CN116131261A - Micro-grid dynamic equivalent modeling method considering model robustness - Google Patents
Micro-grid dynamic equivalent modeling method considering model robustness Download PDFInfo
- Publication number
- CN116131261A CN116131261A CN202310130423.4A CN202310130423A CN116131261A CN 116131261 A CN116131261 A CN 116131261A CN 202310130423 A CN202310130423 A CN 202310130423A CN 116131261 A CN116131261 A CN 116131261A
- Authority
- CN
- China
- Prior art keywords
- model
- micro
- grid
- equivalent
- characteristic data
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000007787 long-term memory Effects 0.000 claims abstract description 15
- 230000006403 short-term memory Effects 0.000 claims abstract description 9
- 238000012216 screening Methods 0.000 claims abstract description 6
- 230000001360 synchronised effect Effects 0.000 claims description 39
- 239000011159 matrix material Substances 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 18
- 230000015654 memory Effects 0.000 claims description 11
- 230000035945 sensitivity Effects 0.000 claims description 11
- 230000005284 excitation Effects 0.000 claims description 10
- 230000001052 transient effect Effects 0.000 claims description 10
- 230000003068 static effect Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 9
- 239000013598 vector Substances 0.000 claims description 9
- 238000004088 simulation Methods 0.000 claims description 8
- 238000013178 mathematical model Methods 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 6
- 230000001186 cumulative effect Effects 0.000 claims description 4
- 239000002245 particle Substances 0.000 claims description 3
- 238000013016 damping Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/25—Integrating or interfacing systems involving database management systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
-
- 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
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Software Systems (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Geometry (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Power Engineering (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses a dynamic equivalent modeling method of a micro-grid taking model robustness into consideration, which comprises the steps of firstly establishing an operation characteristic database of the micro-grid to represent time-varying and random characteristics of the micro-grid; then determining the equivalent model structure of the micro-grid, and determining the equivalent parameters of the equivalent model of the micro-grid corresponding to each characteristic data by adopting an equivalent model parameter identification method based on key parameter screening; secondly, generalizing equivalent model parameters by adopting a long-term and short-term memory network; and finally, establishing an equivalent model parameter online matching method based on Fisher discriminant criteria.
Description
Technical Field
The invention belongs to the technical field of power systems, and particularly relates to a dynamic equivalent modeling method of a micro-grid in consideration of model robustness.
Background
In recent years, with the massive access of distributed power sources and flexible loads, the problem of safety and stability of micro-grids is increasingly prominent. Due to the reasons of multiple internal elements, various types, difficult parameter acquisition and the like of the micro-grid, when the transient characteristics of the micro-grid are analyzed and the influence of the micro-grid on the safe and stable operation of the power grid connected with the micro-grid is explored, a dynamic equivalent modeling method is generally adopted to carry out simplified modeling on the micro-grid.
However, in practice, it is found that due to the randomness of the distributed power supply and the time-varying characteristics of the flexible load, the conventional equivalent model established based on a certain operation state of the system has the problem of low robustness, that is, the equivalent model can only characterize the dynamic characteristics of the micro-grid at a certain or certain operation points, and cannot accurately reflect all the characteristics of the system. Therefore, under the large background that the distributed power supply and the flexible load are continuously increased, the establishment of the dynamic equivalent modeling method of the micro-grid taking the robustness of the model into consideration has important theoretical and engineering significance.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a dynamic equivalent modeling method of a micro-grid taking model robustness into consideration, wherein the robustness of the micro-grid equivalent model is improved by clustering typical running states of the micro-grid and generalizing equivalent model parameters by adopting a long-period memory network.
