WO2019056753A1 - 一种分布式光伏电站集群的动态等值建模方法 - Google Patents

一种分布式光伏电站集群的动态等值建模方法 Download PDF

Info

Publication number
WO2019056753A1
WO2019056753A1 PCT/CN2018/084941 CN2018084941W WO2019056753A1 WO 2019056753 A1 WO2019056753 A1 WO 2019056753A1 CN 2018084941 W CN2018084941 W CN 2018084941W WO 2019056753 A1 WO2019056753 A1 WO 2019056753A1
Authority
WO
WIPO (PCT)
Prior art keywords
photovoltaic power
time series
cluster
dynamic
clustering
Prior art date
Application number
PCT/CN2018/084941
Other languages
English (en)
French (fr)
Inventor
顾伟
李培鑫
曹戈
Original Assignee
东南大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 东南大学 filed Critical 东南大学
Publication of WO2019056753A1 publication Critical patent/WO2019056753A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers

Definitions

  • the invention discloses a dynamic equivalent modeling method for distributed photovoltaic power plant clusters, and belongs to the technical field of distributed energy grid-connected modeling and simulation.
  • the object of the present invention is to provide a dynamic equivalence modeling method for a distributed photovoltaic power plant cluster, aiming at dynamic equivalence modeling of a photovoltaic cluster system, which simplifies the complexity and reduction of the detailed model in view of the deficiencies of the above background art.
  • the data needs and the time of simulation calculation are reduced, and the detailed modeling of each power station in the cluster increases the complexity of the simulation model and consumes a lot of time and effort in the data preparation phase and the model simulation calculation phase. technical problem.
  • Extract the clustering index of each photovoltaic power station in the cluster sample the data of the clustering indicators of each photovoltaic power station, and normalize the sampling data of the clustering indicators of each photovoltaic power station to obtain the time series of the clustering indicators of each photovoltaic power station;
  • the clustering indicators of each photovoltaic power plant include but are not limited to the output voltage of the array, the output current of the array, the DC side voltage of the inverter, and the inverter The input current, the active power output from the photovoltaic power station, and the reactive power output from the photovoltaic power station.
  • the dynamic time bending distance between the photovoltaic power plants is calculated according to the time series of the clustering indicators of the photovoltaic power plants according to the following method: the same clustering according to the two photovoltaic power plants
  • the time series of the indicator calculates the dynamic distance between each element in one time series and each element in another time series to determine the distance matrix of the same clustering index of the two photovoltaic power stations, and then according to the distance matrix of the same clustering index of the two photovoltaic power stations And determining the optimal path of each element in one time series to another time series with the minimum of cumulative dynamic distortion, and accumulating the elements of one time series nonlinearly to another time series
  • the optimal path obtains the dynamic time bending distance of similar clustering indicators between two photovoltaic power plants, and the dynamic time bending distance of all similar clustering indicators between two photovoltaic power plants is obtained to obtain the dynamic time bending distance between two photovoltaic power plants.
  • each element in one time series and each element in another time series are calculated according to the time series of the same clustering index of the two photovoltaic power stations according to the following method.
  • Dynamic distance one time series of the same clustering index of two photovoltaic power stations is Q, another time series of the same clustering index of two photovoltaic power stations is C, and the wth element q w and time series C in time series Q
  • the dynamic distance D(q w ,c v ) of the vth element c v is:
  • abs(q 1 , c 1 ) represents the absolute value of the first element q 1 in the time series Q and the first element c 1 in the time series C
  • abs(q 1 , c v ) represents the time series Q a first time series elements q 1 and v-th element of C
  • v is the absolute value of C
  • abs (q w, c 1 ) Q represents the time series taking the first element of w and q w of the time series in a C
  • the absolute value of the element c 1 , abs(q w , c v ) represents the absolute value of the wth element q w in the time series Q and the vth element c v in the time series C
  • D(q 1 ,c v- 1 ) represents the dynamic distance of the first element q 1 in the time series Q and the v-1th element c v-1 in the time series C
  • D(q w-1 , c 1 ) represents the w
  • the distance matrix of the same clustering index of the two photovoltaic power stations is determined, and the elements of a time series are nonlinearly mapped according to the minimum cumulative dynamic distortion target.
  • the specific method of the optimal path to another time series is to extract the shortest dynamic distance between the first element in the time series Q and each element in the time series C as a nonlinear mapping of the lth element in the time series Q to
  • the dynamic distance of the lth element q l from the first element c 1 in the time series C, D(q l , c J ) is the lth element q l in the time series Q and the J element c in the time series C
  • the clustering method of the photovoltaic power plant according to the dynamic time bending distance between the photovoltaic power plants is: according to the dynamic between the two photovoltaic power plants
  • the time-bending distance is used for clustering and clustering of photovoltaic power plants.
  • the clustering result evaluation index DBI and the modeling complexity are determined to determine the optimal clustering number.
  • R h max(RD hg ), Where k is the number of clusters, R h is the maximum similarity of cluster h, RD hg is the similarity between cluster h and cluster g, SD h is the compactness inside cluster h, and SD g is the compactness inside cluster g Degree, MD hg is the dispersion of cluster h and cluster g, T h is the number of photovoltaic power plants in cluster h, X d is the time series of the clustering index of the dth photovoltaic power station in cluster h , and A h is the cluster h center The time series of the clustering index of the photovoltaic power plant, a bh is the bth sample value of the clustering index time series of the cluster h center photovoltaic power plant, and a bg is the bth sample value of the clustering index time series of the cluster g central photovoltaic power station, M The number of sampling points included in the time series of clustering indicators for photovolt
