Active distribution network expansion planning method and system considering 5G base station backup energy storage

Abstract

The present invention relates to a method and system for active distribution network expansion planning considering 5G base station backup energy storage, the method comprising: obtaining a 5G base station backup energy storage power evaluation result according to the reliability index of the distribution network power supply node and the communication load characteristics of the 5G base station; obtaining the distribution network historical data, constructing an initial data matrix, using a variational Bayesian Gaussian mixture model to perform cluster analysis on the initial data matrix, and constructing a typical daily scenario set of the distribution network; establishing an active distribution network expansion planning model considering 5G base station backup energy storage; converting the active distribution network expansion planning model into a mixed integer second-order cone programming model and solving it, and obtaining the optimal distribution network expansion planning scheme and the simulated operation strategy of the 5G base station backup energy storage. Compared with the prior art, the present invention can make full use of the backup energy storage resources of 5G base stations, realize a more accurate and reliable expansion planning of the active distribution network, and reduce the electricity cost of communication network operators while reducing the distribution network planning cost.

Description

Technical Field

[0001] The present invention relates to the field of distribution network planning, and more particularly to a distribution network expansion planning method and system that takes into account 5G base station backup energy storage. Background Art

[0002] Active distribution network refers to a distribution network with distributed power sources, energy storage devices, intelligent loads and advanced control systems, which can achieve efficient, reliable, economical and environmentally friendly operation. The expansion planning of active distribution network refers to the optimization of the construction and expansion of substations, the construction of new lines and the site selection and capacity determination of distributed power sources according to future load demand and new energy access, so as to reduce the planning and simulation operation costs of the system. 5G base station refers to a base station using the fifth generation of mobile communication technology, which has the characteristics of high speed, low latency and large connection, but also faces the challenges of high density and high energy consumption. 5G base stations are usually equipped with backup batteries to ensure their power reliability, but these batteries are idle most of the time, resulting in a waste of resources. If the backup batteries of 5G base stations can be used as a flexible energy storage resource to participate in the dispatch of active distribution networks, it can not only reduce the operating costs of 5G base stations, but also improve the new energy absorption capacity and voltage stability of active distribution networks, and achieve a win-win situation for communication operators and power grid operators. Although there are distribution network dispatching models that consider 5G base stations in the existing technology, they usually only consider traditional energy storage resources in the expansion planning of active distribution networks, and cannot fully utilize the backup energy storage resources of 5G base stations. There is a problem that the expansion planning of active distribution networks is not accurate and reliable enough. Therefore, it is necessary to propose a study on the application of 5G base station backup energy storage in the expansion planning of active distribution networks. Summary of the invention

[0003] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide an active distribution network expansion planning method and system taking into account the backup energy storage of 5G base stations.

[0004] The purpose of the present invention can be achieved by the following technical solutions:

[0005] A method for planning active distribution network expansion considering 5G base station backup energy storage includes the following steps:

[0006] According to the reliability index of the power supply node of the distribution network and the communication load characteristics of the 5G base station, the evaluation results of the backup energy storage capacity of the 5G base station are obtained;

[0007] Obtaining historical data of the distribution network, constructing an initial data matrix, performing cluster analysis on the initial data matrix using a variational Bayesian Gaussian mixture model, and constructing a typical daily scenario set of the distribution network;

[0008] Based on the evaluation results of the 5G base station backup energy storage and the typical daily scenario set, an active distribution network expansion planning model considering the 5G base station backup energy storage is established;

[0009] The active distribution network expansion planning model considering 5G base station backup energy storage is converted into a mixed integer second-order cone programming model and solved, and the optimal distribution network expansion planning scheme and the simulated operation strategy of 5G base station backup energy storage are obtained. The distribution network expansion planning scheme includes the new construction and expansion scheme of substations, the new construction scheme of lines, and the wind power / photovoltaic site selection and capacity determination scheme. The simulated operation strategy of 5G base station backup energy storage includes the 5G base station backup energy storage charging / discharging strategy considering the 5G base station communication load and the reliability of the active distribution network power supply nodes.

[0010] Furthermore, the 5G base station backup energy storage power evaluation result includes the minimum backup power and dispatchable power of the dynamically divided base station backup energy storage.

[0011] Furthermore, the distribution network historical data includes load in a certain area and output data of wind power and photovoltaic power.

[0012] Furthermore, the clustering analysis of the initial data matrix using the variational Bayesian Gaussian mixture model includes:

[0013] The historical output data set of load, wind power and photovoltaic power generation is obtained as X = [x1, x2, ..., x i ,...,x N ] T Assuming that it is a mixture of K Gaussian distributions, there exists:

[0014]

[0015] Where: θ={π1,...,π K ,μ1,...,μ K ∑1,...,∑ K} is the model parameter of Gaussian distribution; π k is the mixing coefficient of the kth Gaussian distribution, satisfying μ k and∑ k are the mean and covariance matrix of the kth Gaussian distribution respectively; is the probability density function of Gaussian distribution;

[0016] By solving the optimal variational distribution, the data points are divided into typical days with the highest probability, and K typical day scenario sets considering load, wind power and photovoltaic output are obtained.

[0017] Furthermore, the mixing coefficient π k The conjugate prior distribution of follows the Dirichlet distribution, μ kand∑ k The conjugate prior distribution of follows a normal-inverse Wiechter distribution.