In order to achieve the purpose of the invention, the dynamic equivalent modeling method of the micro-grid taking the robustness of the model into consideration is characterized by comprising the following steps:
(1) Establishing an operation characteristic database of the micro-grid;
recording N at power system common connection point under different external fault conditions 1 And the transient characteristic data are set and serve as an operation characteristic database of the micro-grid, wherein each set of characteristic data comprises the following information:
[P i Q i S i C i ]
wherein i=1, 2, …, N 1 ,N 1 The number of the characteristic data sets; p (P) i and Qi Active power and reactive power at the public connection point of the power system recorded in the ith group of characteristic data are respectively recorded; p (P) SGi The active power sum of all synchronous power supplies in the micro-grid in the ith group of characteristic data; p (P) VSCi The active power sum of all distributed power sources which are connected with the grid through the converter in the micro-grid in the ith group of characteristic data; p (P) i,0 Is P i The corresponding steady state value; s is S i and Ci The method comprises the steps that the proportion of the total active output of a source distribution network occupied by all synchronous power supplies and all distributed power supplies connected through a converter in the ith group of characteristic data is respectively expressed;
(2) Establishing a micro-grid equivalent model;
the micro-grid is simplified and modeled into a static load model, a synchronous generator model and a voltage source type converter model which are connected in parallel at a common connection point of an electric power system; meanwhile, a line impedance model is connected in series between the public connection point of the electric power system, the parallel synchronous generator model and the voltage source converter model;
the output active power and reactive power of the micro-grid equivalent model at the public connection point of the power system are as follows:
P=P eq,t +I·R eq,t +P Gen,t +P VSC,t
Q=Q eq,t +I·X eq,t +Q Gen,t +Q VSC,t
wherein P and Q are respectively the output active power and reactive power of the micro-grid equivalent model at the public connection point of the power system, and P eq,t and Qeq,t Respectively equivalent active power and reactive power of static load, R eq,t and Xeq,t Equivalent impedance and equivalent reactance of the line impedance, P Gen,t and QGen,t Equivalent active power and reactive power of synchronous generator, P VSC,t and QVSC,t Respectively equivalent active power and reactive power of the voltage source type converter, wherein I is equivalent current flowing through a line impedance model;
(3) Identifying equivalent model parameters based on key parameter screening;
(3.1) calculating a track sensitivity value Ts of the equivalent model parameters of the micro-grid:
wherein ,θj A j-th parameter representing the equivalent model of the micro-grid; θ j0 Represents θ j K is the number of active power sampling points; p (P) 0 Representing steady state value of active power, delta theta j Representing the parameter θ j Offset from the standard value;
(3.2) setting a key parameter selection threshold epsilon, setting parameters with track sensitivity values larger than the threshold as key parameters, and setting other parameters as non-key parameters;
(3.3) identifying key parameters;
setting non-key parameters of the equivalent model of the micro-grid as theoretical values, identifying the key parameters by adopting a particle swarm algorithm, wherein the objective function of the parameter identification is as follows:
wherein , and />The root mean square error and delta P of the active power and reactive power measured values and the model simulation output value at the public connection point of the power system in the ith group of characteristic data are respectively i (k)、ΔQ i (k) Respectively obtaining differences between active power and reactive power measured values and model simulation output values at a kth sampling point at a public connection point of an electric power system in an ith group of characteristic data;
(3.4) identifying equivalent model parameters by utilizing each group of characteristic data recorded by the micro-grid operation characteristic database to obtain N 1 Identification results of the equivalent model key parameters corresponding to the group feature data;
(4) Clustering the feature data sets;
(4.1) operating N in characteristic database of micro-grid based on k-means++ algorithm 1 Clustering the group feature data, and clustering the feature data with similar dynamic characteristics into one type, wherein an input feature matrix of a k-means++ algorithm is as follows:
wherein ,Qi,0 Is Q i Corresponding steady state value, P i,max and Qi,max The maximum values of active power and reactive power are measured by the public connection point of the power system under the fault condition;
(4.2) determining an optimal cluster number;
based on Silhouette criterion, calculating the optimal cluster number index of the k-means++ algorithm as follows:
wherein n is the number of clusters, a k For sample k and the residuals in the same clusterAverage distance of the remaining samples; b k The minimum value of the sample distance between the sample k and the samples in the non-same cluster;
taking the number N of clusters corresponding to the maximum value of kappa as the optimal cluster number, and recording as N 2 Thereby N is arranged 1 Group feature data clustering as N 2 A set of group feature data;
(5) Generalizing equivalent model parameters;
(5.1) setting an input matrix for long-term and short-term memory network training as follows:
setting an output matrix of long-term and short-term memory network training as follows:
wherein ,is->The number of feature data sets in the set of feature data sets; />Is->Group feature data set +.>The mth key parameter of the equivalent model corresponding to the characteristic data,/th key parameter of the equivalent model corresponding to the characteristic data>M is the number of key parameters;
(5.2) repeating the step (5.1), and training a long-short-period memory network through each group of characteristic data set to realize the generalization of equivalent model parameters;
(6) On-line matching of equivalent model parameters;
(6.1), N-based 2 A set of characteristic data sets, and a discriminant function of Fisher discriminant criteria is established;
(6.2) judging a characteristic data set to which the new characteristic data belongs according to a judging criterion for the new characteristic data, and then selecting a corresponding long-period and short-period memory network;
(6.3) invoking a long-term memory network, the input of the long-term memory network being [ P ] 0 Q 0 P max Q max S C]The output of the long-short-period memory network is the equivalent parameter of the equivalent model, wherein P 0 and Q0 Steady state values corresponding to active power and reactive power of the new characteristic data, P max and Qmax The maximum values of active power and reactive power are measured at the public connection point of the power system through the new characteristic data under the fault condition respectively; s and C respectively represent the proportion of the total active output of the source distribution network occupied by all synchronous power supplies and all distributed power supplies connected through the converter in the new characteristic data.