  • the nodes with non-zero voltage values are boundary nodes
  • C pvCE and C dcCE are the capacitance values of the PV array and the converter in the cluster center PV power plant respectively, and L dcEQ and L fEQ are respectively single. value converter and power plant filter inductance, L dcCE and inductance L fCE cluster centers are PV plant converters and filters, S tEQ and Z tEQ are equivalent single transformer power plant Rated capacity and impedance, S tCE and Z tCE are the rated capacity and impedance of the transformer in the cluster center photovoltaic power station, respectively.
  • the invention adopts the above technical solution, and has the following beneficial effects: the invention selects six electric quantity waveforms capable of clearly and accurately describing the dynamic characteristics of the photovoltaic power station as a clustering index, first constructs a dynamic distance matrix between the two photovoltaic power stations, and then according to The dynamic distance matrix needs to find the optimal path to map a clustering indicator of a PV power plant to another PV cluster clustering index, and add the dynamic time bending distance of all similar clustering indicators between the two PV power plants to obtain two The dynamic time bending distance between photovoltaic power plants is improved by the dynamic time bending distance to the clustering algorithm based on Euclidean distance.
  • the dynamic time bending distance between two photovoltaic power plants is defined as the similarity between the two photovoltaic power plants and clustered.
  • the distance metric of the algorithm is calculated according to the clustering result, and the equivalent parameters of the cluster PV power plant are calculated and the equivalent simplification of the distribution network is performed.
  • the dynamic similarity between the photovoltaic power plants is accurately captured while overcoming the convergence based on the Euclidean distance.
  • the algorithm of the class class has the disadvantage of low accuracy of clustering results when communication is not synchronized, and details Compared molding method, dynamic equivalence modeling method disclosed in the present invention can significantly reduce the complexity of the model and the simulation time is substantially reduced.
  • Figure 1 is a flow chart of a dynamic equivalence modeling method for photovoltaic power plant clusters.
  • FIG. 2 is a schematic diagram of a PV cluster node number based on clustering results.
  • FIG. 3 is a schematic diagram of a lossless REI equivalent network of cluster i.
  • Figure 4 is a schematic diagram of a Ward equivalent network without internal nodes.
  • Figure 5 is a one-line diagram of an example model.
  • the invention selects six electric quantity waveforms of the photovoltaic power station as clustering indicators to describe the dynamic characteristics of the photovoltaic power station clearly and accurately.
  • the dynamic time bending distance is used based on the Euclidean distance.
  • the clustering algorithm is improved, and a new dynamic clustering algorithm is proposed.
  • FIG. 1 A dynamic equivalence modeling method for a distributed photovoltaic power plant cluster based on clustering technology disclosed in the present invention is shown in FIG. 1 and includes the following steps:
  • Step 10 performing data sampling on the clustering index of the photovoltaic power station in the cluster and performing normalization processing on the data;
  • Step 20 calculating a dynamic time bending distance between the respective photovoltaic power stations and performing clustering grouping of the power stations;
  • Step 30 Calculate the aggregate equivalent parameters of the same cluster photovoltaic power plant and perform equivalent simplification of the distribution network.
  • step 10 As a data preparation phase, the process of data sampling and normalization in step 10) is:
  • Step 101) After the system operating conditions change, the clustering indicators of each photovoltaic power station in the cluster are extracted: the output voltage u pv of the array, the output current i pv of the array, the DC side voltage u dc of the inverter, and the inverter
  • the input current i dc of the device , the active power p output by the photovoltaic power station, and the reactive power q output by the photovoltaic power station, these six variables constitute the clustering indicator matrix WCI, and the WCI is expressed as follows:
  • each variable contains 100 sampling points with equal time distances within the sampling length of 3 power frequency cycles, such as the first variable: [u pv (1),...,u pv (i) ,...,u pv (100)] T ;
  • Step 102 Since the photovoltaic power plants of different capacities may also have close dynamic characteristics, it is necessary to normalize the sampling data of each clustering index in the WCI, and the calculation formula is as follows:
  • x and x norm are the initial value and the normalized value of the sampled data, respectively
  • x max and x min are the maximum and minimum values of the sampled data, respectively.
  • step 20 As the clustering stage of photovoltaic power plants, the photovoltaic power plants with the closest dynamic characteristics in the cluster should be divided into the same group and the optimal number of groups should be determined.
  • the specific process of cluster clustering in step 20) is:
  • Step 201) In order to capture the dynamic trend of the cluster indicator waveform during the clustering process, firstly, the similarity sim(F, G) between the photovoltaic power station F and the photovoltaic power station G needs to be calculated based on the dynamic time bending distance, and the calculation process is as follows:
  • the normalized power plant clustering index waveform obtained in the data sampling stage is regarded as a time series.
  • D(q w , c v ) represents the accumulated value of the object q w in the time series Q and the object c v in the time series C, and abs(q 1 , c 1 ) represents the first element q in the time series Q.
  • abs(q 1 , c v ) represents the absolute value of the first element q 1 in the time series Q and the v element c v in the time series C
  • abs(q w , c 1 ) represents the absolute value of the wth element q w in the time series Q and the first element c 1 in the time series C
  • abs(q w , c v ) represents the time series Q
  • D(q 1 , c v-1 ) represents the first element q 1 in the time series Q and the v- th in the time series C