[0018] Furthermore, when solving the optimal variational distribution, an approximate distribution is introduced, and the optimal variational distribution is obtained by minimizing the KL divergence between the approximate distribution and the true posterior distribution.

[0019] Furthermore, the active distribution network expansion planning model aims to minimize the annual comprehensive cost of the distribution network, and the constraints considered include equipment installation capacity constraints, active distribution network network topology constraints, 5G base station backup energy storage operation constraints, active distribution network operation safety constraints, Distflow flow constraints and active distribution network active management constraints.

[0020] Furthermore, the active distribution network expansion planning model is transformed into a mixed integer second-order cone programming model and then solved by using the second-order cone relaxation method and the Big-M method.

[0021] The present invention also provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the method as described above.

[0022] The present invention also provides an active distribution network expansion planning system taking into account 5G base station backup energy storage, comprising one or more processors, a memory and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the method described above.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1. The present invention evaluates the backup energy storage capacity of 5G base stations, and constructs a typical daily scenario set of the distribution network based on historical data. On the basis of the evaluation results of the backup energy storage capacity of 5G base stations and the typical daily scenario set, an active distribution network expansion planning model considering the backup energy storage of 5G base stations is established, which realizes the coordinated planning of new construction and expansion of substations, new construction of lines and site selection and capacity determination of distributed power sources, can make full use of the backup energy storage resources of 5G base stations, realize more accurate and reliable expansion planning of the active distribution network, and reduce the planning cost of the distribution network while reducing the electricity cost of communication network operators, thereby achieving mutual benefit and win-win results for distribution network operators and communication network operators.

[0025] 2. The present invention adopts variational Bayesian-Gaussian mixture clustering analysis to construct a typical daily scenario set of the distribution network, which can adaptively determine the number and probability of the optimal typical daily scenarios, thereby improving the accuracy and reliability of subsequent active distribution network expansion planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1Schematic diagram of the process of the present invention;

[0027] Figure 2 Flow chart of a typical day scene construction method using variational Bayesian Gaussian mixture clustering in an embodiment of the present invention;

[0028] Figure 3 Evaluation result of 5G base station standby energy storage quantity in an embodiment of the present invention;

[0029] Figure 4 The wind and solar load historical data throughout the year in the embodiment of the present invention;

[0030] Figure 5 Clustering results based on the variational Bayesian-Gaussian mixture clustering algorithm in the embodiment of the present invention;

[0031] Figure 6 The planning result of scheme I in the embodiment of the present invention;

[0032] Figure 7 The planning result of scheme II in the embodiment of the present invention;

[0033] Figure 8 In the embodiment of the present invention, the 5G base station can dispatch the total output power of the backup energy storage. DETAILED DESCRIPTION

[0034] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0035] Active distribution network (ADN) is an advanced grid model that adapts to the construction of new power systems and the needs of "dual carbon" goals. ADN can actively control and manage connected distributed generation (DG) and energy storage equipment to improve the flexibility and economy of the system. With the rapid development of information and communication technology, 5G base stations, as high-density and high-energy consumption communication nodes, their large-scale deployment will increase the equipment investment and operating costs of distribution network operators. Therefore, it is of great significance to consider 5G base stations in ADN planning and operation research.

[0036] This embodiment provides a method for planning active distribution network expansion considering 5G base station backup energy storage, such as Figure 1 As shown, including:

[0037] Step S1, establish a 5G base station standby energy storage power evaluation model, dynamically divide its minimum standby power and dispatchable power according to the reliability of the power supply node of the distribution network and the communication load characteristics of the 5G base station, and obtain the 5G base station standby energy storage power evaluation result;

[0038] Step S2, obtaining historical data of wind speed, light intensity and conventional load in the distribution network to form an initial data matrix, and using a variational Bayesian Gaussian mixture model to construct a typical daily scene set;

[0039] Step S3, taking the minimization of the annual comprehensive cost of the distribution network as the objective function, considering the 5G base station backup energy storage operation constraints, 5G base station operation reliability constraints, distribution network operation safety constraints, network topology constraints and active management constraints, etc., on the basis of the 5G base station backup energy storage power evaluation results and the typical daily scenario set, establish an active distribution network expansion planning model considering the 5G base station backup energy storage;

[0040] Step S4, converting the active distribution network expansion planning model considering 5G base station backup energy storage into a mixed integer second-order cone programming model and solving it, and obtaining the optimal distribution network expansion planning scheme and the simulated operation strategy of 5G base station backup energy storage, wherein the distribution network expansion planning scheme includes the new construction and expansion scheme of substations, the new construction scheme of lines, and the site selection and capacity determination scheme of wind power (windturbine generation, WTG) and photovoltaic (photovoltaic generation, PVG), and the simulated operation strategy of 5G base station backup energy storage includes the 5G base station backup energy storage charging / discharging strategy considering the 5G base station communication load and the reliability of the active distribution network power supply node.

[0041] The above method uses the redundant capacity of 5G base station backup energy storage as a flexible energy storage resource to participate in the active distribution network expansion planning, realizing the coordinated optimization of substation construction and expansion, line construction and distributed power source site selection and capacity determination. This method is suitable for the active distribution network expansion planning of 110kV and below, especially in the case of large-scale deployment of 5G base stations, where it is necessary to coordinate the construction and expansion of substations, the construction of lines and the site selection and capacity determination of distributed power sources.