The invention aims at realizing the following steps:
the invention relates to a dynamic equivalent modeling method of a micro-grid, which considers model robustness, and comprises the steps of firstly establishing an operation characteristic database of the micro-grid so as to represent time-varying and random characteristics of the micro-grid; then determining the equivalent model structure of the micro-grid, and determining the equivalent parameters of the equivalent model of the micro-grid corresponding to each characteristic data by adopting an equivalent model parameter identification method based on key parameter screening; secondly, generalizing equivalent model parameters by adopting a long-term and short-term memory network; and finally, establishing an equivalent model parameter online matching method based on Fisher discriminant criteria.
Meanwhile, the micro-grid dynamic equivalent modeling method considering model robustness also has the following steps of
The beneficial effects are that:
(1) According to the invention, the characteristic data sets are clustered to effectively distinguish typical operating points of the micro-grid, and corresponding equivalent models are respectively built based on different operating points, so that the robustness of the equivalent models of the micro-grid is remarkably improved.
(2) The invention avoids the problem of multiple solutions in the parameter identification process through key parameter selection, and improves the accuracy of parameter identification.
(3) According to the invention, the nonlinear mapping characteristic of the long-term and short-term memory network is utilized to generalize the equivalent model parameters of the micro-grid, so that the robustness of the equivalent model is further improved.
Drawings
FIG. 1 is a simplified mathematical model schematic of a microgrid;
FIG. 2 is a flow chart of a dynamic equivalent modeling method of a micro-grid in consideration of model robustness;
FIG. 3 is a topological structure diagram of a microgrid;
FIG. 4 is an optimal cluster number index map;
FIG. 5 is an active power graph at a common junction of an electrical system under typical fault condition 1;
fig. 6 is an active power graph at a common junction of an electrical system under typical fault condition 2.
Detailed Description
The following description of the embodiments of the invention is presented in conjunction with the accompanying drawings to provide a better understanding of the invention to those skilled in the art. It is to be expressly noted that in the description below, detailed descriptions of known functions and designs are omitted here as perhaps obscuring the present invention.
Examples
In this embodiment, as shown in fig. 2, the method for modeling dynamic equivalence of a micro-grid in consideration of model robustness mainly includes six steps: s1, establishing an operation characteristic database of a micro-grid, S2, establishing a micro-grid equivalent model, S3, identifying equivalent model parameters based on key parameter screening, S4, clustering a characteristic data set, S5, generalizing equivalent model parameters, and S6, and carrying out online matching on equivalent model parameters;
in this embodiment, a real micro-grid test model is built in MATLAB software, and the topology structure of the real micro-grid test model is shown in fig. 3. The test model contained 2.7MW distributed wind power, 1MW distributed photovoltaic, 3.2MW diesel generator, 2KVA energy storage battery and a static load of 6.2MW/0.8 Mvar. The following will describe the above six steps in detail with reference to fig. 3, and as shown in fig. 2, the following steps are specifically included:
s1, establishing an operation characteristic database of a micro-grid;
the micro-grid consists of different distributed power supplies and different loads, and N at the public connection point of the power system is recorded under different external fault conditions 1 =1000 sets of transient feature data, and as an operational feature database for the microgrid, wherein each set of feature data contains the following information:
[P i Q i S i C i ]
wherein i=1, 2, …, N 1 ,N 1 The number of the characteristic data sets; p (P) i and Qi Active power and reactive power at the public connection point of the power system recorded in the ith group of characteristic data are respectively recorded; p (P) SGi The active power sum of all synchronous power supplies in the micro-grid in the ith group of characteristic data; p (P) VSCi The active power sum of all distributed power sources which are connected with the grid through the converter in the micro-grid in the ith group of characteristic data; p (P) i,0 Is P i The corresponding steady state value; s is S i and Ci The method comprises the steps that the proportion of the total active output of a source distribution network occupied by all synchronous power supplies and all distributed power supplies connected through a converter in the ith group of characteristic data is respectively expressed;
in this embodiment, we divide the feature data into 2 groups, the first group of feature data sets contains 900 samples, called modeling feature data sets, for dynamic equivalence modeling of the microgrid; the second set of feature databases contains 100 samples, called test feature data sets, for model generalization capability verification.
S2, establishing a micro-grid equivalent model;
as shown in fig. 1, the micro-grid is modeled in a simplified manner as a static load model, a synchronous generator model and a voltage source type converter model connected in parallel at a common connection point of the power system; meanwhile, a line impedance model is connected in series between the public connection point of the electric power system, the parallel synchronous generator model and the voltage source converter model;
in this embodiment, the mathematical model of the static load model is:
L 1 =P eq,t +jQ eq,t
where j is the imaginary part.