  • the dynamic distance of one element c v-1 , D(q w-1 , c 1 ) represents the dynamic distance of the w-1th element q w-1 in the time series Q and the first element c 1 in the time series C
  • the generation of the distance matrix A starts from calculating the dynamic distance D(q 1 , c 1 ) and is calculated row by row until the calculation of D(q I , c J );
  • the dynamic distance between the lth element q l in sequence Q and the Jth element c J in time series C defines the dynamic time warping distance DTW of sequences Q and C as:
  • the sum (DTW) of the DTW of all clustering indicators between the photovoltaic power station F and the photovoltaic power station G is defined as the similarity sim(F, G);
  • Step 202) Clustering the similarity calculated based on the dynamic time bending distance in step 201) as the distance metric value, and the clustering algorithm may be specifically selected according to requirements, and the present invention uses the dynamic time bending distance instead of the European style.
  • the clustering algorithm of distance is called dynamic clustering algorithm;
  • Step 203 The determination of the optimal cluster number of the photovoltaic power station in the cluster should be based on the clustering compactness of the clustering results, the inter-cluster dispersion degree, and the cluster modeling complexity under a certain clustering number, in order to investigate the clustering result.
  • the compactness and inter-cluster dispersion of the data of the same cluster power station, the calculation result of the clustering result evaluation index DBI (Davies-Bouldin index) is as follows:
  • T h represents the number of photovoltaic power plants in cluster h
  • X d represents the clustering index matrix of the dth photovoltaic power plant in cluster h
  • a h represents the clustering index matrix of cluster h central photovoltaic power plant
  • a bh represents cluster h
  • the value of the b-th clustering indicator of the central photovoltaic power station ie, the b-th value in the matrix WCI
  • M is the total number of sampling points of the photovoltaic power plant clustering index.
  • Step 301) The same cluster power station is equivalent to a single equivalent power station, and the parameters of each part are calculated as follows:
  • the series number N sEQ and the parallel number N pEQ of the PV array components in a single equivalent power station should be calculated as follows:
  • N sEQ N sCE
  • N pEQ ⁇ N pCE (10)
  • N represents the number of series or parallel
  • subscripts s, p, EQ and CE represent series, parallel, equivalence results and cluster center values
  • N sCE is the number of series photovoltaic modules in the cluster center photovoltaic power station
  • N pCE is The number of parallel photovoltaic modules in the cluster center photovoltaic power station
  • the converter and the filter in the photovoltaic power station are rich in capacitance and inductance.
  • the equivalent components should be able to correspond to the equivalent capacity of the PV array, and on the other hand, the dynamic characteristics before and after the equivalent value must be guaranteed.
  • the specific parameters are calculated as follows:
  • C pv and C dc are the capacitance values of the array and the converter, respectively, and L dc and L f are the inductance values of the converter and the filter, respectively;
  • Each photovoltaic power station is connected to the grid through a transformer, and the aggregate equivalent parameters of the capacity and impedance of the cluster power station are calculated as follows:
  • S t and Z t represent the rated capacity and impedance of the transformer, respectively.
  • control parameters of the photovoltaic power station are equivalent to the control parameters of the cluster center power station.
  • Step 302) The network equivalent includes a step 3021 of constructing a REI lossless network to aggregate the same cluster of active nodes and a step 3022 of constructing a Ward equivalent network to eliminate redundant nodes in the system.
  • the active nodes include all nodes (photovoltaic nodes) and pure load nodes connected to the PV power plant, and the rest are passive nodes.
  • Step 3021 First, construct a REI equivalent network of each cluster node according to the clustering result. As shown in FIG. 3, the network does not cause power loss, and the lossy network of the i-th cluster node generates two new i0 i and im i Add the node to make the node i0 voltage 0, and the new admittance value of the lossless network is calculated as follows:
  • Step 3022 After constructing the lossless network for the active nodes of all clusters, construct a Ward equivalent network, define the PCC as an internal node, define all newly added nodes as boundary nodes, and define other nodes in the network as external nodes, Ward, etc.
  • the value network equation is as follows:
  • subscripts E, B, and I represent external nodes, boundary nodes, and internal nodes, respectively.
  • FIG. 5 Taking a city-photovoltaic power plant cluster as an example, its single-line structure diagram is shown in Figure 5.
  • the cluster contains 20 photovoltaic power plants: PV1 ⁇ PV20.
  • the photovoltaic cluster dynamic equivalence modeling method proposed by the present invention is used for modeling, and the proposed modeling method and detailed modeling method are proposed.
  • the single-potential equivalent modeling method is compared to verify the effectiveness and superiority of the proposed modeling method.
  • the simulation errors of the equivalent model compared with the detailed model under different operating conditions are shown in Table 1.
  • IE p and IE q are simulation errors of active power and reactive power at PCC, respectively. It can be seen from Table 1 that under various operating conditions, the simulation error of the proposed equivalent model and the detailed model is less than 5%, and the proposed equivalent model accuracy is in each working condition. Both are higher than the single station equivalent model.
  • the simulation time of the three models under each working condition is shown in Table 2.
  • the simulation was run on a computer with the following parameters: Intel(R) CPU I7-6500U, 2.50GHz, RAM 8GB.
  • the proposed equivalent model can greatly reduce the simulation running time, and can reduce the simulation time by up to 95.2% (irradiance variation study). .
  • the single-station equivalent model does not save a lot of simulation time than the proposed simulation model. For example, in the irradiance variation study, the single-station equivalent model reduces the simulation time by 97.7% compared with the detailed model. The simulation model saved 2.5% of the simulation time.