[0042] 1. 5G base station backup energy storage model

[0043] The 5G base station heterogeneous network is composed of different types of base stations, such as macro base stations, micro base stations, pico base stations, home base stations, etc., which can work together in the same frequency band or different frequency bands to form a flexible network topology. The present invention mainly analyzes the 5G macro base station with the largest coverage and power consumption. According to the test data analysis of 5G base station power consumption, under the maximum communication load, the 5G base station signal transmission unit accounts for about 90% of the total energy consumption; under no-load conditions, it only accounts for 15%. In comparison, the power consumption of other electrical equipment in the base station is relatively stable and does not fluctuate significantly with changes in the communication load. Therefore, the power consumption of 5G base stations can be divided into two parts: one part is the baseline power consumption of the base station, which is a fixed value and has nothing to do with the communication load; the other part is the power consumption of the signal transmission unit of the base station, which is proportional to the communication load:

[0044]

[0045] Where: is the communication power consumption of the 5G base station in period t; is the benchmark power consumption of 5G base stations; is the communication load rate of the 5G base station in period t; is the maximum transmit power consumption of the 5G base station; is the power consumption of the power amplifier in the active antenna unit (AAU) at time t; is the power consumption of the RF unit in the AAU at time t; The power consumption of 5G base station transmission equipment; is the power consumption of the baseband unit (BBU); It is the benchmark power consumption of 5G base stations; is the communication load rate of the 5G base station at time t, ranging from 0 to 1; is the dynamic power consumption at maximum communication load.

[0046] In order to ensure that the 5G base station backup energy storage battery meets the normal backup power demand and maximizes the use of its idle capacity. This embodiment takes into account the reliability of the ADN power supply node and the reliability of the 5G base station, and dynamically divides the backup energy storage battery into minimum backup capacity and dispatchable capacity. Among them, the minimum backup capacity is used to respond to emergencies and short-term city power failures to ensure continuous power supply to the base station; and the dispatchable capacity refers to the capacity of the backup energy storage battery that can participate in ADN optimization scheduling during periods of low communication load or in an idle state.

[0047] The reliability index of 5G base stations refers to the probability of normal operation of base station communication equipment. According to the regulations of the Next Generation Mobile Network Alliance, the reliability index of 5G base stations is The reliability index of 5G base stations involves two scenarios: one is that the base station load node is powered by the distribution network when it is operating normally, and the failure rate of the load node needs to be evaluated; the other is that when the base station load node fails and the power is cut off, it is necessary to evaluate whether the reserved capacity of the base station backup energy storage is sufficient to support the restoration of power supply. for:

[0048]

[0049] Where: i is the failure rate of node i; P s,i The probability that the backup energy storage battery power supply time for node i is greater than the fault power outage time.

[0050] The present invention considers the repairable components such as feeders, transformers, disconnectors, and circuit breakers in the ADN, and calculates the failure rate of node k in the ADN as follows:

[0051]

[0052] Where: m is the number of minimum paths; n is the number of network elements; λ i is the failure rate of the i-th minimum path; j is the failure rate of the jth component; P ij ∈{0,1}, indicating whether the i-th minimum path contains the j-th element, 1 for inclusion and 0 for non-inclusion; B i is the probability of the distributed generation on the i-th minimum path forming an island.

[0053] According to the historical data of power outage time in the distribution network, the Weibull distribution is used to approximate the power outage time probability distribution curve Q. s (t), according to the probability distribution of power outage time, we can get the probability P that the base station backup energy storage power supply time is greater than the fault power outage time s As follows:

[0054]

[0055] From equations (2)-(4), we can get the minimum backup time of 5G base station at load node i: The calculation formula is as follows:

[0056]

[0057] Based on the 5G base station power consumption model, the telecommunications network operator (TNO) can use the predicted data of the base station communication load rate to calculate the power consumption of the 5G base station at time t, combined with the minimum backup time The minimum spare capacity of 5G base station n at time t can be obtained and dispatchable capacity As shown in formula (6) and formula (7) respectively:

[0058]

[0059]

[0060] Where: is the minimum reserve power of 5G base station n in period t; Minimum backup time for base station energy storage; is the dispatchable power of 5G base station n in period t; The rated charge / discharge energy of the backup energy storage battery. is the power consumption of 5G base station n at time τ.

[0061] 2. Typical day scene construction method based on variational Bayesian Gaussian mixture model

[0062] In power grid planning, due to the uncertainty of load, wind power and photovoltaic output, typical day scenario construction is an effective processing method. Typical day scenario construction refers to selecting a certain number of representative daily output curves from the load, WTG and PVG output data throughout the year as the input of the planning model. The traditional k-means clustering algorithm is a commonly used typical day scenario construction method, but it requires manual specification of the number of typical day scenarios. If the number of typical days is too large, it will increase the difficulty of solving the planning model; if the number of typical days is too small, it will reduce the accuracy and credibility of the planning model. Variational Bayesian Gaussian mixture clustering is an unsupervised clustering algorithm based on a probability model. The parameters of the model are estimated by the variational Bayesian expectation maximization (VBEM) algorithm, which can adaptively determine the number and probability of the optimal typical day scenarios. The calculation process is as follows Figure 2 As shown, the calculation steps and formulas are as follows:

[0063] (1) Assume that the output data set of load, WTG and PVG in a certain region is X = [x1, x2, ..., x i ,...,x N ] T It is a mixture of K Gaussian distributions. are data points, representing the 24h output scenarios of load, WTG and PVG. Assume that each output scenario is represented by the mean and covariance matrix of Gaussian distribution, and N represents the number of days of data.