The mathematical model of the line impedance model is:
L 2 =R eq,t +jX eq,t 。
the mathematical model of the voltage source type converter model is as follows:
wherein ,Kp,t 、K q,t 、K pc,t and Kqc,t Gain coefficients, T, of the voltage source converters respectively p,t 、T q,t 、T pc,t and Tqc,t Respectively the time constants, m of the voltage source type converter d and mq Duty cycle of d-axis and q-axis, P ref and Qref The reference active power and the reference reactive power, respectively, s represents the laplace operator.
The synchronous generator model comprises a synchronous generator body model, a synchronous generator speed regulator model and a synchronous generator excitation model;
the mathematical expression of the synchronous generator body model is as follows:
wherein ,Xd,t and Xq,t D-axis and q-axis reactance, X' d,t and Xq ′ ,t Respectively, the transient reactance of d axis and q axis, T d ′ 0,t and Tq ′ 0,t Open-loop transient time constants of d axis and q axis respectively, H t and Dt Rotor inertia and damping factor, respectively; e's' d and Eq ' d-axis and q-axis transient potentials, E f For excitation potential, i d and iq The currents of the d axis and the q axis are respectively, delta is the included angle between the q axis and the reference axis of the synchronous generator, omega is the rotating speed of the rotor, and M m and Me Respectively a mechanical torque and an electromagnetic torque;
the mathematical expression of the synchronous generator speed regulator model is:
wherein ,Ptur To output active power omega for speed regulator R To reference rotor speed, K d,t To synchronize the gain coefficient of the generator, T s,t and T0,t U is the time constant of the speed regulator 0 Is the opening reference valueS represents the laplace operator;
the mathematical expression of the excitation model of the synchronous generator is as follows:
wherein ,T1,t 、T 2,t 、T 3,t 、T 4,t 、T r,t and Ta,t Respectively the time constant, K of the synchronous generator t 、K a,t and Kc,t Respectively are excitation gain coefficients E fd and Ifd Respectively the exciting voltage and exciting current, V ref For reference voltage u c To measure the voltage.
In this embodiment, the output active power and reactive power of the micro-grid equivalent model at the system public connection point are:
P=P eq,t +i·R eq,t +P Gen,t +P VSC,t
Q=Q eq,t +i·X eq,t +Q Gen,t +Q VSC,t
wherein P and Q are respectively the output active power and reactive power of the micro-grid equivalent model at the system public connection point, and P eq,t and Qeq,t Respectively equivalent active power and reactive power of static load, R eq,t and Xeq,t Equivalent impedance and equivalent reactance of the line impedance, P Gen,t and QGen,t Equivalent active power and reactive power of synchronous generator, P VSC,t and QVSC,t The equivalent active power and reactive power of the voltage source type converter are respectively, and i is the equivalent current flowing through the line impedance model.
S3, identifying equivalent model parameters based on key parameter screening;
s3.1, calculating a track sensitivity value Ts of the equivalent model parameters of the micro-grid:
wherein ,θj A j-th parameter representing the equivalent model of the micro-grid; θ j0 Represents θ j K is the number of active power sampling points; p (P) 0 Representing steady state value of active power, delta theta j Representing the parameter θ j Offset from the standard value;
in the present embodiment, sensitivity values of the synchronous generator body parameter, the excitation system parameter, the governor parameter, and the voltage source converter parameter are shown in tables 1 to 4, respectively.
Table 1 is the synchronous generator body parameter sensitivity.
TABLE 1
Table 2 is the synchronous generator excitation system parameter sensitivity.
TABLE 2
Table 3 is synchronous generator governor parameter sensitivity.
TABLE 3 Table 3
Table 4 shows the voltage source converter parameter sensitivities.
TABLE 4 Table 4
S3.2, setting a key parameter selection threshold epsilon=0.3, setting parameters with track sensitivity values larger than the threshold as key parameters, and setting other parameters as non-key parameters;
in this embodiment, key parameters of the micro-grid equivalent model are shown in table 5:
table 5 is the micro grid equivalent model key parameters.