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

一种分布式光伏电站集群的动态等值建模方法，属于分布式能源并网建模与仿真的技术领域。本方法选择能够准确描述光伏电站动态特性的电气量波形作为聚类指标，以光伏电站之间所有同类聚类指标的动态时间弯曲距离作为聚类分簇的度量，再根据聚类结果计算同簇光伏电站的聚合等值参数并进行配电网的等值化简，不仅能够准确捕捉光伏电站之间动态相似度，还能降低模型复杂度并缩减仿真计算时间。

Description

一种分布式光伏电站集群的动态等值建模方法 技术领域
本发明公开了一种分布式光伏电站集群的动态等值建模方法，属于分布式能源并网建模与仿真的技术领域。
背景技术
与传统化石能源相比，光伏发电具有资源丰富、可再生和无污染等优势，因此，近年来光伏产业发展迅速。根据国际能源局(IEA)发布的数据，2015年全球新增光伏装机容量接近50GW，相较2014年增长25％，全球累计装机容量约230GW，其中，中国、日本和美国为最大市场，共占据约三分之二的新增装机容量。预计到2050年，光伏发电将占据全球总用电量的16％。
由于光伏发电本身的优点和政府政策的鼓励，分布式光伏电站大量且集中地出现在如工业园区的屋顶以及农村地区的荒地等应用环境中，从而在局部地区配网中形成光伏电站集群。随着电力系统光伏渗透率的增加，光伏电站集群中可包含数十甚至上百个分布式光伏电站，由光伏电站导致的如电力系统电能质量和稳定性运行等问题也逐渐增多。为对这些问题展开研究，首先就需要建立精准的模型来表征光伏集群在系统发生扰动时公共耦合点(PCC)的动态响应特性。但如果对集群中每个电站进行详细建模，不仅增加了仿真模型的复杂度，而且会在数据准备阶段和模型仿真计算阶段耗费大量的时间和精力，这种详细的建模方法无疑限制了其在实际工程中的应用。所以，有必要研究光伏集群系统的动态等值建模以简化详细模型的复杂度、减少数据的需求并减少仿真计算的时间。
未来二三十年，将是我国能源生产消费方式和能源结构调整变革的关键时期，对于分布式可再生能源发电技术而言，将会迎来更加广阔的发展前景和发展机遇。分布式光伏电站集群的动态等值建模将会为高光伏渗透率配网相关问题的分析和处理铺平道路，有助于加快我国智能、绿色、坚强的能源互联网的建设。
发明内容
本发明的发明目的是针对上述背景技术的不足，提供了一种分布式光伏电站集群的动态等值建模方法，对光伏集群系统进行动态等值建模，简化了详细模型的复杂度、减少了数据的需求并减少了仿真计算的时间，解决了对集群中每个电 站进行详细建模增加了仿真模型的复杂度并且会在数据准备阶段和模型仿真计算阶段耗费大量的时间和精力这一技术问题。
本发明为实现上述发明目的采用如下技术方案：
一种分布式光伏电站集群的动态等值建模方法，
提取集群中各光伏电站的聚类指标，采样各光伏电站聚类指标的数据，对各光伏电站聚类指标的采样数据进行归一化处理得到各光伏电站聚类指标的时间序列；
依据各光伏电站聚类指标的时间序列计算各光伏电站之间的动态时间弯曲距离；
依据各光伏电站之间的动态时间弯曲距离对光伏电站进行聚类分簇；
计算同簇光伏电站的聚合等值参数并进行配电网的等值化简。
进一步的，分布式光伏电站集群的动态等值建模方法中，各光伏电站的聚类指标包含但不限于阵列的输出电压、阵列的输出电流、逆变器的直流侧电压、逆变器的输入电流、光伏电站输出的有功功率、光伏电站输出的无功功率。
进一步的，分布式光伏电站集群的动态等值建模方法中，按照如下方法依据各光伏电站聚类指标的时间序列计算各光伏电站之间的动态时间弯曲距离：依据两个光伏电站相同聚类指标的时间序列计算一个时间序列中每一个元素与另一个时间序列中各元素的动态距离进而确定两个光伏电站相同聚类指标的距离矩阵，再根据两个光伏电站相同聚类指标的距离矩阵并以累计动态失真最小为目标确定将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径，累加将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径得到两个光伏电站之间同类聚类指标的动态时间弯曲距离，累加两个光伏电站之间所有同类聚类指标的动态时间弯曲距离得到两个光伏电站之间的动态时间弯曲距离。
再进一步的，分布式光伏电站集群的动态等值建模方法中，按照如下方法依据两个光伏电站相同聚类指标的时间序列计算一个时间序列中每一个元素与另一个时间序列中各元素的动态距离：记两个光伏电站相同聚类指标的一个时间序列为Q，记两个光伏电站相同聚类指标的另一个时间序列为C，时间序列Q中第w个元素q w与时间序列C中第v个元素c v的动态距离D(q w,c v)为：
Figure PCTCN2018084941-appb-000001
其中，abs(q 1,c 1)表示取时间序列Q中第1个元素q 1与时间序列C中第1个元素c 1的绝对值，abs(q 1,c v)表示取时间序列Q中第1个元素q 1与时间序列C中第v个元素c v的绝对值，abs(q w,c 1)表示取时间序列Q中第w个元素q w与时间序列C中第1个元素c 1的绝对值，abs(q w,c v)表示取时间序列Q中第w个元素q w与时间序列C中第v个元素c v的绝对值，D(q 1,c v-1)表示时间序列Q中第1个元素q 1与时间序列C中第v-1个元素c v-1的动态距离，D(q w-1,c 1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第1个元素c 1的动态距离，D(q w-1,c v)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v个元素c v的动态距离，D(q w-1,c v-1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v-1个元素c v-1的动态距离，D(q w,c v-1)表示时间序列Q中第w个元素q w与时间序列C中第v-1个元素c v-1的动态距离，w＝1,…,I，v＝1,…,J，I、J为正整数。
更进一步的，分布式光伏电站集群的动态等值建模方法中，根据两个光伏电站相同聚类指标的距离矩阵并以累计动态失真最小为目标确定将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径的具体方法为：提取时间序列Q中第l个元素与时间序列C中各元素的最短动态距离作为将时间序列Q中第l个元素非线性地映射到时间序列C上的最优路径p l，p l＝min{D(q l,c 1),…,D(q l,c J)}，D(q l,c 1)为时间序列Q中第l个元素q l与时间序列C中第1个元素c 1的动态距离，D(q l,c J)为时间序列Q中第l个元素q l与时间序列C中第J个元素c J的动态距离。
再进一步的，分布式光伏电站集群的动态等值建模方法中，依据各光伏电站之间的动态时间弯曲距离对光伏电站进行聚类分簇的方法为：根据光伏电站两两 之间的动态时间弯曲距离对光伏电站进行聚类分簇处理，综合考虑了同簇紧致度、簇间离散度的聚类结果评价指标DBI以及建模复杂程度确定最优聚类数，
Figure PCTCN2018084941-appb-000002
R h＝max(RD hg)，
Figure PCTCN2018084941-appb-000003
Figure PCTCN2018084941-appb-000004