[0064]

[0065] Where: θ={π1,...,π K ,μ1,...,μ K ∑1,...,∑ K} is the model parameter of Gaussian distribution; π k is the mixing coefficient of the kth Gaussian distribution, satisfying μ k and∑ k are the mean and covariance matrix of the kth Gaussian distribution respectively; is the probability density function of the Gaussian distribution.

[0066] (2) Introduce hidden variables Z = {z 1k ,z 2k ,...,z NK} represents the attribution of the daily force data set of load, WTG and PVG to the Gaussian mixture distribution. nk ∈{0,1}, satisfying z nk =1 means that the nth data belongs to the kth Gaussian distribution, z nk =0 The nth data does not belong to the kth Gaussian distribution. Assuming that the data points are independent of each other, we have:

[0067]

[0068]

[0069]

[0070] Where: is the attribution of the mixing coefficient of the kth Gaussian distribution in the latent variable. If z nk =1, then If z nk =0, then

[0071] (3) By selecting an appropriate conjugate prior distribution and iteratively updating the parameters, the calculation process of Bayesian inference can be simplified, and the form of the posterior distribution can be kept the same as the prior distribution without the need for complex integral calculations. In this embodiment, assuming that the mixing coefficient π k The conjugate prior distribution of follows the Dirichlet distribution, μ k and∑ k The conjugate prior distribution of follows the Normal-Inverse-Wishart distribution, that is:

[0072]

[0073]

[0074] Where: α={α1,...,α K} represents the hyperparameters of Dirichlet distribution; β0, m0, W0, v0 are the hyperparameters of normal-inverse Vicht distribution.

[0075] Then the joint probability distribution of the model is:

[0076] p(X,Z,θ)=p(X|Z,μ,Σ)p(Z|π)p(π)p(μ|Σ)p(Σ) (14)

[0077] (4) The clustering goal is to solve the posterior probability distribution p(Z,θ|X), but this distribution is difficult to calculate, so we use the variational inference method to introduce an approximate variational distribution q(Z,θ) to replace the true posterior distribution:

[0078]

[0079] (5) In order to solve the optimal variational distribution, the coordinate ascent method is used to alternately update q(Z) and q(θ):

[0080]

[0081]

[0082] Where: E q(θ) [·] indicates the expectation of q(θ); const is a constant term independent of the variable.

[0083] In the optimization process of the VBGMM (Variational Bayesian Gaussian Mixture Model) clustering algorithm, the main idea of ​​using the coordinate ascent method is to gradually improve the value of the objective function by alternately optimizing each variable until convergence. Constant terms refer to those terms that do not change with the optimized variables. These constant terms ensure the integrity and accuracy of the mathematical expression during the derivation process, but can be ignored in the actual optimization process because they do not affect the optimization results. The value of the constant term depends on the parameters of the probability distribution involved in the model and the normalization constant, which ensures the convergence and stability of the algorithm.

[0084]

[0085]

[0086]

[0087] Where: r nk is the probability that the nth data point belongs to the kth Gaussian distribution.

[0088] (6) Repeat step (5) to minimize the KL (Kullback-Leibler) divergence between the approximate distribution and the true posterior distribution until the variational lower bound L(q) is less than the set threshold ξ:

[0089] L(q)=|E q(Z,θ) [lnp(X,Z,θ)]-E q(Z,θ) [lnq(Z)q(θ)]| (21)

[0090] (7) According to the output after the iteration stops The optimal parameter r nk , divide the data points into the typical days with the highest probability, that is:

[0091]

[0092] Finally, we get K typical daily scenario sets considering load, WTG and PVG output, π k* Represents the probability of a typical day scenario:

[0093]

[0094] 3. Active Distribution Network (ADN) Expansion Planning Model Considering 5G Base Station Backup Energy Storage

[0095] Based on the evaluation results of 5G base station backup energy storage and the typical daily scenario constructed by the variational Bayesian Gaussian mixture model clustering method, an ADN expansion planning model considering 5G base station backup energy storage is established.

[0096] ADN annual comprehensive cost is divided into annual investment cost C Inv and annual simulated operating cost C Ope Among them, C Inv Including the substation investment cost, line investment cost, WTG investment cost and PVG investment cost converted to annual cost; C Ope Including equipment operation and maintenance costs C Om , Electricity purchase cost C Pur , Penalty fees for wind and solar power abandonment C Cur and base station energy storage dispatching cost C BS The decision variables in the model include the construction and expansion of substations, the construction of new lines, the site selection and capacity determination of WTGs and PVGs, the gear adjustment of the on-line tap changer (OLTC), and the charging / discharging power of the dispatchable power of the 5G base station backup energy storage.