TABLE 5
S3.3, identifying key parameters;
setting non-key parameters of the equivalent model of the micro-grid as theoretical values, identifying the key parameters by adopting a particle swarm algorithm, wherein the objective function of the parameter identification is as follows:
wherein , and />The root mean square error and delta P of the active power and reactive power measured values and the model simulation output value at the public connection point of the power system in the ith group of characteristic data are respectively i (k)、ΔQ i (k) Respectively obtaining differences between active power and reactive power measured values and model simulation output values at a kth sampling point at a public connection point of an electric power system in an ith group of characteristic data;
s3.4, repeating the step S3.3 until identification of the equivalent model key parameters corresponding to 900 groups of feature data in the modeling feature data set is completed;
s4, clustering the characteristic data sets;
s4.1, clustering 900 groups of feature data in the modeling feature data set based on a k-means++ algorithm to distinguish different running states of the micro-grid, and clustering the feature data with similar dynamic characteristics into a class, wherein an input feature matrix of the k-means++ algorithm is as follows:
wherein ,Qi,0 Is Q i Corresponding steady state value, P i,max and Qi,max The maximum values of active power and reactive power are obtained by measuring the public connection point of the system under the fault condition;
s4.2, determining the optimal clustering number;
based on Silhouette criterion, calculating the optimal cluster number index of the k-means++ algorithm as follows:
wherein n is the number of clusters, a k The average distance between the sample k and the rest samples in the same cluster; b k The minimum value of the sample distance between the sample k and the samples in the non-same cluster;
taking the number N of clusters corresponding to the maximum value of kappa as the optimal cluster number, and recording as N 2 Thereby N is arranged 1 Group feature data clustering as N 2 A set of group feature data;
in this embodiment, as shown in fig. 4, it is most appropriate to group 900 sets of feature data into 7 types. Thus, 7 typical operating states can be used to classify the modeled feature dataset, and the clustering results for the feature data are shown in Table 6.
Table 6 is a table of clustering results for the modeled feature dataset.
Feature data set | Feature data | Feature dataCollection set | Feature data | Feature data set | |
1 | 362 | 4 | 57 | 7 | 64 |
2 | 53 | 5 | 78 | / | / |
3 | 245 | 6 | 41 | / | / |
TABLE 6
S5, generalizing equivalent model parameters;
s5.1, setting an input matrix for long-term and short-term memory network training as follows:
setting an output matrix of long-term and short-term memory network training as follows:
wherein ,is->The number of feature data sets in the set of feature data sets; />Is->Group feature data set +.>The mth key parameter of the equivalent model corresponding to the characteristic data,/th key parameter of the equivalent model corresponding to the characteristic data>M is the number of key parameters;
s5.2, repeating the step S5.1 until training of the long-period memory network corresponding to all 7 sets of characteristic data sets is completed, so that 7 long-period memory networks are obtained, and equivalent model parameter generalization is realized.
S6, online matching of equivalent model parameters;
s6.1, establishing a discrimination function of Fisher discrimination criteria based on the 7-group characteristic data set;
s6.1.1 for 7 sets of characteristic data, the thGroup feature data set +.>Observation vector of individual characteristic data->The arrangement is as follows:
s6.1.4, calculating an inter-class divergence matrix as follows:
s6.1.5, constructing a discriminant function generalized feature matrix is as follows:
w b c=λw a c
wherein lambda is the eigenvalue of the generalized eigenvalue matrix of the discriminant function, and c is the corresponding eigenvector;
s6.1.6 the eigenvalues λ are arranged in descending order of magnitude, i.e. λ 1 ≥λ 2 ≥…≥λ m Not less than 0, and its correspondent characteristic vector is c 1 ,c 2 ,…,c m Wherein m is the number of characteristic values, and m is calculated before 0 The cumulative contribution rate of the individual eigenvalues is:
current m 0 The cumulative contribution rate of the individual characteristic values is greater than a threshold epsilon 2 When using the first m 0 The individual feature vectors construct a discriminant function as:
wherein l=1, 2, …, m 0 。
S6.2, judging a characteristic data set to which the new characteristic data belongs according to a judging criterion for the newly acquired characteristic data, and then selecting a corresponding long-period and short-period memory network, wherein the specific process is as follows:
s6.2.1, based on the established discriminant function, calculating a central value matrix of 7 groups of characteristic data sets as follows:
wherein , representing the +.sup.th calculated from the first discriminant function>A central value of the set of group feature data;
s6.2.2, sample observation vector x of the new feature data is set as follows:
s6.2.3, calculating the distance from the sample to the central value matrix of each group of characteristic data set is as follows:
S6.3, calling a long-term memory network, wherein the input of the long-term memory network is as follows:
the output of the long-short-period memory network is the equivalent parameter of the equivalent model.
In order to verify the effectiveness of the method, a traditional micro-grid equivalent modeling method is introduced as a control group, and the method is set as follows: and (3) performing simplified modeling on the micro-grid by adopting the equivalent model shown in fig. 1, and setting the equivalent model parameters as statistical means of equivalent model parameters corresponding to 900 groups of characteristic data in the modeling characteristic data set. In this embodiment, the equivalent model corresponding to the 2 sets of feature data is randomly selected from the second set of feature database, and the active power at the common connection point of the system is obtained through simulation as shown in fig. 5 and fig. 6.