其中，k为聚类数，R h为簇h的最大相似度，RD hg为簇h与簇g的相似度，SD h为簇h内部的紧致度，SD g为簇g内部的紧致度，MD hg为簇h与簇g的离散度，T h为簇h中光伏电站的个数，X d为簇h中第d个光伏电站聚类指标的时间序列，A h为簇h中心光伏电站聚类指标的时间序列，a bh为簇h中心光伏电站聚类指标时间序列的第b个采样值，a bg为簇g中心光伏电站聚类指标时间序列的第b个采样值，M为光伏电站聚类指标时间序列包含的采样点个数。
作为分布式光伏电站集群的动态等值建模方法的更进一步优化方案，按照如下方法计算同簇光伏电站的聚合等值参数并进行配电网的等值化简：将同簇中的光伏电站等效为单个等值电站，N sEQ＝N sCE，N pEQ＝ρN pCE，C pvEQ＝ρC pvCE，L dcEQ＝L dcCE/ρ，C dcEQ＝ρC dcCE，L fEQ＝L fCE/ρ，S tEQ＝ρS tCE，Z tEQ＝Z tCE/ρ，将纯负荷节点归为一簇并根据光伏电站聚类分簇的结果构建REI等值网络，以公共耦合点为内部节点、以构建REI网络时新增的电压值非零的节点为边界节点、以其它节点为外部节点构建Ward等值网络方程，其中，N sEQ和N pEQ为单个等值电站中光伏组件的串联数和并联数，N sCE为聚类中心光伏电站内串联光伏组件的数目，N pCE为聚类中心光伏电站内并联光伏组件的数目，ρ为该簇光伏电站总额定容量S GR与聚类中心光伏电站额定容量S CE之比，C pvEQ和C dcEQ分别为单个等值电站中光伏阵列和变流器的电容值，C pvCE、C dcCE分别为聚类中心光伏电站内光伏阵列和变流器的电容值，L dcEQ和L fEQ分别为单个等值电站中变流器和滤波器的电感值，L dcCE和L fCE分别为聚类中心光伏电站中变流器和滤波器的电感值，S tEQ和Z tEQ分别为单个等值电站中变压器的额定容量和阻抗，S tCE和Z tCE分别为 聚类中心光伏电站内变压器的额定容量和阻抗。
本发明采用上述技术方案，具有以下有益效果：本发明选择能够清晰且准确地描述光伏电站动态特性的六个电气量波形作为聚类指标，首先构建两光伏电站之间的动态距离矩阵，然后依据动态距离矩阵需找将一光伏电站某一聚类指标非线性映射到另一光伏电站聚类指标的最优路径，累加两个光伏电站之间所有同类聚类指标的动态时间弯曲距离得到两个光伏电站之间的动态时间弯曲距离采用动态时间弯曲距离对基于欧式距离的聚类算法进行改进，定义两个光伏电站之间的动态时间弯曲距离为两光伏电站的相似度并以此为聚类算法的距离度量，再根据聚类结果计算同簇光伏电站的聚合等值参数并进行配电网的等值化简，在准确捕捉光伏电站之间动态相似度的同时克服了基于欧式距离进行聚类的算法在有通信不同步时聚类结果准确率较低的缺点，与详细建模方法相比，本发明公开的动态等值建模方法能够大幅降低模型复杂度并大幅缩减仿真计算时间。
附图说明
图1是光伏电站集群动态等值建模方法流程图。
图2是基于聚类结果的光伏集群节点编号示意图。
图3是簇i的无损REI等值网络示意图。
图4是不含内部节点的Ward等值网络示意图。
图5是示例模型的单线图。
具体实施方式
下面结合附图对发明的技术方案进行详细说明。本发明选择光伏电站的六个电气量波形作为聚类指标可以清晰且准确地描述光伏电站的动态特性，为了捕捉不同电站聚类指标之间的动态相似度，采用动态时间弯曲距离对基于欧式距离的聚类算法进行改进，从而提出了一种新型的动态聚类算法。
本发明公开的一种基于聚类技术的分布式光伏电站集群的动态等值建模方法如图1所示，包括以下步骤：
步骤10)对集群中光伏电站的聚类指标进行数据采样并进行数据归一化处理；
步骤20)计算各个光伏电站之间的动态时间弯曲距离并进行电站的聚类分组；
步骤30)计算同簇光伏电站的聚合等值参数并进行配电网的等值化简。
作为数据准备阶段，步骤10)中数据采样和归一化处理的过程为：
步骤101)在系统运行工况发生变化后，提取集群中各光伏电站的聚类指标包括：阵列的输出电压u pv、阵列的输出电流i pv、逆变器的直流侧电压u dc、逆变器的输入电流i dc、光伏电站输出的有功功率p、光伏电站输出的无功功率q，这六个变量组成了聚类指标矩阵WCI，WCI表示如下：
Figure PCTCN2018084941-appb-000005
其中，上标“～”表示每个变量包含3个工频周期采样长度内等时间距离的100个采样点，如第一个变量为：[u pv(1),…,u pv(i),…,u pv(100)] T；
步骤102)由于不同容量的光伏电站之间也可能拥有接近的动态特性，因此需要将WCI中各聚类指标的采样数据分别进行归一化，其计算公式如下式给出：
Figure PCTCN2018084941-appb-000006
其中，x和x norm分别为采样数据的初始值和归一化值，x max和x min分别为采样数据的最大值和最小值。
作为光伏电站聚类分组阶段，应将集群中动态特性最为相近的光伏电站分为同一组并确定最优的分组数，步骤20)中电站聚类的具体过程为：
步骤201)为在聚类过程中捕捉到聚类指标波形的动态走势，首先需要基于动态时间弯曲距离计算光伏电站F与光伏电站G之间的相似度sim(F,G)，计算过程如下：
首先，将数据采样阶段中获得的归一化电站聚类指标波形看作是时间序列，现设有两个光伏电站的相同聚类指标的时间序列分别为Q＝q 1,…,q e,…,q I和C＝c 1,…,c e,…,c J，序列长度分别为I和J；
然后，定义动态距离D(q w,c v)为：
Figure PCTCN2018084941-appb-000007
D(q w,c v)表示时间序列Q中对象q w和时间序列C中对象c v相异距离的累加值，abs(q 1,c 1)表示取时间序列Q中第1个元素q 1与时间序列C中第1个元素c 1的绝对值，abs(q 1,c v)表示取时间序列Q中第1个元素q 1与时间序列C中第v个元素c v的绝对值，abs(q w,c 1)表示取时间序列Q中第w个元素q w与时间序列C中第1个元素c 1的绝对值，abs(q w,c v)表示取时间序列Q中第w个元素q w与时间序列C中第v个元素c v的绝对值，D(q 1,c v-1)表示时间序列Q中第1个元素q 1与时间序列C中第v-1个元素c v-1的动态距离，D(q w-1,c 1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第1个元素c 1的动态距离，D(q w-1,c v)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v个元素c v的动态距离，D(q w-1,c v-1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v-1个元素c v-1的动态距离，D(q w,c v-1)表示时间序列Q中第w个元素q w与时间序列C中第v-1个元素c v-1的动态距离，随后可以定义距离矩阵A为：
Figure PCTCN2018084941-appb-000008
距离矩阵A的生成从计算动态距离D(q 1,c 1)开始，逐行计算，直至计算到D(q I,c J)为止；
距离矩阵A生成后，需要找到一条最优路径P *＝{p 1,…,p l,…,p I}，该路径可将Q序列非线性地映射到C序列上以使总的累计失真量最小，路径P *的选取规则如下：第l段路径p l为距离矩阵A中第l行的最小值，即p l＝min{D(q l,c 1),…,D(q l,c J)}，D(q l,c 1)为时间序列Q中第l个元素q l与时间序列C中第1个元素c 1的动态距离，D(q l,c J)为时间序列Q中第l个元素q l与时间序列C中第J个元素c J的动态距离，定义序列Q和C的动态时间弯曲距离DTW 为：