[0097] Therefore, the objective function of the ADN expansion planning model is expressed as:

[0098] minF=C Inv +COpe (twenty four)

[0099]

[0100] C Ope =C Om +C Pur +C Cur +C BS (26)

[0101]

[0102]

[0103]

[0104]

[0105] Where: represents the coefficient of converting the present value of equipment a to its equivalent annual value; d is the discount rate; γ a is the depreciation period of equipment a; Ω D and Ω T represents a typical daily scene and a set of simulation operation time periods; s and t represent the current typical daily scene and time period; Δt is the simulation step length, which is 1h in this embodiment; D s is the probability of the sth typical day scene; Φ SC , Φ SE , Φ WTG and Φ PVG They represent the set of substations to be newly built, the set of substations to be expanded, and the set of nodes to be selected for WTG and PVG respectively; Ω L is a line branch set; x ij , and are 0-1 decision variables for new substation construction, substation expansion, new line construction, WTG and PVG construction, respectively. and Indicates whether to invest in equipment a; c Line , and They represent the investment cost of new substation, investment cost of expanded substation, investment cost of new line per unit length, and investment cost per unit capacity of WTG and PVG respectively; and represents the installed capacity of WTG and PVG at node i; L ij is the length of line ij; f WTG and f PVGIt represents the operation and maintenance cost per unit of electricity generated by WTG and PVG; and Indicates the active output of WTG and PVG; and Indicates the output upper limit of WTG and PVG; It is the annual operation and maintenance cost of the substation; is the electricity purchase price during period t; is the power purchased from the upper grid; c W,Cur and c P,Cur Indicates the penalty cost for the unit's abandonment of wind and solar power; Ω BS is the access node set of 5G base station; f BS Subsidies for distribution network operators to provide 5G base station energy storage for dispatching; and Represents the active power of charging and discharging the 5G base station energy storage.

[0106] The constraints considered in the ADN expansion planning model include equipment installation capacity constraints, ADN network topology constraints, 5G base station backup energy storage operation constraints, ADN operation safety constraints, Distflow constraints, and ADN active management constraints.

[0107] (1) Equipment installation capacity constraints

[0108] 1) Capacity constraints for new and expanded substations

[0109]

[0110] Where: is the rated capacity of the new substation to be built at the node; and They represent the rated capacity of the transformer of the existing substation at node i and the rated capacity of the transformer to be expanded respectively.

[0111] 2) DG installation capacity constraints

[0112]

[0113] Where: and They represent the number of WTG and PVG installed at node i, respectively; and They represent the unit installed capacity of WTG and PVG respectively; and They represent the upper limits of the installed capacity of WTG and PVG at node i respectively.

[0114] (2) ADN network topology constraints

[0115] 1) Radial constraints:

[0116] When the distribution network is operating normally, the network structure is a tree structure without loops, that is, there is a unique path from the root node to any load point. The tree structure constraint can be expressed by determining the number of loops in the network.

[0117]

[0118] Where: Indicates an existing line; Ω N is a node set; |·| represents the number of elements in the set ·.

[0119] 2) Connectivity constraints:

[0120] Connectivity constraints ensure that there is at least one power supply path between any two nodes in the distribution network, avoiding isolated nodes or disconnected paths and ensuring that DG nodes are not isolated.

[0121]

[0122]

[0123]

[0124]

[0125] Where: y ij is the virtual power flow from node i to node j, y ij is an integer variable, y ij =-y ji ; The virtual load of the load node is 1; N(i) is the set of nodes connected to node i.

[0126] (3) 5G base station backup energy storage operation constraints

[0127] 1) Base station energy storage charging / discharging energy limit constraints

[0128] In a charging / discharging cycle, the amount of energy stored in the base station at each time period should be kept within the assessed dispatchable amount:

[0129]

[0130] Where: is the power consumption of 5G base station n during period t in a typical day s.

[0131] 2) Base station energy storage charging / discharging energy continuity constraints

[0132] The charging / discharging energy of the base station energy storage follows strict time series continuity. It is accumulated and calculated in chronological order according to the size of the charging / discharging power, as shown in the following formula:

[0133]

[0134] Where: η c and η d Respectively represent the charging / discharging efficiency of the base station energy storage, and They respectively represent the charging / discharging power of the base station energy storage n during period t in a typical day s.

[0135]

[0136] Where: and They represent the power of the energy storage of base station n at time 0 and time 24 in a typical day s respectively; is the initial charge / discharge energy of base station n in a typical day s.

[0137] 3) Base station energy storage charging / discharging power constraints

[0138]

[0139] Where: and They represent the upper limits of the energy storage charging / discharging power of base station n respectively; Indicates the charging / discharging status of the base station energy storage, 1 for charging and 0 for discharging.

[0140] (4) ADN operation safety constraints

[0141] 1) Node voltage constraints

[0142]

[0143] Where: U i,s,t represents the voltage amplitude at node i; and They represent the upper and lower limits of the voltage amplitude at node i respectively.

[0144] 2) Branch current constraints

[0145]

[0146] Where: I ij,s,t represents the current amplitude of the branch (ij); It indicates the upper limit of the current amplitude allowed to flow through the branch (ij).