As can be seen from fig. 5 and fig. 6, compared with the conventional method, the method of the present invention can more accurately characterize the dynamic characteristics of the system, and has stronger robustness. Simulation results show that the robustness of the equivalent model of the micro-grid is remarkably improved by the method.
While the foregoing describes illustrative embodiments of the present invention to facilitate an understanding of the present invention by those skilled in the art, it should be understood that the present invention is not limited to the scope of the embodiments, but is to be construed as protected by the accompanying claims insofar as various changes are within the spirit and scope of the present invention as defined and defined by the appended claims.
Claims (7)
1. The utility model relates to a micro-grid dynamic equivalent modeling method considering model robustness, which is characterized by comprising the following steps:
(1) Establishing an operation characteristic database of the micro-grid;
recording N at power system common connection point under different external fault conditions 1 And the transient characteristic data are set and serve as an operation characteristic database of the micro-grid, wherein each set of characteristic data comprises the following information:
[P i Q i S i C i ]
wherein i=1, 2, …, N 1 ,N 1 The number of the characteristic data sets; p (P) i and Qi Active power and reactive power at the public connection point of the power system recorded in the ith group of characteristic data are respectively recorded; p (P) SGi The active power sum of all synchronous power supplies in the micro-grid in the ith group of characteristic data; p (P) VSCi The active power sum of all distributed power sources which are connected with the grid through the converter in the micro-grid in the ith group of characteristic data; p (P) i,0 Is P i The corresponding steady state value; s is S i and Ci The method comprises the steps that the proportion of the total active output of a source distribution network occupied by all synchronous power supplies and all distributed power supplies connected through a converter in the ith group of characteristic data is respectively expressed;
(2) Establishing a micro-grid equivalent model;
the micro-grid is simplified and modeled into a static load model, a synchronous generator model and a voltage source type converter model which are connected in parallel at a common connection point of an electric power system; meanwhile, a line impedance model is connected in series between the public connection point of the electric power system, the parallel synchronous generator model and the voltage source converter model;
the output active power and reactive power of the micro-grid equivalent model at the public connection point of the power system are as follows:
P=P eq,t +I·R eq,t +P Gen,t +P VSC,t
Q=Q eq,t +I·X eq,t +Q Gen,t +Q VSC,t
wherein P and Q are respectively the output active power and reactive power of the micro-grid equivalent model at the public connection point of the power system, and P eq,t and Qeq,t Respectively equivalent active power and reactive power of static load, R eq,t and Xeq,t Equivalent impedance and equivalent reactance of the line impedance, P Gen,t and QGen,t Equivalent active power and reactive power of synchronous generator, P VSC,t and QVSC,t Respectively equivalent active power and reactive power of the voltage source type converter, wherein I is equivalent current flowing through a line impedance model;
(3) Identifying equivalent model parameters based on key parameter screening;
(3.1) calculating a track sensitivity value Ts of the equivalent model parameters of the micro-grid:
wherein ,θj A j-th parameter representing the equivalent model of the micro-grid; θ j0 Represents θ j K is the number of active power sampling points; p (P) 0 Representing steady state value of active power, delta theta j Representing the parameter θ j Offset from the standard value;
(3.2) setting a key parameter selection threshold epsilon, setting parameters with track sensitivity values larger than the threshold as key parameters, and setting other parameters as non-key parameters;
(3.3) identifying key parameters;
setting non-key parameters of the equivalent model of the micro-grid as theoretical values, identifying the key parameters by adopting a particle swarm algorithm, wherein the objective function of the parameter identification is as follows:
wherein , and />The root mean square error and delta P of the active power and reactive power measured values and the model simulation output value at the public connection point of the power system in the ith group of characteristic data are respectively i (k)、ΔQ i (k) Respectively obtaining differences between active power and reactive power measured values and model simulation output values at a kth sampling point at a public connection point of an electric power system in an ith group of characteristic data;
(3.4) identifying equivalent model parameters by utilizing each group of characteristic data recorded by the micro-grid operation characteristic database to obtain N 1 Identification results of the equivalent model key parameters corresponding to the group feature data;
(4) Clustering the feature data sets;
(4.1) operating N in characteristic database of micro-grid based on k-means++ algorithm 1 Clustering the group feature data, and clustering the feature data with similar dynamic characteristics into one type, wherein an input feature matrix of a k-means++ algorithm is as follows:
wherein ,Qi,0 Is Q i Corresponding steady state value, P i,max and Qi,max The maximum values of active power and reactive power are measured by the public connection point of the power system under the fault condition;
(4.2) determining an optimal cluster number;
based on Silhouette criterion, calculating the optimal cluster number index of the k-means++ algorithm as follows:
wherein n is the number of clusters, a k The average distance between the sample k and the rest samples in the same cluster; b k The minimum value of the sample distance between the sample k and the samples in the non-same cluster;
taking the number N of clusters corresponding to the maximum value of kappa as the optimal cluster number, and recording as N 2 Thereby N is arranged 1 Group feature data clustering as N 2 A set of group feature data;
(5) Generalizing equivalent model parameters;
(5.1) setting an input matrix for long-term and short-term memory network training as follows:
setting an output matrix of long-term and short-term memory network training as follows:
wherein ,is->The number of feature data sets in the set of feature data sets; />Is->Group feature data set ofThe mth key parameter of the equivalent model corresponding to the characteristic data,/th key parameter of the equivalent model corresponding to the characteristic data>N 2 M is the number of key parameters;
(5.2) repeating the step (5.1), and training a long-short-period memory network through each group of characteristic data set to realize the generalization of equivalent model parameters;
(6) On-line matching of equivalent model parameters;
(6.1), N-based 2 A set of characteristic data sets, and a discriminant function of Fisher discriminant criteria is established;
(6.2) judging a characteristic data set to which the new characteristic data belongs according to a judging criterion for the new characteristic data, and then selecting a corresponding long-period and short-period memory network;
(6.3) invoking a long-term memory network, the input of the long-term memory network being [ P ] 0 Q 0 P max Q max SC]The output of the long-short-period memory network is the equivalent parameter of the equivalent model, wherein P 0 and Q0 Steady state values corresponding to active power and reactive power of the new characteristic data, P max and Qmax The maximum values of active power and reactive power are measured at the public connection point of the power system through the new characteristic data under the fault condition respectively; s and C respectively represent the proportion of the total active output of the source distribution network occupied by all synchronous power supplies and all distributed power supplies connected through the converter in the new characteristic data.
2. The method for modeling dynamic equivalence of a micro-grid taking model robustness into consideration as defined in claim 1, wherein the mathematical model of the static load model is:
L 1 =P eq,t +jQ eq,t
where j is the imaginary part.
3. The method for modeling dynamic equivalence of a micro-grid taking model robustness into consideration as defined in claim 1, wherein the mathematical model of the line impedance model is:
L 2 =R eq,t +jX eq,t 。
4. the method for modeling dynamic equivalence of a micro-grid taking model robustness into consideration as defined in claim 1, wherein the mathematical model of the voltage source converter model is:
wherein ,Kp,t 、K q,t 、K pc,t and Kqc,t Gain coefficients, T, of the voltage source converters respectively p,t 、T q,t 、T pc,t and Tqc,t Respectively the time constants, m of the voltage source type converter d and mq Duty cycle of d-axis and q-axis, P ref and Qref The reference active power and the reference reactive power, respectively, s represents the laplace operator.
5. The micro-grid dynamic equivalent modeling method considering model robustness according to claim 1, wherein the synchronous generator model comprises a synchronous generator body model, a synchronous generator speed regulator model and a synchronous generator excitation model;
the mathematical expression of the synchronous generator body model is as follows:
wherein ,Xd,t and Xq,t D-axis and q-axis reactance, X' d,t and X′q,t Transient reactance of d-axis and q-axis, T' d0,t and T′q0,t Open-loop transient time constants of d axis and q axis respectively, H t and Dt Rotor inertia and damping factor, respectively; e's' d and E′q Transient potentials of d-axis and q-axis, E f For excitation potential, i d and iq The currents of the d axis and the q axis are respectively, delta is the included angle between the q axis and the reference axis of the synchronous generator, omega is the rotating speed of the rotor, and M m and Me Respectively a mechanical torque and an electromagnetic torque;
the mathematical expression of the synchronous generator speed regulator model is:
wherein ,Ptur To output active power omega for speed regulator R To reference rotor speed, K d,t To synchronize the gain coefficient of the generator, T s,t and T0,t U is the time constant of the speed regulator 0 S represents a Laplacian operator for an opening reference value;
the mathematical expression of the excitation model of the synchronous generator is as follows:
wherein ,T1,t 、T 2,t 、T 3,t 、T 4,t 、T r,t and Ta,t Respectively the time constant, K of the synchronous generator t 、K a,t and Kc,t Respectively are excitation gain coefficients E fd and Ifd Respectively the exciting voltage and exciting current, V ref For reference voltage u c To measure the voltage.