Figure PCTCN2018084941-appb-000009
随后将光伏电站F和光伏电站G之间的所有聚类指标的DTW之和sum(DTW)定义为相似度sim(F,G)；
步骤202)将步骤201)中基于动态时间弯曲距离计算而得的相似度作为距离度量值进行聚类，聚类算法可以根据需要具体地选取，本发明将这种基于动态时间弯曲距离而非欧式距离的聚类算法称为动态聚类算法；
步骤203)集群中光伏电站的最优聚类数的确定应该综合聚类结果的同簇紧致度、簇间离散度以及某次聚类数下的集群建模复杂度，为了考察聚类结果中同簇电站数据的紧致度和簇间离散度，聚类结果评价指标DBI的(Davies-Bouldin index)计算式如下：
Figure PCTCN2018084941-appb-000010
其中，k为聚类数，R h为簇h的最大相似度。
为获取R h，首先定义簇h内部的紧致度SD h、簇h与簇g的离散度MD hg以及簇h与簇g的相似度RD hg分别为：
Figure PCTCN2018084941-appb-000011
Figure PCTCN2018084941-appb-000012
Figure PCTCN2018084941-appb-000013
其中，T h表示簇h中光伏电站的个数，X d表示簇h中第d个光伏电站的聚类指标矩阵，A h表示簇h中心光伏电站的聚类指标矩阵，a bh表示簇h中心光伏电站的第b个聚类指标的值(即矩阵WCI中第b个值),M是光伏电站聚类指标采样点的总数。
随后选取与簇h相关的相似度中的最大值为RD h，即R h＝max(RD hg)，其中，g＝1,…,k且g≠h。
DBI的值越小说明聚类结果的组内紧致度和组间离散度越好，综合DBI指标与建模复杂程度来确定最优聚类分组数，随后调整聚算法以使最终聚类数与最优聚类分组数一致。
在获得最终聚类结果之后，需要计算同簇光伏电站的聚合等值参数并进行配电网的等值化简，具体过程为：
步骤301)将同簇电站等效为单个等值电站，其各个部分的参数的计算方法如下：
对于光伏阵列而言，同簇电站等值前后应尽量保持阵列的总容量和出口电压不变，所以单个等值电站中光伏阵列组件的串联数N sEQ和并联数N pEQ应按下式计算：
N sEQ＝N sCE，N pEQ＝ρN pCE     式(10)，
其中，N表示串联或并联数，下标s、p、EQ和CE分别表示串联、并联、等值结果和簇中心值，N sCE为聚类中心光伏电站内串联光伏组件的数目，N pCE为聚类中心光伏电站内并联光伏组件的数目，ρ为该簇光伏电站总额定容量S GR与聚类中心光伏电站额定容量S CE之比，ρ＝S GR/S CE；
光伏电站中的变流器及滤波器中含有丰富的电容及电感，等值后的元件一方面应能对应光伏阵列等值后的容量，另一方面需保证其等值前后的动态特性不变，具体参数按下式计算：
Figure PCTCN2018084941-appb-000014
其中，C pv和C dc分别为阵列和变流器的电容值，L dc和L f分别为变流器和滤波器的电感值；
各光伏电站均通过变压器并网，同簇电站的容量及阻抗的聚合等值参数如下式计算：
S tEQ＝ρS tCE，Z tEQ＝Z tCE/ρ   式(12)，
其中，S t和Z t分别表示变压器的额定容量和阻抗。
对于光伏电站的控制参数而言，各簇电站变流器控制回路的控制参数等值于 聚类中心电站的控制参数。
步骤302)网络等值包括构建REI无损网络以聚合同簇有源节点的步骤3021)以及构建Ward等值网络以消去系统中的冗余节点的步骤3022)。如图2所示，将光伏集群网络中所有节点的下标按照聚类结果编号，有源节点包括所有接有光伏电站的节点(光伏节点)及纯负荷节点，剩下的均为无源节点，根据聚类结果对有源节点进行排序，簇i的节点下标为i1,i2,…,in i(i＝1,2…,k,k+1)，n i表示簇i包含n i个节点，k表示光伏节点的聚类个数为k个，第k+1类为纯负荷节点。
步骤3021)：首先根据聚类结果构造各簇节点的REI等值网络，如图3所示，该网络不会造成功率损耗，第i簇节点的无损网络会产生i0 i和im i两个新增节点，令节点i0电压为0，而无损网络的新增导纳值计算如下：
Figure PCTCN2018084941-appb-000015
Figure PCTCN2018084941-appb-000016
和
Figure PCTCN2018084941-appb-000017
分别为：
Figure PCTCN2018084941-appb-000018
其中，y ij、
Figure PCTCN2018084941-appb-000019
分别为第i簇中第j个节点、第m i个节点的导纳，
Figure PCTCN2018084941-appb-000020
和
Figure PCTCN2018084941-appb-000021
为第i簇中第j个节点的视在功率及其共轭值，
Figure PCTCN2018084941-appb-000022
和 为第i簇的视在功率及其共轭值，U ij、
Figure PCTCN2018084941-appb-000024
分别为第i簇中第j个节点、第m i个节点的电压，j＝1,…,n i，i＝1,…,k+1。
步骤3022)在对所有簇的有源节点构造无损网络之后构建Ward等值网络，将PCC定义为内部节点，将所有新增节点定义为边界节点，将网络中其它节点定义为外部节点，Ward等值网络方程如下所示：
Figure PCTCN2018084941-appb-000025
其中，下标E、B和I分别表示外部节点、边界节点和内部节点。
根据该式便可搭建如图4所示的不含内部节点的简化网络。对于含有大量节点的系统，该等值过程可以消去绝大部分节点。
下面列举一实施例以辅助说明本发明公开的动态等值建模方法的技术优 势。
以某城市一光伏电站集群为例，其单线结构图如图5所示，该集群含有20个光伏电站：PV1～PV20。分别在辐照度变动、负荷变动以及三相短路故障三个算例中釆用本发明提出的光伏集群动态等值建模方法进行建模，并将所提出的建模方法与详细建模方法和单电站等值建模方法进行对比以验证所提建模方法的有效性和优越性。等值模型在不同运行状态下与详细模型相比的仿真误差如表1所示。
表1等值模型在各种运行工况下的仿真误差
Figure PCTCN2018084941-appb-000026
其中，IE p和IE q分别为PCC处有功功率和无功功率的仿真误差。由表1可知，在各种运行工况下，本发明所提出的等值模型与详细建模的模型相比的仿真误差均小于5％，且所提出的等值模型准确度在各工况下均比单电站等值模型要高。
在各工况下3种模型的仿真时间如表2所示。仿真是在具有如下参数的计算机上运行的：Intel(R)CPU I7-6500U,2.50GHz,RAM 8GB。
表2各运行工况下不同模型的仿真时间
Figure PCTCN2018084941-appb-000027
由表2可知，在各种运行工况下，相对于详细建模而言，所提出的等值模型可以大幅减少仿真运行时间，最多可减少95.2％的仿真时间(辐照度变动算例)。 而单电站等值模型并未比所提出的仿真模型节省大量的仿真时间，如在辐照度变动算例中，单电站等值模型比详细模型减少了97.7％的仿真时间，仅比所提出的仿真模型节约了2.5％的仿真时间。

Claims (7)

  1. 一种分布式光伏电站集群的动态等值建模方法，其特征在于，
    提取集群中各光伏电站的聚类指标，采样各光伏电站聚类指标的数据，对各光伏电站聚类指标的采样数据进行归一化处理得到各光伏电站聚类指标的时间序列；
    依据各光伏电站聚类指标的时间序列计算各光伏电站之间的动态时间弯曲距离；
    依据各光伏电站之间的动态时间弯曲距离对光伏电站进行聚类分簇；
    计算同簇光伏电站的聚合等值参数并进行配电网的等值化简。
  2. 根据权利要求1所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，各光伏电站的聚类指标包含但不限于阵列的输出电压、阵列的输出电流、逆变器的直流侧电压、逆变器的输入电流、光伏电站输出的有功功率、光伏电站输出的无功功率。
  3. 根据权利要求1所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，按照如下方法依据各光伏电站聚类指标的时间序列计算各光伏电站之间的动态时间弯曲距离：依据两个光伏电站相同聚类指标的时间序列计算一个时间序列中每一个元素与另一个时间序列中各元素的动态距离进而确定两个光伏电站相同聚类指标的距离矩阵，再根据两个光伏电站相同聚类指标的距离矩阵并以累计动态失真最小为目标确定将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径，累加将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径得到两个光伏电站之间同类聚类指标的动态时间弯曲距离，累加两个光伏电站之间所有同类聚类指标的动态时间弯曲距离得到两个光伏电站之间的动态时间弯曲距离。
  4. 根据权利要求3所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，按照如下方法依据两个光伏电站相同聚类指标的时间序列计算一个时间序列中每一个元素与另一个时间序列中各元素的动态距离：记两个光伏电站相同聚类指标的一个时间序列为Q，记两个光伏电站相同聚类指标的另一个时间序 列为C，时间序列Q中第w个元素q w与时间序列C中第v个元素c v的动态距离D(q w,c v)为：
    Figure PCTCN2018084941-appb-100001
    其中，abs(q 1,c 1)表示取时间序列Q中第1个元素q 1与时间序列C中第1个元素c 1的绝对值，abs(q 1,c v)表示取时间序列Q中第1个元素q 1与时间序列C中第v个元素c v的绝对值，abs(q w,c 1)表示取时间序列Q中第w个元素q w与时间序列C中第1个元素c 1的绝对值，abs(q w,c v)表示取时间序列Q中第w个元素q w与时间序列C中第v个元素c v的绝对值，D(q 1,c v-1)表示时间序列Q中第1个元素q 1与时间序列C中第v-1个元素c v-1的动态距离，D(q w-1,c 1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第1个元素c 1的动态距离，D(q w-1,c v)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v个元素c v的动态距离，D(q w-1,c v-1)表示时间序列Q中第w-1个元素q w-1与时间序列C中第v-1个元素c v-1的动态距离，D(q w,c v-1)表示时间序列Q中第w个元素q w与时间序列C中第v-1个元素c v-1的动态距离，w＝1,…,I，v＝1,…,J，I、J为正整数。
  5. 根据权利要求4所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，根据两个光伏电站相同聚类指标的距离矩阵并以累计动态失真最小为目标确定将一个时间序列上各元素非线性地映射到另一个时间序列上的最优路径的具体方法为：提取时间序列Q中第l个元素与时间序列C中各元素的最短动态距离作为将时间序列Q中第l个元素非线性地映射到时间序列C上的最优路径p l，p l＝min{D(q l,c 1),…,D(q l,c J)}，D(q l,c 1)为时间序列Q中第l个元素q l与时间序列C中第1个元素c 1的动态距离，D(q l,c J)为时间序列Q中第l个元素 q l与时间序列C中第J个元素c J的动态距离。
  6. 根据权利要求5所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，依据各光伏电站之间的动态时间弯曲距离对光伏电站进行聚类分簇的方法为：根据光伏电站两两之间的动态时间弯曲距离对光伏电站进行聚类分簇处理，综合考虑了同簇紧致度、簇间离散度的聚类结果评价指标DBI以及建模复杂程度确定最优聚类数，
    Figure PCTCN2018084941-appb-100002
    R h＝max(RD hg)，
    Figure PCTCN2018084941-appb-100003
    Figure PCTCN2018084941-appb-100004
    其中，k为聚类数，R h为簇h的最大相似度，RD hg为簇h与簇g的相似度，SDh为簇h内部的紧致度，SD g为簇g内部的紧致度，MD hg为簇h与簇g的离散度，T h为簇h中光伏电站的个数，X d为簇h中第d个光伏电站聚类指标的时间序列，A h为簇h中心光伏电站聚类指标的时间序列，a bh为簇h中心光伏电站聚类指标时间序列的第b个采样值，a bg为簇g中心光伏电站聚类指标时间序列的第b个采样值，M为光伏电站聚类指标时间序列包含的采样点个数。
  7. 根据权利要求1至6中任意一项所述一种分布式光伏电站集群的动态等值建模方法，其特征在于，按照如下方法计算同簇光伏电站的聚合等值参数并进行配电网的等值化简：将同簇中的光伏电站等效为单个等值电站，N sEQ＝N sCE，N pEQ＝ρN pCE，C pvEQ＝ρC pvCE，L dcEQ＝L dcCE/ρ，C dcEQ＝ρC dcCE，L fEQ＝L fCE/ρ，S tEQ＝ρS tCE，Z tEQ＝Z tCE/ρ，将纯负荷节点归为一簇并根据光伏电站聚类分簇的结果构建REI等值网络，以公共耦合点为内部节点、以构建REI网络时新增的电压值非零的节点为边界节点、以其它节点为外部节点构建Ward等值网络方程，其中，N sEQ和N pEQ为单个等值电站中光伏组件的串联数和并联数，N sCE为聚类中心光伏电站内串联光伏组件的数目，N pCE为聚类中心光伏电站内并联光伏组件的数目，ρ为该簇光伏电站总额定容量S GR与聚类中心光伏电站额定容量S CE 之比，C pvEQ和C dcEQ分别为单个等值电站中光伏阵列和变流器的电容值，C pvCE、C dcCE分别为聚类中心光伏电站内光伏阵列和变流器的电容值，L dcEQ和L fEQ分别为单个等值电站中变流器和滤波器的电感值，L dcCE和L fCE分别为聚类中心光伏电站中变流器和滤波器的电感值，S tEQ和Z tEQ分别为单个等值电站中变压器的额定容量和阻抗，S tCE和Z tCE分别为聚类中心光伏电站内变压器的额定容量和阻抗。
PCT/CN2018/084941 2017-09-20 2018-04-27 一种分布式光伏电站集群的动态等值建模方法 WO2019056753A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710853068.8 2017-09-20
CN201710853068.8A CN107565601A (zh) 2017-09-20 2017-09-20 一种分布式光伏电站集群的动态等值建模方法

Publications (1)

Publication Number Publication Date
WO2019056753A1 true WO2019056753A1 (zh) 2019-03-28

Family

ID=60982180

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/084941 WO2019056753A1 (zh) 2017-09-20 2018-04-27 一种分布式光伏电站集群的动态等值建模方法

Country Status (2)

Country Link
CN (1) CN107565601A (zh)
WO (1) WO2019056753A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116865336A (zh) * 2023-07-06 2023-10-10 南方电网科学研究院有限责任公司 大规模光伏电站实时仿真的并行化单元级建模方法及装置
CN117094206A (zh) * 2023-06-19 2023-11-21 安徽云图信息技术有限公司 一种仿真数据生成系统和生成方法