[0147] (5) Distflow Constraints

[0148]

[0149]

[0150]

[0151]

[0152]

[0153] Where: P jk,s,t and Q jk,s,t Respectively represent the active and reactive power at the head end of branch (jk); P ij,s,t and Q ij,s,t Respectively represent the active and reactive power of the first section of the branch (ij); R ij and X ij Respectively represent the resistance and reactance of the branch (ij); U i,s,t and U j,s,t Represent the voltage amplitudes of nodes i and j respectively; M is a sufficiently large positive integer.

[0154] P j,s,t and Q j,s,t The expressions of are shown in equations (49) and (50) respectively. If the corresponding device is not installed at node i, the corresponding item is taken as 0.

[0155]

[0156]

[0157] Where: is the active power consumption of 5G base station n at node j; and They represent the active and reactive power of the load at node j respectively.

[0158] (6) ADN Active Management Constraints

[0159] 1) OLTC tap position adjustment constraints

[0160]

[0161] Where: It is the lower limit of the square of the substation voltage amplitude; 0~2K represents the lowest to highest OLTC tap position; is a 0-1 decision variable, Indicates that the OLTC tap position is k during period t t , Indicates that the OLTC tap position is not k t .

[0162] 2) DG power regulation constraints

[0163]

[0164] Where: and They represent the power factor angles of WTG and PVG installed at node i respectively; the adjustable range is from lagging phase 0.95 to leading phase 0.95.

[0165] 4. Model Conversion

[0166] Since the above model is a mixed integer non-convex model, it cannot be solved directly using a commercial solver. To improve the solving efficiency, this embodiment uses the Big-M method and second-order cone relaxation to make the model convex.

[0167] (1) Conversion of base station energy storage charging / discharging power constraints

[0168] Using the Big-M method, a sufficiently large number M is introduced to transform the base station energy storage charging / discharging power constraint into the following form:

[0169]

[0170] (2) Second-order cone relaxation of Distflow flow constraints

[0171] Introducing auxiliary variables and The non-convex quadratic constraints in the Distflow power flow constraints and Performing second-order cone relaxation, we obtain the following form:

[0172]

[0173]

[0174]

[0175]

[0176]

[0177] The node voltage constraint and branch current constraint are equivalent to:

[0178]

[0179] Finally, the original model is transformed into a MISOCP problem, which can be solved using a commercial solver.

[0180] V. Implementation Cases

[0181] In order to verify the effectiveness of the above active distribution network expansion planning method considering 5G base station backup energy storage, the following example scenario is set:

[0182] 1. The relevant planning parameters are set as follows:

[0183] (1) Line data:

[0184] The investment cost per unit length of the line is RMB 350,000 / km; the depreciation period is 20 years and the discount rate is 6%.

[0185] (2) DG data:

[0186] The nodes to be installed for WTG are 9, 18, 20, 39, 44 and 47, with an investment cost of 6,000 yuan / kW per unit capacity, an operation and maintenance cost of 0.03 yuan / (kW·h) per unit electricity, a penalty cost for wind abandonment of 0.56 yuan / (kW·h), a depreciation period of 20 years, and a discount rate of 7%; the nodes to be installed for PVG are 6, 13, 16, 24, 33 and 49, with an investment cost of 4,000 yuan / kW per unit capacity, an operation and maintenance cost of 0.02 yuan / (kW·h) per unit electricity, a penalty cost for solar abandonment of 0.4 yuan / (kW·h), a depreciation period of 20 years, and a discount rate of 10%.

[0187] (3) Substation data:

[0188] The relevant parameters of the substation to be newly built and expanded are shown in Table 1; the depreciation period is 20 years, and the discount rate is 7%; the electricity purchase price refers to the industrial and commercial time-of-use electricity price; the number of OLTC upper and lower adjustment gears is 7, and the adjustment range is 0.97pu~1.03pu.

[0189] Table 1 Substation parameters

[0190]

[0191] (4) 5G base station data:

[0192] Table 2 5G base station equipment parameters

[0193]

[0194] The relevant parameters of 5G base stations are detailed in Table 2. The present invention assumes that all 5G base station equipment parameters are consistent; the minimum backup time of the base station energy storage battery is uniformly set to 3h; the 5G base station deployment density is set to 50 / km 2 ; The area of ​​the 54-node case is about 20km 2, assuming that the number of 5G base stations participating in ADN planning optimization in the case area is 280; the proportion of 5G base station access nodes in four typical communication load characteristics of residential areas, office areas, commercial areas and university areas is 43%, 37%, 15% and 5% respectively, and the distribution network operator pays a dispatch subsidy of 0.25 yuan / kW·h to the energy storage electricity actually participating in the dispatch.

[0195] 2. Evaluation results of 5G base station backup energy storage capacity:

[0196] According to the communication load characteristics of four typical 5G base stations in residential areas, office areas, commercial areas and university areas, the evaluation results of the backup energy storage capacity of 5G base stations are obtained as follows: Figure 3 As shown. As can be seen from the figure, during the peak communication load period, in order to ensure the communication reliability of the 5G base station, the energy storage battery has a higher reserve power and a lower dispatchable power; while during the low communication load period, the reserve power is lower and the dispatchable power is higher. It is worth noting that the peak communication load periods of 5G base stations in different typical areas are also different. The peak communication load period of base stations in office areas and university areas is usually during the day, while the peak communication load period of base stations in residential and commercial areas is usually at night. This makes the dispatchable power of backup energy storage of different 5G base stations complementary in time and space.

[0197] 3. Use the variational Bayesian-Gaussian mixture clustering algorithm to construct typical day scene results:

[0198] Based on the historical data of 8760 hours of wind and solar load in a distribution network in a certain area in South China, Figure 4 As shown, the composition X 365×72 The variational Bayesian Gaussian mixture clustering algorithm is used to construct four typical daily scenarios such as Figure 5 As shown in Table 3, the number and probability of original data represented by each typical day scene are shown in Table 3.

[0199] Table 3 Number and probability of typical day scene representatives

[0200]

[0201] In order to verify the superiority of the typical day scene set constructed by the variational Bayesian Gaussian mixture clustering algorithm in the present invention, the indicators are compared with the typical day scene set constructed by the widely used k-means algorithm. Three clustering effectiveness indicators are calculated for the typical day scene sets constructed by the two clustering methods, and the specific results are shown in Table 2. The silhouette coefficient reflects the similarity of the same typical day, the closer to 1, the better; the DB index reflects the difference between different typical days, the smaller the better; the CH index reflects the clustering effect of the typical day, the larger the better. It can be seen from Table 2 that the typical day scene constructed by the variational Bayesian Gaussian mixture clustering algorithm is superior to the k-means clustering algorithm in all three clustering effectiveness indicators. In addition, the variational Bayesian Gaussian mixture clustering algorithm can adaptively determine the optimal number and probability of typical day scenes, while the k-means clustering algorithm requires manual setting of the number of typical day scenes. Therefore, the use of the variational Bayesian Gaussian mixture clustering algorithm to construct typical day scenes can more effectively handle the complexity of massive data and provide more accurate and reliable results for power grid planning.

[0202] Table 4 Comparison of clustering effectiveness indicators

[0203]

[0204] 4. Planning results and analysis

[0205] In order to verify the effectiveness of the ADN expansion planning method proposed in this invention, two planning schemes are set:

[0206] Solution I: ADN expansion planning without considering 5G base station backup energy storage;

[0207] Solution II: Consider ADN expansion planning for 5G base station backup energy storage.

[0208] After simulation calculation, the various costs of the two planning schemes are shown in Table 5. The planning results of Scheme I and Scheme II are shown in Figure 6 and Figure 7 The overall planning results of each scheme are shown in Table 6. As can be seen from Table 5, after considering the participation of 5G base station backup energy storage in regulation, Scheme II is more economical than Scheme I as a whole. The annual comprehensive cost of ADN in Scheme II is 102.724 million yuan, which is 1.515 million yuan less than that of Scheme I, a decrease of 1.47%. At the same time, the annual regulation income of communication network operators through participating in ADN collaborative optimization is 2.059 million yuan, achieving mutual benefit and win-win results for distribution network operators and communication network operators.

[0209] Comparing the substation planning results of the two schemes, Scheme I expanded the capacity of substations S1 and S2, and built new substations S3 and S4. After considering the dispatchable energy storage of 5G base stations to participate in regulation, Scheme II can meet the load demand in the power supply area, and the construction capacity is reduced by 16.7MV·A, saving an annual investment cost of 770,000 yuan. In terms of grid planning, the new locations of 5 lines in Scheme I and Scheme II are different, and the investment cost is reduced by 148,000 yuan compared with Scheme I. In terms of DG planning, the installed capacity of WTG in Scheme I is 3.6MW, the installed capacity of PVG is 13.5MW, and the annual investment cost of DG is 8.08 million yuan. In Scheme II, the installed capacity of WTG is 5.9MW, the installed capacity of PVG is 17.2MW, and the annual investment cost of DG is 10.93 million yuan. The installed capacity of DG in Scheme II is 6MW more than that in Scheme I, indicating that the installed capacity of DG can be increased after the dispatchable energy storage of 5G base stations participates in regulation.

[0210] Table 5 ADN expansion planning costs for Scheme I and Scheme II

[0211]

[0212]

[0213] Table 6 ADN expansion planning results under different planning schemes

[0214]

[0215] Table 7 Comparison of ADN partition results

[0216]

[0217] Table 7 compares the results of load partitioning and average load rate of substations after planning. As can be seen from Table 7, in Scheme I, the newly built substations S3 and S4 only bear 8 load nodes, with load rates of 26.47% and 24.17%, respectively, and the resource utilization rate is low. However, substation S1 bears 19 load nodes, with a load rate of 68.63%, which is prone to overload at this time, endangering the safe operation of the power grid. The maximum difference in the load rate of the substations in Scheme I is 44.46%, and the power flow distribution is extremely unbalanced. In Scheme II, the newly built substations S3 and S4 access more load nodes, while reducing the load nodes accessed by S1 and S2. The maximum difference in the load rate of the substations is only 19.31%, making the ADN partitioning more reasonable. On the one hand, the dispatchable energy storage of 5G base stations participates in regional load peak shaving and valley filling to make the power flow distribution more balanced; on the other hand, Scheme II installs more capacity DG, reduces the load demand in some areas, and improves resource utilization.