6. The method for modeling dynamic equivalence of a micro-grid with consideration of model robustness according to claim 1, wherein in the step (6.1), the method is based on N 2 The method for establishing the Fisher discriminant function by the group characteristic data set comprises the following steps:
6.1 For N) 2 Group characteristic data set, itemGroup feature data set +.>Observation vector of individual characteristic data->The settings were as follows:
6.4 Calculating an inter-class divergence matrix as:
6.5 Constructing a discriminant function generalized feature matrix as follows:
w b c=λw a c
wherein lambda is the eigenvalue of the generalized eigenvalue matrix of the discriminant function, and c is the corresponding eigenvector;
6.6 The eigenvalues λ are arranged in descending order of magnitude, i.e. λ 1 ≥λ 2 ≥…≥λ m Not less than 0, and its correspondent characteristic vector is c 1 ,c 2 ,…,c m Wherein m is the number of characteristic values, and m is calculated before 0 The cumulative contribution rate of the individual eigenvalues is:
current m 0 The cumulative contribution rate of the individual characteristic values is greater than a threshold epsilon 2 When using the first m 0 The individual feature vectors construct a discriminant function as:
wherein l=1, 2, …, m 0 。
7. The method for modeling dynamic equivalence of a micro-grid with consideration of model robustness according to claim 1, wherein in the step (6.2), for a new feature data, the method for judging the feature data set to which the new feature data belongs according to the sample observation vector and the established Fisher criterion is as follows:
7.1 Based on the established discriminant function, calculate N 2 The central value matrix of the group characteristic data set is as follows:
wherein , representing the +.sup.th calculated from the first discriminant function>A central value of the set of group feature data;
7.2 For a new feature data, its sample observation vector x is:
x=[P 0 Q 0 P max Q max S C]
7.3 Calculating the distance from the sample to the central value matrix of each group of characteristic data set is as follows:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310130423.4A CN116131261A (en) | 2023-02-17 | 2023-02-17 | Micro-grid dynamic equivalent modeling method considering model robustness |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310130423.4A CN116131261A (en) | 2023-02-17 | 2023-02-17 | Micro-grid dynamic equivalent modeling method considering model robustness |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116131261A true CN116131261A (en) | 2023-05-16 |
Family
ID=86311573
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310130423.4A Pending CN116131261A (en) | 2023-02-17 | 2023-02-17 | Micro-grid dynamic equivalent modeling method considering model robustness |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116131261A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117559518A (en) * | 2023-10-16 | 2024-02-13 | 国家电网有限公司华东分部 | Equivalent model construction method and device for wind power generation system |
-
2023
- 2023-02-17 CN CN202310130423.4A patent/CN116131261A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117559518A (en) * | 2023-10-16 | 2024-02-13 | 国家电网有限公司华东分部 | Equivalent model construction method and device for wind power generation system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103887815B (en) | Based on wind energy turbine set parameter identification and the Dynamic Equivalence of service data | |
Wu et al. | A data mining approach combining $ K $-Means clustering with bagging neural network for short-term wind power forecasting | |
CN102074955B (en) | Method based on knowledge discovery technology for stability assessment and control of electric system | |
Zou et al. | A survey of dynamic equivalent modeling for wind farm | |
CN111914486B (en) | Power system transient stability evaluation method based on graph attention network | |
CN108336739A (en) | A kind of Probabilistic Load Flow on-line calculation method based on RBF neural | |
Yi et al. | An integrated model-driven and data-driven method for on-line prediction of transient stability of power system with wind power generation | |
Chen et al. | Data-driven robust state estimation through off-line learning and on-line matching | |
CN116131261A (en) | Micro-grid dynamic equivalent modeling method considering model robustness | |
CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
CN110705831A (en) | Power angle instability mode pre-judgment model construction method after power system fault and application thereof | |
Lin et al. | Parameters identification of reduced governor system model for diesel‐engine generator by using hybrid particle swarm optimisation | |
Vasudevarao et al. | Methodology for creating composite standard load profiles based on real load profile analysis | |
CN103632314A (en) | Probability statistics-based method for modeling generalized node characteristics | |
CN111861256B (en) | Active power distribution network reconstruction decision method and system | |
CN105701265A (en) | Double-fed wind generator modeling method and apparatus | |
Wang et al. | Cluster division in wind farm through ensemble modelling | |
Chenxu et al. | An improved cumulant method for probabilistic load flow calculation | |
Ramirez-Gonzalez et al. | Power System Inertia Estimation Using A Residual Neural Network Based Approach | |
Wang et al. | Multistage Parameter Identification Featured Generic Wind Farm Dynamic Equivalent Modeling | |
CN112290538A (en) | Load model parameter online correction method based on aggregation-identification double-layer framework | |
Yu et al. | Probabilistic stability of small disturbance in wind power system based on a variational Bayes and Lyapunov theory using PMU data | |
CN116257998A (en) | Active power distribution network dynamic equivalent modeling method based on time-varying parameter identification | |
Zhang et al. | Derivation of load model parameters using improved genetic algorithm | |
CN109599897A (en) | A kind of power-less compensation control method considering offshore wind farm fluctuation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |