function, φ2(h i ,h j ,ρ′ ij ) is the random variable h i and h j The joint probability density function of .

[0326] Based on Cholesky decomposition, the correlation coefficient matrix C h Decompose into upper triangular matrix B T and the lower triangular matrix B

[0327] C h =BB T (12)

[0328] Based on the lower triangular matrix B and formula (13), the correlated standard normal random vector H is transformed into an independent standard normal random vector L

[0329] L=B -1 H (13)

[0330] Through the Nataf inverse transformation, the independent standard normal random variable samples obtained by sampling can be converted into correlated wind speed random variable samples.

[0331] Based on the generated correlation wind speed samples, the output power of the wind farm can be obtained through the functional relationship between wind power and wind speed (14):

[0332]

[0333] Among them, v is the real-time wind speed of the wind farm, v in and v out are the cut-in and cut-out wind speeds of the wind turbines in the wind farm, v N is the rated wind speed of the wind turbine in the wind farm, P N is the rated power of the wind turbines in the wind farm.

[0334] 2) Based on the variational modal decomposition (VMD) algorithm, the generated wind farm power is decomposed into high-frequency, medium-frequency and low-frequency components. The high-frequency component is assigned to the supercapacitor, the medium-frequency component is assigned to the battery, and the low-frequency component is used as the direct grid-connected power. The configured energy storage capacity and power are calculated. The details are as follows

[0335] For the output power P(t) of a wind farm, the VMD algorithm can decompose the power signal into K modal functions with limited bandwidth. The variational problem is expressed as

[0336]

[0337] Among them, u k (t) is the kth modal component after decomposition of the wind farm output power P(t), represents the gradient operation, ‖·‖2 represents L 2 -norm.

[0338] For the K modal components after VMD decomposition, their frequencies are from ω2 to ω th The intermediate frequency components between are distributed to the battery, and the frequency between ω th and ω K The high frequency components between are allocated to the supercapacitor. Therefore, the power allocation task is expressed as

[0339] P con (t)=u1(t) (16)

[0340] P BESS (t)=u2(t)+u3(t)+...+u th (t) (17)

[0341] P SC (t) = u th+1 (t)+u th+2 (t)...+u K (t) (18)

[0342] Among them, P con (t) is the grid-connected power at time t, PBESS (t) and P SC (t) are the power leveling task reference values ​​of the battery and supercapacitor, u th (t) represents the wind power at frequency ω th The modal components at .

[0343] Considering the charging and discharging efficiency of the energy storage system, the actual charging and discharging power of the battery and supercapacitor is

[0344]

[0345]

[0346] Among them, η BESS,c and η BESS,d Indicates the charging and discharging efficiency of the battery, η SC,c and η SC,d Indicates the charging and discharging efficiency of the supercapacitor.

[0347] Assume that the initial capacities of the battery energy storage system and the supercapacitor are E B0 and E SC0 , then the capacity of the battery energy storage system and the supercapacitor at time n relative to the initial time is

[0348]

[0349]

[0350] Among them, Δt is the sampling period of wind power.

[0351] The state of charge (SOC) of the energy storage system at any time must meet

[0352]

[0353]

[0354] In the formula, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery SOC constraint, SOC SC,max and SOC SC,min represents the upper and lower limits of the supercapacitor SOC constraint, E B,rated and E SC,rated Indicates the rated capacity of batteries and supercapacitors.

[0355] The charging and discharging power during the energy storage operation process does not exceed its rated power, that is,

[0356] P B,rated =max(|P′ BESS(t)|) (25)

[0357] P SC,rated =max(|P′ SC (t)|) (26)

[0358] Among them, P B,rated and P SC,rated is the rated power of the battery and supercapacitor.

[0359] Calculate the rated capacity of batteries and supercapacitors based on the maximum and minimum capacity and SOC constraints during energy storage operation

[0360]

[0361]

[0362] The initial capacity E of the energy storage system is calculated by formulas (21), (23) and (27): B0 Similarly, the initial capacity of the supercapacitor can also be obtained by formulas (22), (24) and (28).

[0363] 3) Calculate the grid reliability evaluation index, the details are as follows:

[0364] In order to describe the sensitivity of system node voltage to voltage collapse events, the voltage stability index (VSI) is introduced. Figure 3 As shown, according to the Thevenin equivalent circuit of any node in the power grid, we can get

[0365]

[0366] P i -jQ i =V i * ×I i (30)

[0367] in, represents the Thevenin equivalent voltage at grid node i, V i represents the voltage at grid node i, and are the Thevenin equivalent resistance and reactance at grid node i, respectively, P i and Q i represents the active and reactive power fed into by node i, Represents the node voltage V i The conjugate voltage.