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107565601A (zh) * 2017-09-20 2018-01-09 东南大学 一种分布式光伏电站集群的动态等值建模方法
CN109145503B (zh) * 2018-09-18 2023-07-21 东南大学 一种光伏电站集群的高精度动态建模方法
CN109193649A (zh) * 2018-10-23 2019-01-11 东南大学 基于深度信念网络的分布式光伏集群动态等值建模方法
CN109167395B (zh) * 2018-11-22 2021-10-15 国网宁夏电力有限公司电力科学研究院 基于adpss的光伏发电系统的暂态等值电势辨识方法
CN109638892B (zh) * 2019-01-25 2022-04-26 合肥工业大学 一种基于改进模糊聚类算法的光伏电站等值建模方法
CN110571794B (zh) * 2019-08-26 2023-07-25 国家电网公司东北分部 一种适用于双馈风机风电场的暂态模型等值计算方法
CN110544039A (zh) * 2019-09-06 2019-12-06 阳光电源股份有限公司 一种光伏组串的阴影遮挡识别方法及装置
CN111199016B (zh) * 2019-09-29 2023-11-07 国网湖南省电力有限公司 一种基于DTW的改进K-means的日负荷曲线聚类方法
CN111525612B (zh) * 2020-04-28 2021-08-24 东南大学 适用于新能源并网系统的实时稳定控制装置
CN112966364B (zh) * 2021-02-03 2023-03-28 中国大唐集团科学技术研究院有限公司华东电力试验研究院 一种用于特征值计算的光伏电站等值模型建模方法及装置
CN112993984B (zh) * 2021-02-26 2022-08-16 南瑞集团有限公司 一种分布式光伏电站在有电气连接情况下的大扰动聚合方法
CN116545037B (zh) * 2023-07-03 2023-09-26 广东电网有限责任公司中山供电局 一种面向分布式资源集群的轻量化边缘控制方法及装置
CN116599159B (zh) * 2023-07-13 2023-10-13 湖北长江电气有限公司 光伏并网逆变器的智能控制方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106054665A (zh) * 2016-05-27 2016-10-26 国网浙江省电力公司电力科学研究院 一种大规模光伏逆变器系统的分群等值建模方法
US20160344194A1 (en) * 2011-07-28 2016-11-24 Tigo Energy, Inc. Systems and methods to reduce the number and cost of management units of distributed power generators
CN106451567A (zh) * 2016-10-14 2017-02-22 国网浙江省电力公司电力科学研究院 一种分布式电源集群动态划分的方法及系统
CN106529731A (zh) * 2016-11-17 2017-03-22 云南电网有限责任公司电力科学研究院 一种区域电网光伏电站集群划分方法
CN106778824A (zh) * 2016-11-24 2017-05-31 河南理工大学 一种面向时间序列数据的增量模糊c中心点聚类方法
CN107565601A (zh) * 2017-09-20 2018-01-09 东南大学 一种分布式光伏电站集群的动态等值建模方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105938578A (zh) * 2016-04-13 2016-09-14 合肥工业大学 一种基于聚类分析的大规模光伏电站等值建模方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160344194A1 (en) * 2011-07-28 2016-11-24 Tigo Energy, Inc. Systems and methods to reduce the number and cost of management units of distributed power generators
CN106054665A (zh) * 2016-05-27 2016-10-26 国网浙江省电力公司电力科学研究院 一种大规模光伏逆变器系统的分群等值建模方法
CN106451567A (zh) * 2016-10-14 2017-02-22 国网浙江省电力公司电力科学研究院 一种分布式电源集群动态划分的方法及系统
CN106529731A (zh) * 2016-11-17 2017-03-22 云南电网有限责任公司电力科学研究院 一种区域电网光伏电站集群划分方法
CN106778824A (zh) * 2016-11-24 2017-05-31 河南理工大学 一种面向时间序列数据的增量模糊c中心点聚类方法
CN107565601A (zh) * 2017-09-20 2018-01-09 东南大学 一种分布式光伏电站集群的动态等值建模方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117094206A (zh) * 2023-06-19 2023-11-21 安徽云图信息技术有限公司 一种仿真数据生成系统和生成方法
CN117094206B (zh) * 2023-06-19 2024-01-26 安徽云图信息技术有限公司 一种仿真数据生成系统和生成方法
CN116865336A (zh) * 2023-07-06 2023-10-10 南方电网科学研究院有限责任公司 大规模光伏电站实时仿真的并行化单元级建模方法及装置

Also Published As

Publication number Publication date
CN107565601A (zh) 2018-01-09

Similar Documents

Publication Publication Date Title
WO2019056753A1 (zh) 一种分布式光伏电站集群的动态等值建模方法
CN109193649A (zh) 基于深度信念网络的分布式光伏集群动态等值建模方法
CN105656031B (zh) 基于高斯混合分布特征的含风电电力系统安全风险评估方法
CN111428355B (zh) 一种电力负荷数字统计智能综合的建模方法
CN109086928A (zh) 基于saga-fcm-lssvm模型的光伏电站实时功率预测方法
CN110556871B (zh) 基于结构保持方法的大规模光伏发电系统聚合等值方法
WO2013174145A1 (zh) 基于改进人工鱼群混合优化算法的大规模风电并网无功电压优化方法
CN102684201B (zh) 一种基于电压越限概率的含风电场电网无功优化方法
CN110321919A (zh) 一种用于低压单相用户户相关系辨识的聚类方法
CN105069236B (zh) 考虑风电场节点空间相关性的广义负荷联合概率建模方法
CN106651660B (zh) 基于g1-熵权法寻找电网静态薄弱点的综合评价方法
CN105938578A (zh) 一种基于聚类分析的大规模光伏电站等值建模方法
CN105512766A (zh) 一种风电场功率预测方法
CN109409575A (zh) 基于Gap Statistic的风电场机群划分方法
CN109638892B (zh) 一种基于改进模糊聚类算法的光伏电站等值建模方法
CN115392133A (zh) 采用高斯混合模型的风电场最优聚类等值方法
CN115564142A (zh) 一种混合储能系统的选址定容优化方法及系统
CN112993984B (zh) 一种分布式光伏电站在有电气连接情况下的大扰动聚合方法
Ni et al. A review of line loss analysis of the low-voltage distribution system
CN109494787B (zh) 一种光伏多馈入电力系统并网容量极限计算方法
CN108460228B (zh) 一种基于多目标优化算法进行风电场等值的方法
CN105207255B (zh) 一种适用于风电出力的电力系统调峰计算方法
CN116664870A (zh) 一种基于迁移学习的电能质量扰动识别方法
CN109802383A (zh) 基于聚类算法的分布式发电系统等值建模方法
CN114357787B (zh) 一种海上风电场等值建模方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18858020

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18858020

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 18858020

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 17/12/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18858020

Country of ref document: EP

Kind code of ref document: A1