[0218] In the simulation operation stage, the DG output of each typical day scenario in Scheme II was fully absorbed, and the annual electricity purchase cost from the superior power grid was reduced by 6.05 million yuan compared with Scheme I. In Scheme I, wind and solar power were abandoned in the typical day scenario with large DG output and small load demand, and the corresponding penalty fee for abandoning wind and solar power was 109,000 yuan. Using the dispatchable power of 5G base station energy storage to optimize the system flow can reduce the network loss on the line and promote the absorption of new energy.

[0219] Figure 8 It is the total charge / discharge power of the dispatchable power of 5G base station backup energy storage under four typical daily scenarios. As can be seen from the figure, the 5G base station backup energy storage in different regions adopts the charge / discharge strategy of "charging during valley hours and discharging during peak hours" under four typical daily scenarios. During the low-valley electricity price period at night from 00:00 to 08:00, the load demand is small, while the WTG output is large. Using this time period to charge the backup energy storage can promote the consumption of wind power. During the daytime peak electricity price periods from 08:00 to 12:00 and 14:00 to 17:00, the load demand is large. Discharging the dispatchable power of the base station energy storage can alleviate the peak electricity demand. During the midday basic electricity price period from 12:00 to 14:00, the load demand is reduced, while the PVG output reaches the maximum value. Charging the base station backup energy storage system can promote the consumption of photovoltaic power generation. During the basic electricity price period of 17:00-24:00, the communication load is also at its peak. The backup energy storage needs to give priority to ensuring the continuous power supply of 5G base stations. Therefore, the dispatchable power is reduced and the discharge power is also reduced. By analyzing the charging / discharging strategy of 5G base station backup energy storage, communication network operators participating in ADN collaborative optimization can maximize the utilization rate and economic benefits of base station backup energy storage while ensuring the reliability of 5G base stations, which will help future communication network operators accelerate the construction of a new digital infrastructure system.

[0220] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for planning the expansion of an active distribution network considering 5G base station backup energy storage, characterized in that: The following steps are involved: According to the reliability index of the power supply node of the distribution network and the communication load characteristics of the 5G base station, the evaluation results of the backup energy storage capacity of the 5G base station are obtained; Obtaining historical data of the distribution network, constructing an initial data matrix, performing cluster analysis on the initial data matrix using a variational Bayesian Gaussian mixture model, and constructing a typical daily scenario set of the distribution network; Based on the evaluation results of the 5G base station backup energy storage and the typical daily scenario set, an active distribution network expansion planning model considering the 5G base station backup energy storage is established; The active distribution network expansion planning model considering 5G base station backup energy storage is converted into a mixed integer second-order cone programming model and solved, and the optimal distribution network expansion planning scheme and the simulated operation strategy of 5G base station backup energy storage are obtained. The distribution network expansion planning scheme includes the new construction and expansion scheme of substations, the new construction scheme of lines, and the wind power / photovoltaic site selection and capacity determination scheme. The simulated operation strategy of 5G base station backup energy storage includes the 5G base station backup energy storage charging / discharging strategy considering the 5G base station communication load and the reliability of the active distribution network power supply nodes; The cluster analysis of the initial data matrix using the variational Bayesian Gaussian mixture model includes: The historical output data set of load, wind power and photovoltaic power generation is obtained as X = [x1, x2, ..., x i ,...,x N ] T Assuming that it is a mixture of K Gaussian distributions, there exists: Where: θ={π1,...,π K ,μ1,...,μ K ∑1,...,∑ K } is the model parameter of Gaussian distribution; π k is the mixing coefficient of the kth Gaussian distribution, satisfying μ k and∑ k are the mean and covariance matrix of the kth Gaussian distribution respectively; is the probability density function of Gaussian distribution; By solving the optimal variational distribution, the data points are divided into typical days with the highest probability, and K typical day scenario sets considering load, wind power and photovoltaic output are obtained.

2. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: The 5G base station backup energy storage power evaluation result includes the minimum backup power and dispatchable power of the base station backup energy storage that is dynamically divided.

3. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: The distribution network historical data includes the load in a certain area and the output data of wind power and photovoltaic power.

4. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: The mixing coefficient π k The conjugate prior distribution of follows the Dirichlet distribution, μ k and∑ k The conjugate prior distribution of follows a normal-inverse Wiechter distribution.

5. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: When solving the optimal variational distribution, an approximate distribution is introduced, and the optimal variational distribution is obtained by minimizing the KL divergence between the approximate distribution and the true posterior distribution.

6. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: The active distribution network expansion planning model aims to minimize the annual comprehensive cost of the distribution network. The constraints considered include equipment installation capacity constraints, active distribution network network topology constraints, 5G base station backup energy storage operation constraints, active distribution network operation safety constraints, Distflow flow constraints and active distribution network active management constraints.

7. The active distribution network expansion planning method considering 5G base station backup energy storage according to claim 1 is characterized in that: The active distribution network expansion planning model is transformed into a mixed integer second-order cone programming model and then solved by using the second-order cone relaxation method and the Big-M method.

8. A computer-readable storage medium, characterized in that: The method comprises one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs comprise instructions for executing the method according to any one of claims 1 to 7.

9. An active distribution network expansion planning system considering 5G base station backup energy storage, characterized in that: The system comprises one or more processors, a memory and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the method according to any one of claims 1 to 7.

Images