[0368] According to formulas (29) and (28), we can get

[0369] |V i | 4 -b i |V i | 2 +c i =0 (31) where coefficient b i and c i It can be obtained from formulas (32) and (33), that is,

[0370]

[0371]

[0372] If formula (34) is satisfied, formula (31) will have a feasible solution

[0373]

[0374] Voltage stability index VSI at node i i Defined as

[0375]

[0376] In order to describe the voltage stability of the entire power grid, the voltage stability index VSI of the entire network is defined, that is,

[0377]

[0378] Among them, N n is a collection of grid nodes.

[0379] Wind power penetration increases the probability of power outages in distribution networks. Power outages are generally caused by disconnection of power grid lines, and line failures that have a greater impact on the power grid are an important cause of cascading failures in the power grid. Therefore, the dielectric number is introduced to identify vulnerable branches in the distribution network. Its calculation process is as follows.

[0380] The node impedance matrix of a distribution network system with N nodes is

[0381]

[0382] Among them, U i is the voltage at node i, I i is the current injected into node i, Z ij is the impedance between nodes i and j.

[0383] Assume that there is a unit current flowing from the generator node g to the load node l in the power grid, and k is any node on the shortest path between the generator node and the load node, then the voltage at node k is

[0384] Uk =Z kg -Z kl (38)

[0385] Among them, Z kg is the impedance between nodes k and g in the shortest path, Z kl is the impedance between nodes k and l in the shortest path.

[0386] For branch (m,n), its dielectric constant is defined as the sum of the currents generated on branch (m,n) when unit current flows into path (i,j) between generator and load node, that is,

[0387]

[0388] Among them, V G is the set of generator nodes in the power grid, V L is the set of load nodes in the power grid, W g and W l are the weights of the generator node g and the load node l, respectively. The values ​​can be selected from the actual power of the generator and the load. mn (m,n) represents the current generated in branch (m,n) when unit current flows from generator node g into load node l.

[0389] I mn (m,n) can be calculated by formula (40)

[0390] I mn (m,n)=y mn (U m -U n )=y mn [(Z mg -Z ml )-(Z ng -Z nl )] (40)

[0391] Among them, y mn is the admittance of branch (m,n).

[0392] In order to describe the overall vulnerability of the power grid, the maximum value of the dielectric constant in all branches is defined as B e ,Right now

[0393]

[0394] A dynamic partitioning method is proposed, which divides the distribution network into several arbitrary areas at each reconstruction, while trying to keep the power in each area balanced locally. Figure 4As shown in Figure 1, when a branch in the power grid fails, the power in the remaining areas can still be balanced, thereby improving the reliability of power grid operation. After each network reconstruction, the unbalanced power of the jth area is

[0395]

[0396] Among them, A j is the number of nodes in the jth region, P G,i (t) represents the active output of the ith generator at time t, P D,i (t) represents the load power at bus i at time t.

[0397] The unbalanced power of the entire power grid is defined as P un ,Right now

[0398]

[0399] Among them, N A The number of regions after the distribution network is divided.

[0400] 4) Establish a two-stage optimization configuration model, the details are as follows:

[0401] Based on the wind power scenario generated in the first part and the power allocation strategy proposed in the second part, the first stage aims to minimize the average daily investment and operating cost of the hybrid energy storage system, that is,

[0402]

[0403] Among them, r is the funding discount rate, T L It is the entire life cycle of energy storage.

[0404] The investment cost of supercapacitors and batteries can be calculated using formula (45):

[0405]

[0406] Among them, C B,inv and C SC,inv are the investment costs of batteries and supercapacitors, and are the unit power costs of batteries and supercapacitors, respectively, in RMB / kW. and They are the unit capacity costs of batteries and supercapacitors respectively, and their units are RMB / kWh.

[0407] The daily operating cost of the hybrid energy storage system can be calculated using formula (46):

[0408]

[0409] Among them, C op is the daily operating cost of hybrid energy storage, T is the scheduling period of hybrid energy storage, and are the unit capacity operating costs of supercapacitors and batteries, respectively, P' SC (t) and P' BESS (t) are the actual charging and discharging powers of the supercapacitor and the battery, respectively, and Δt represents the sampling period.

[0410] Based on the hybrid energy storage system configured in the first stage, the network topology is continuously changed in the second stage through the dynamic reconstruction partitioning method proposed in the third part, so that the grid voltage stability index, dielectric constant and unbalanced power are minimized, that is,

[0411] min f v =ρ1VSI+ρ2B e +ρ3P un (47)

[0412] Among them, ρ1, ρ2 and ρ3 are weight coefficients of reliability indicators, and satisfy ρ3=2ρ1=2ρ2.

[0413] The proposed two-stage optimization configuration method needs to meet the power grid flow constraints:

[0414]

[0415]

[0416] Among them, P W,i (t) and Q W,i (t) is the wind power at bus i at time t, P G,i (t) and Q G,i (t) is the generator power at bus i at time t, P D,i (t) and Q D,i (t) is the load power at bus i at time t, P HESS,i (t) represents the charging and discharging power of the hybrid energy storage at bus i at time t, V i (t) and θ i (t) is the voltage amplitude and phase at bus i at time t, C ij and B ij is the conductance and susceptance between busbars i and j, N b (i) is the set of other buses connected to bus i.

[0417] The voltage and phase of any bus node in the distribution network must satisfy the constraints

[0418]

[0419] in, and are the upper and lower limits of the voltage amplitude at bus node i, and are the upper and lower limits of the voltage phase at bus i.

[0420] The generator output at time t in the distribution network must satisfy

[0421]

[0422] in, and are the upper and lower limits of the active output of the generator at bus i, and Indicates the upper and lower limits of the reactive output of the generator at bus i.

[0423] The load power, wind power, generator power, hybrid energy storage system power and power purchased from the upper grid at any time in the distribution network should satisfy the equation

[0424] P D (t) = P grid (t)+P W (t)+P G (t)+P HESS (t) (52)

[0425] Among them, P D (t) represents the total load power in the distribution network at time t, P grid (t) represents the power purchased by the distribution network from the upper grid at time t, P W (t) represents the wind power in the distribution network at time t, P G (t) represents the active output of the generator in the distribution network at time t, P HESS (t) represents the output power of the hybrid energy storage system in the distribution network at time t.

[0426] The charging and discharging power of supercapacitors and batteries must meet the constraints

[0427]

[0428] Among them, P' SC (t) and P' BESS (t) is the charge and discharge power of the supercapacitor and battery at time t, P SC,rated and P B,rated is the rated power of the supercapacitor and battery.

[0429] The capacity of supercapacitors and batteries at any time must satisfy the constraints

[0430]

[0431] Among them, E SC (t) and E B (t) is the capacity of the supercapacitor and battery at time t, E SC,rated and E B,rated is the rated capacity of the supercapacitor and battery.

[0432] The SOC of the supercapacitor and the battery at any time should satisfy the constraint

[0433]

[0434] Among them, SOC SC (t) and SOC B (t) is the state of charge of the supercapacitor and battery at time t, SOC SC,max and SOC SC,min is the upper and lower limits of the SOC of the supercapacitor, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery's SOC.

[0435] The initial capacity and terminal capacity of hybrid storage during the dispatch period need to be consistent, that is,

[0436] E SC (0) = E SC (T),E B (0) = E B (T) (56)

[0437] Among them, E SC (0) and E B (0) represents the capacity of the energy storage system at the initial moment, E SC (T) and E B (T) represents the energy storage system capacity at the termination time.

[0438] After installing hybrid energy storage, the direct grid-connected wind power must meet the constraints

[0439]

[0440] Among them, F Δt is the wind power fluctuation rate within Δt, and is the maximum and minimum wind power in the Δt period, P N is the rated output power of the fan, F lim Indicates the fluctuation rate limit that the wind power grid-connected power needs to meet.

[0441] The branches in the distribution network must satisfy the constraints

[0442]

[0443] Among them, the binary variable U (m,n) (t) represents the switch state of branch (m,n) at time t, U (m,n) (t) = 0 means the branch switch is disconnected, U (m,n) (t) = 1 means the branch switch is closed; N b is the set of busbar nodes in the distribution network, L i is the number of branches of loop i in the distribution network, N Li is the set of branches in loop i in the distribution network.

Claims

1. A method for improving the reliability of a distribution network based on hybrid energy storage and dynamic network reconstruction, characterized in that it comprises the following steps: 1) Generate wind farm wind power scenario; 2) Based on the variational mode decomposition algorithm, the generated wind farm power is decomposed into high-frequency, medium-frequency and low-frequency components, and then the high-frequency and medium-frequency components are allocated to supercapacitors and batteries respectively, and the low-frequency components are used as direct grid-connected power, and the energy storage capacity and power of supercapacitors and batteries are calculated; 3) Calculate the grid reliability assessment index; 4) Establish a two-stage optimization configuration model; 5) Solve the two-stage optimization configuration model to optimize the grid reliability assessment index; The steps to generate a wind farm wind power scenario include: 1.1) The Gaussian mixture model is used to describe the probability distribution of wind speed in the wind farm f(x), that is: Among them, K represents the total number of Gaussian components, p k (x|θ k ) is given by the parameter θ k The k-th Gaussian component expression is determined by k ∈[0,1] is the weight corresponding to the kth Gaussian component, μ is the mathematical expectation of the random variable x, σ 2 is the variance of the random variable x, θ k =(μ,σ j ) represents the set of expected and variance distribution functions; x is the observed sample; 1.2) Based on the probability distribution of wind speed f(x), construct the correlation coefficient matrix C of the relevant random vector Y in the wind farm y ,Right now: Among them, ρ ij are any two variables y in the random vector Y i and j The correlation coefficient between 1.3) By using formula (5), the correlated random vector Y is converted into a correlated standard normal random vector H = (h1, h2, ..., h c ); Among them, Φ i (h i ) is the variable h i The standard normal distribution function of is the variable y i The cumulative distribution function F i The inverse function of (.); Among them, the correlation coefficient matrix C of the related standard normal random vector H h As shown below: Among them, ρ i ' j are any two variables h in the vector H i and h j The correlation coefficient between 1.4) Based on Cholesky decomposition, the correlation coefficient matrix C h Decompose into upper triangular matrix B T And the lower triangular matrix B, we get: C h =BB T (7) 1.5) Convert the correlated standard normal random vector H into an independent standard normal random vector L, that is: L=B -1 H (8) 1.6) Using the Nataf inverse transform, the standard normal random vector L is converted into a sample of related wind speed random variables to generate a wind farm wind power scenario; in, The expression of wind power in a wind farm is as follows: Among them, v is the real-time wind speed of the wind farm, v in and v out are the cut-in and cut-out wind speeds of the wind turbines in the wind farm, v N is the rated wind speed of the wind turbine in the wind farm, P N is the rated power of the wind turbine in the wind farm; P W is the output power of the wind farm; In the Gaussian mixture model, the parameter ω k and parameter θ k Solved by expectation maximization algorithm; Solving for parameter ω k and parameter θ k The steps include: a1) Initialization parameter ω k and θ k , calculate the jth observation sample x j The corresponding k hidden variables γ jk ,Right now: a2) Based on latent variable γ jk , update the parameter ω k and θ k ,get: Where N is the number of observed samples; μ k , are the mathematical expectation and variance; a3) Repeat steps 1) to 2) until the parameter ω k and parameter θ k convergence; Correlation coefficient ρ of correlated random variables Y ij The correlation coefficient ρ′ with the related standard normal random variable H ij The following relationship is satisfied: in, is the random variable y i and j The joint probability density function, φ2(h i ,h j ,ρ′ ij ) is the random variable h i and h j The joint probability density function of i , σ j is the random variable y i and j The corresponding standard deviation; μ i , μ j is the random variable y i and j The corresponding mathematical expectation; ρ' ij is the relationship coefficient; The steps of decomposing the generated wind farm power into high-frequency, medium-frequency and low-frequency components based on the variational mode decomposition algorithm include: 2.1) Decompose the wind farm wind power into K modal functions with limited bandwidth, and obtain: Among them, u k (t) is the wind farm output power P W (t) the kth modal component after decomposition, represents the gradient operation, ‖·‖2 represents L 2 -norm; ω k is the weight corresponding to the kth Gaussian component; t is the time; P W (t) is the wind farm output power at time t; δ(t) is the pulse function; 2.2) For K mode functions with limited bandwidth, the components with frequencies less than ω2 are regarded as low-frequency components, and the frequency range is [ω2, ω th ) as the intermediate frequency component, and the frequency range is [ω th ,ω K ) as the high-frequency component; 2.3) The high frequency and medium frequency are respectively distributed to the supercapacitor and the battery, and the low frequency component is used as the direct grid-connected power, that is: P con (t)=u1(t) (16) P BESS (t)=u2(t)+u3(t)+...+u th (t) (17) P SC (t)=u th+1 (t)+u th+2 (t)...+u K (t) (18) Among them, P con (t) is the grid-connected power at time t, P BESS (t) and P SC (t) are the power leveling task reference values ​​of the battery and supercapacitor, u th (t) represents the wind power at frequency ω th The modal components at ; The power of supercapacitors and batteries is as follows: Among them, η BESS,c and η BESS,d Indicates the charging and discharging efficiency of the battery, η SC,c and η SC,d Indicates the charging and discharging efficiency of the supercapacitor; P B ' ESS (t) and P S ' C (t) represents the actual charging and discharging power of the battery and supercapacitor; The steps to calculate the energy storage capacity of supercapacitors and batteries include: b1) Calculate the capacity of the supercapacitor and battery at time n relative to the initial time, that is: Where Δt is the sampling period of wind power; E B (n) and E SC (n) represents the capacity of the battery energy storage system and supercapacitor at time n relative to the initial time; E B0 and E SC0 Indicates the initial capacity of the battery energy storage system and supercapacitor; b2) Construct the state of charge constraints of supercapacitors and batteries at any time, namely: In the formula, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery SOC constraint, SOC SC,max and SOC SC,min represents the upper and lower limits of the supercapacitor SOC constraint, E B,rated and E SC,rated Indicates the rated capacity of batteries and supercapacitors; The charging and discharging power during the energy storage operation process does not exceed its rated power, that is, P B,rated =max(|P' BESS (t)|) (25) P SC,rated =max(|P' SC (t)|) (26) Among them, P B,rated and P SC,rated is the rated power of the battery and supercapacitor; b3) Calculate the rated capacity of the battery and supercapacitor, that is: In the formula, E B,rated 、E SC,rated is the rated capacity of the battery and supercapacitor; The grid reliability evaluation index includes voltage stability index VSI, maximum value of grid branch dielectric constant Be, unbalanced power P of the whole grid, and un ; The voltage stability indicators are as follows: Among them, N n is a collection of grid nodes; Voltage stability index VSI at node i i As shown below: |V i | 4 -b i |V i | 2 +c i =0 (32) in, represents the Thevenin equivalent voltage at grid node i, V i represents the voltage at grid node i, and are the Thevenin equivalent resistance and reactance at the grid node i, respectively, P i and Q i represents the active and reactive power fed into by node i, Represents the node voltage V i The conjugate voltage of i represents the current at the grid node i; b i and c i is the coefficient; The maximum value of the dielectric constant of the power grid branch B e As shown below: When the unit current flows into the path (i, j) between the generator and the load node, the current generated on the branch (m, n) and B e (m,n) is as follows: I mn (g,l)=y mn (U m -U n )=y mn [(Z mg -Z ml )-(Z ng -Z nl )] (39) AT k =Z kg -WITH kl (40) Where U i is the voltage at node i, I i is the current injected into node i, Z ij is the impedance between nodes i and j; Z kg is the impedance between nodes k and g in the shortest path, Z kl is the impedance between nodes k and l in the shortest path; U k , U m , U n is the voltage of node k, node m, and node n; k is any node on the shortest path between the generator node and the load node; V G is the set of generator nodes in the power grid, V L is the set of load nodes in the power grid, W g and W l are the weights of the generator node g and the load node l respectively; I mn (g, l) represents the current generated in branch (m, n) when unit current flows from generator node g into load node l; y mn represents the admittance of branch (m,n); Z NN Represents impedance; Z mg is the impedance between nodes m and g; Z ml is the impedance between nodes m and l; Z ng is the impedance between nodes n and g; Z nl is the impedance between nodes n and l; I N Indicates current; U N Indicates voltage; The overall unbalanced power of the power grid P un As shown below: Where P un,j is the unbalanced power of the jth region; A j is the number of nodes in the jth region, P G,i (t) represents the active output of the ith generator at time t, P D,i (t) represents the load power at bus i at time t; P W,i (t) represents the output power of the i-th wind farm at time t; N A The number of regions after the distribution network is divided; The two-stage optimization configuration model includes a first-stage optimization configuration model and a second-stage optimization configuration model; Among them, the objective function minf of the first stage optimization configuration model c As shown below: Among them, r is the funding discount rate, T L is the entire life cycle of energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors respectively; C op is the daily operating cost of hybrid energy storage; C B,inv and C SC,inv are the investment costs of batteries and supercapacitors, and are the unit power costs of batteries and supercapacitors, respectively; and are the unit capacity costs of batteries and supercapacitors respectively; P B,rated and P SC,rated is the rated power of the battery and supercapacitor; E B,rated and E SC,rated Indicates the rated capacity of the battery and supercapacitor; T indicates the time period; and is the unit operating cost of supercapacitors and batteries; P' BESS (t) and P' SC (t) represents the actual charging and discharging power of the battery and supercapacitor; Δt is the sampling period; The objective function minf of the second-stage optimization configuration model v As shown below: minf v =ρ1VSI+ρ2B e +ρ3P un (47) Among them, ρ1, ρ2 and ρ3 are the weight coefficients of reliability index, and satisfy ρ3=2ρ1=2ρ2; VSI, B e , P un It is the voltage stability index, the maximum value of the dielectric constant of the power grid branch, and the unbalanced power of the entire power grid; The constraints of the two-stage optimization configuration model include power flow constraints, bus node voltage and phase constraints, generator output constraints, load power, wind power, generator power, hybrid energy storage system power and power purchase from the upper grid, supercapacitor and battery charging and discharging power constraints, supercapacitor and battery capacity constraints, supercapacitor and battery SOC constraints, hybrid energy storage during the dispatch period The initial capacity and terminal capacity balance constraints, wind power direct grid-connected power constraints, and distribution network branch constraints; The power flow constraints are as follows: Among them, P W,i (t) and Q W,i (t) is the wind power at bus i at time t, P G,i (t) and Q G,i (t) is the generator power at bus i at time t, P D,i (t) and Q D,i (t) is the load power at bus i at time t, P HESS,i (t) represents the charging and discharging power of the hybrid energy storage at bus i at time t, V i (t) and θ i (t) is the voltage amplitude and phase at bus i at time t, G ij and B ij is the conductance and susceptance between busbars i and j, N b (i) is the set of other buses connected to bus i; V j (t) and θ j (t) is the voltage amplitude and phase at bus j at time t; The voltage and phase constraints of the bus node are as follows: in, and are the upper and lower limits of the voltage amplitude at bus node i, and are the upper and lower limits of the voltage phase at bus i; The generator output constraints are as follows: in, and are the upper and lower limits of the active output of the generator at bus i, and Indicates the upper and lower limits of the reactive output of the generator at bus i; The constraints on the relationship between load power, wind power, generator power, hybrid energy storage system power and power purchased from the upper grid are as follows: P D (t)=P grid (t)+P W (t)+P G (t)+P HESS (t) (52) Among them, P D (t) represents the total load power in the distribution network at time t, P grid (t) represents the power purchased by the distribution network from the upper grid at time t, P W (t) represents the wind power in the distribution network at time t, P G (t) represents the active output of the generator in the distribution network at time t, P HESS (t) represents the output power of the hybrid energy storage system in the distribution network at time t; The charging and discharging power constraints of supercapacitors and batteries are as follows: Among them, P' SC (t) and P' BESS (t) is the charge and discharge power of the supercapacitor and battery at time t, P SC,rated and P B,rated is the rated power of the supercapacitor and battery; The supercapacitor and battery capacity constraints are as follows: Among them, E SC (t) and E B (t) is the capacity of the supercapacitor and battery at time t, E SC,rated and E B,rated is the rated capacity of the supercapacitor and battery; The supercapacitor and battery SOC constraints are as follows: Among them, SOC SC (t) and SOC B (t) is the state of charge of the supercapacitor and battery at time t, SOC SC,max and SOC SC,min is the upper and lower limits of the SOC of the supercapacitor, SOC B,max and SOC B,min Indicates the upper and lower limits of the battery's SOC; The initial capacity and terminal capacity balance constraints of hybrid energy storage during dispatch are as follows: AND SC (0)=And SC (TEA B (0)=And B (T) (56) Among them, E SC (0) and E B (0) represents the capacity of the energy storage system at the initial moment, E SC (T) and E B (T) represents the capacity of the energy storage system at the end time; The power constraints for direct grid connection of wind power are as follows: Among them, F Δt is the wind power fluctuation rate within Δt, and is the maximum and minimum wind power in the Δt period, P N is the rated output power of the fan, F lim Indicates the fluctuation rate limit that the wind power grid-connected power needs to meet; The distribution network branch constraints are as follows: Among them, the binary variable U (m,n) (t) represents the switch state of branch (m,n) at time t, U (m,n) (t) = 0 means the branch switch is disconnected, U (m,n) (t) = 1 means the branch switch is closed; N b is the set of busbar nodes in the distribution network, L i is the number of branches of loop i in the distribution network, N Li is the set of branches in loop i in the distribution network.

Images