CN114583767B - Data-driven wind power plant frequency modulation response characteristic modeling method and system - Google Patents
Data-driven wind power plant frequency modulation response characteristic modeling method and system Download PDFInfo
- Publication number
- CN114583767B CN114583767B CN202210229805.8A CN202210229805A CN114583767B CN 114583767 B CN114583767 B CN 114583767B CN 202210229805 A CN202210229805 A CN 202210229805A CN 114583767 B CN114583767 B CN 114583767B
- Authority
- CN
- China
- Prior art keywords
- data
- frequency modulation
- power plant
- wind power
- working condition
- 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.)
- Active
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
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
- H02J3/48—Controlling the sharing of the in-phase component
-
- 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
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/381—Dispersed generators
-
- 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]
-
- 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
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/28—The renewable source being wind energy
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/76—Power conversion electric or electronic aspects
Abstract
The invention relates to a data-driven wind power plant frequency modulation response characteristic modeling method and a data-driven wind power plant frequency modulation response characteristic modeling system, wherein the method comprises the following steps: analyzing and preprocessing the measured data of the wind power plant frequency modulation response characteristic based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition; establishing a transfer function model for each working condition according to the processed data, and measuring a gap value between the models by using a gap measurement method to determine a working condition area represented by a nonlinear autoregressive neural network model; and merging the frequency modulation data of the working condition according to the gap value, and training the nonlinear autoregressive neural network model according to the merged data to obtain the trained nonlinear autoregressive neural network model. According to the wind power plant frequency modulation response data evaluation method, the actual frequency modulation response data of the wind power plant are utilized, a modeling scheme capable of well representing the wind power plant grid-connected point frequency modulation response characteristics under different working conditions is designed, and the accuracy and the frequency modulation effect of the wind power plant frequency response evaluation can be improved.
Description
Technical Field
The invention relates to the technical field of wind power plant frequency modulation response characteristics, in particular to a data-driven wind power plant frequency modulation response characteristic modeling method and system.
Background
An important problem in the research of frequency stability in a wind power plant is to establish a model of wind plant frequency modulation response characteristics, and generally, the modeling method is divided into two types, one type is component-based modeling and the other type is data-based modeling. The modeling based on the components usually needs parameters of internal components of the wind turbine and a dynamic mathematical model during operation, and in fact, due to the existence of a confidential protocol, a fan manufacturer cannot provide detailed fan information, so that the component modeling has certain limitation. The modeling based on the wind field operation data focuses on the input and the output of a research object, avoids the physical description of wind turbine components and the acquisition of corresponding parameters, and the model can also achieve satisfactory precision. Therefore, the data-driven modeling method is more and more applied to the modeling process of the wind turbine generator and the wind power plant.
The nonlinear autoregressive neural network model with the external input can well represent a nonlinear time sequence system. The wind power plant grid-connected point frequency modulation response is a time sequence nonlinear process. According to the related available documents, the transfer function method of the existing wind power plant frequency modulation response characteristic modeling method which is most similar to the method is usually based on components and considers more single working conditions, the nonlinear autoregressive neural network is mostly used for predicting wind speed and power of the wind power plant, and the frequency modulation modeling of the wind power plant is rarely applied. It is known that when a frequency modulation command from a grid-connected point is issued, a wind power plant is not necessarily in a maximum power operation state, and if frequency modulation is performed in other active power operation states, a single model may not be characterized. The method can cause the problems that the wind power plant frequency response evaluation performed on a single model is inaccurate, the frequency modulation effect is not ideal and the like.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a data-driven wind power plant frequency modulation response characteristic modeling method and system.
In order to achieve the purpose, the invention provides the following scheme:
a data-driven wind power plant frequency modulation response characteristic modeling method comprises the following steps:
analyzing and preprocessing the measured data of the wind power plant frequency modulation response characteristic based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
establishing a transfer function model for each working condition according to the processed data, and measuring a gap value between the models by using a gap measurement method to determine a working condition area represented by a nonlinear autoregressive neural network model;
and merging the frequency modulation data of the working condition according to the gap value, and training the nonlinear autoregressive neural network model according to the merged data to obtain the trained nonlinear autoregressive neural network model.
Preferably, the step response dynamic performance index solving algorithm is used for analyzing and preprocessing the measured data of the frequency modulation response characteristic of the wind power plant to obtain processed data under various working conditions, and the method comprises the following steps:
dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant, and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
calculating the maximum value of the upper peak value set and the minimum value of the lower peak value set;
calculating an upper error band bound and a lower error band bound based on data points in the dataset;
judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if yes, determining the adjustment time according to the sampling interval;
judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step disturbance, if so, determining overshoot according to the maximum value of the upper peak value set;
a rise time or fall time is determined from the data set.
Preferably, the establishing a transfer function model for each operating condition according to the processed data includes:
respectively constructing an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
setting a numerator order and a denominator order of the transfer function;
and carrying out model identification on the initial transfer function model, and adjusting the numerator order and the denominator order to obtain the optimal transfer function model.
Preferably, the measuring the gap value between the models by using a gap measurement method to determine the operating condition region characterized by the nonlinear autoregressive neural network model includes:
determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
calculating the gap value according to the gap measurement formula; the clearance value is used to determine the operating region.
Preferably, the merging the frequency modulation data of the working condition according to the gap value includes:
calculating the distance between the gap value between the transfer function models under different working conditions and 0 to obtain a first distance;
calculating the distance between the gap value between the transfer function models under different working conditions and 1 to obtain a second distance;
judging whether the first distance is smaller than or equal to the second distance, if so, combining the frequency modulation data of the two working conditions according to time sequence to obtain combined data; if not, selecting the frequency modulation data of any working condition to carry out time sequence combination with the existing frequency modulation data to obtain the combined data.
Preferably, the training the nonlinear autoregressive neural network model according to the merged data to obtain a trained nonlinear autoregressive neural network model includes:
constructing an initial neural network;
training the initial neural network according to the merged data based on a Levenberg-Markquark algorithm;
and testing the trained neural network by using the frequency modulation data of each working condition, and determining the nonlinear autoregressive neural network model according to the test result.
A data-driven wind power plant frequency modulation response characteristic modeling system comprises:
the data processing module is used for analyzing and preprocessing the measured data of the frequency modulation response characteristic of the wind power plant based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
the transfer function building module is used for building a transfer function model for each working condition according to the processed data and measuring a gap value among the models by using a gap measurement method so as to determine a working condition area represented by the nonlinear autoregressive neural network model;
and the neural network modeling module is used for merging the frequency modulation data of the working condition according to the clearance value and training the nonlinear autoregressive neural network model according to the merged data to obtain the trained nonlinear autoregressive neural network model.
Preferably, the data processing module specifically includes:
the working condition division unit is used for dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
the peak value calculating unit is used for acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
a maximum value calculation unit for calculating the maximum value of the upper peak value set and the minimum value of the lower peak value set;
an upper and lower bound calculation unit for calculating an upper bound and a lower bound of an error band based on the data points in the data set;
the first judgment unit is used for judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if the judgment results are yes, determining the adjustment time according to the sampling interval;
the second judgment unit is used for judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step lower disturbance or not, and if yes, determining the overshoot according to the maximum value of the upper peak value set;
a time determination unit for determining a rise time or a fall time from the data set.
Preferably, the transfer function constructing module specifically includes:
the initial model building unit is used for building an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
the order setting unit is used for setting the numerator order and the denominator order of the transfer function;
and the model determining unit is used for carrying out model identification on the initial transfer function model and adjusting the numerator order and the denominator order to obtain the optimal transfer function model.
Preferably, the transfer function constructing module specifically includes:
the formula determining unit is used for determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
a gap calculation unit for calculating the gap value according to the gap measurement formula; the clearance value is used to determine the operating region.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a data-driven wind power plant frequency modulation response characteristic modeling method and system, wherein the method comprises the following steps: analyzing and preprocessing the measured data of the wind power plant frequency modulation response characteristic based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition; establishing a transfer function model for each working condition according to the processed data, and measuring a gap value between the models by using a gap measurement method to determine a working condition area represented by a nonlinear autoregressive neural network model; and combining the frequency modulation data of the working condition according to the gap value, and training the nonlinear autoregressive neural network model according to the combined data to obtain the trained nonlinear autoregressive neural network model. According to the wind power plant frequency modulation response data evaluation method, the actual frequency modulation response data of the wind power plant are utilized, a modeling scheme capable of well representing the wind power plant grid-connected point frequency modulation response characteristics under different working conditions is designed, and the accuracy and the frequency modulation effect of the wind power plant frequency response evaluation can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a flow chart of a method for modeling a frequency modulation response characteristic of a wind farm in an embodiment provided by the invention;
FIG. 2 is a graph comparing the output of a transfer function model with a first output of actual data in an embodiment provided by the present invention;
FIG. 3 is a second comparison of the output of the transfer function model with the actual data in an embodiment provided by the present invention;
FIG. 4 is a third output comparison plot of the output of the transfer function model and actual data in an embodiment provided by the present invention;
FIG. 5 is a graph comparing the output of the transfer function model with a fourth output of actual data in an embodiment provided by the present invention;
FIG. 6 is a block diagram of a non-linear autoregressive neural network with external inputs in an embodiment provided by the present invention;
FIG. 7 is a first result diagram of a multi-condition data-trained nonlinear autoregressive neural network model tested with single-condition frequency interference data in an embodiment provided by the present invention;
FIG. 8 is a second result graph of the frequency interference data test for the single condition of the nonlinear autoregressive neural network model trained with multi-condition data in the embodiment provided by the present invention;
FIG. 9 is a third result diagram of the frequency interference data test of the single condition for the nonlinear autoregressive neural network model trained with multi-condition data in the embodiment provided by the present invention;
FIG. 10 is a fourth result diagram of the nonlinear autoregressive neural network model trained on multi-condition data tested by single-condition frequency interference data in the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
The terms "first," "second," "third," and "fourth," etc. in the description and claims of this application and in the accompanying drawings are used for distinguishing between different objects and not for describing a particular order. Furthermore, the terms "include" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, the inclusion of a list of steps, processes, methods, etc. is not limited to only those steps recited, but may alternatively include additional steps not recited, or may alternatively include additional steps inherent to such processes, methods, articles, or devices.
The invention aims to provide a data-driven wind power plant frequency modulation response characteristic modeling method and system, which can improve the accuracy and frequency modulation effect of wind power plant frequency response evaluation.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method for modeling a frequency modulation response characteristic of a wind farm in an embodiment provided by the present invention, and as shown in fig. 1, the present invention provides a data-driven method for modeling a frequency modulation response characteristic of a wind farm, including:
step 100: analyzing and preprocessing the actually measured data of the frequency modulation response characteristic of the wind power plant based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
step 200: establishing a transfer function model for each working condition according to the processed data, and measuring a gap value between the models by using a gap measurement method to determine a working condition area represented by a nonlinear autoregressive neural network model;
step 300: and combining the frequency modulation data of the working condition according to the gap value, and training the nonlinear autoregressive neural network model according to the combined data to obtain the trained nonlinear autoregressive neural network model.
Preferably, the step 100 comprises:
dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant, and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
calculating the maximum value of the upper peak value set and the minimum value of the lower peak value set;
calculating an upper error band bound and a lower error band bound based on data points in the dataset;
judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if yes, determining the adjustment time according to the sampling interval;
judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step lower disturbance or not, and if yes, determining overshoot according to the maximum value of the upper peak value set;
a rise time or fall time is determined from the data set.
The first step in this embodiment is to analyze and preprocess the measured data of the wind farm frequency modulation response characteristics, and in practice, the index describing the dynamic performance of one system is generally the rise (fall) time t r Adjusting the time t s And overshoot δ%. Wherein t is r For response values from zero to a final value p ∞ Time taken, t s The time required for the response value to settle within an artificially specified error band from zero to δ% is determined by the following equation
Wherein p is ex To adjust the maximum offset of the response value over time.
Because the wind power plant frequency step response cannot be described by a simple first-order or second-order mathematical model, the dynamic performance index of the response cannot be accurately solved by an analytical expression, in order to solve the problem, a step response dynamic performance index solving algorithm based on measured data is provided, and for convenience of understanding, the algorithm is embodied in a pseudo code form, and the steps of the algorithm are as follows:
selecting data in the adjustment time in the index for modeling, and processing the data as follows: at the moment t of wind farm frequency interference 0 Selecting t as the initial time 0 Active power p corresponding to time 0 The active power variation corresponding to each moment in the frequency modulation process of the wind power plant is
Δp i =p i -p 0 (2)
Where i =0,1, \8230;, n, n is the number of data for which a full frequency step event occurred. p is a radical of i Is t i And the active power value of the wind power field in the corresponding measured data at the moment. Taking 50Hz as the standard frequency of the grid connection, the frequency deviation amount corresponding to each moment in the frequency step process is
Δf i =f i -50 (3)
Wherein f is i Is t i And (4) the frequency value of the wind field in the measured data corresponding to the moment. For ease of calculation, Δ f = (Δ f) is defined 0 ,Δf 1 ,...,Δf n ) T In Δ p = (Δ p) 0 ,Δp 1 ,...,Δp n ) T 。
Further, by taking frequency disturbance as an example, the initial active power during frequency modulation of the wind power plant is divided into different working conditions, and according to measured data of the wind power plant, the working conditions are specifically divided into working conditions 1: frequency step up-disturbance under the limit power of 25%; working condition 2: frequency step up-disturbance under the limit power of 50%; working condition 3: frequency step up-disturbance under unlimited power; working condition 4: a composite frequency up-perturbation; according to the algorithm provided by the invention, the wind farm frequency modulation response characteristic indexes under different working conditions can be obtained, as shown in table 1, table 1 is the frequency step up disturbance response characteristic index of the wind farm under each working condition, and the data required by modeling is intercepted according to the descending time in table 1, and it is noted that the working condition 4 is complex frequency disturbance and is not typical step response, so that the performance index obtained has no actual value, the index of the working condition 4 is not given in table 1, and the data in the complete frequency modulation response time is selected according to the modeling data of the working condition 4. After the data required for modeling are selected, the data are preprocessed according to the preprocessing method in the first step of the scheme.
TABLE 1
Preferably, the step 200 comprises:
respectively constructing an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
setting the numerator order and denominator order of the transfer function;
and carrying out model identification on the initial transfer function model, and adjusting the numerator order and the denominator order to obtain the optimal transfer function model.
Preferably, the step 200 further comprises:
determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
calculating the gap value according to the gap measurement formula; the clearance value is used to determine the operating region.
Specifically, in this embodiment, the second step is to establish a transfer function model for each working condition according to the processed data under each working condition and measure the gap value between the models by using a gap measurement method to determine the modeling working condition region, and the specific method is as follows:
the following transfer function model is given by taking the frequency variation as input and the corresponding power variation as output
Where m and k represent the order of the transfer function numerator and denominator, and assume that k ≧ m, j =1,2 8230n, n, represents the jth frequency fluctuation event, θ j =[b 1,j ,…,b k,j ,a 0,j ,…,a m,j ] T Is a vector of discrete transfer function parameters.
Then (4) the formula can be:
ΔP j (z)(1+b 1,j z -1 +…+b k,j z -k )=Δf j (z)(a 0,j +a 1,j z -1 +…+a m,j z -m ) (5)
depending on the nature of the z-transform, (5) can be directly converted to a difference equation as follows:
ΔP j (t)=-b 1,j ΔP j (t-1)-…-b k,j ΔP j (t-k)+a 0,j Δf j (t)+…+a m,j Δf j (t-m) (6)
suppose that the jth frequency fluctuation event has N measured sample points, and let
Y j =[ΔP j (t),…,ΔP j (N)] T ,
Then equation (6) can be expressed from the above matrix as a compact form as follows:
Y j =Ξ j θ j (7)
therefore, the linear least square method can be used for solving theta j I.e. by
In the modeling process, it can be assumed that the orders m and k of the transfer function numerator and denominator are known, as can be seen from the nature of the transfer function: the values of m and k can reflect the extreme value and the zero number of the output vector curve. In the process of analyzing the frequency modulation response characteristics of the wind power plant, the number of poles and zeros of a transfer function is preliminarily judged according to a curve of active power variation in the frequency modulation event process, values of m and k are given, then model parameters are identified, and finally an optimal model is obtained by continuously adjusting the values of m and k.
After the obtained transfer function model, the dynamic characteristic difference between the models is compared by utilizing a clearance measurement method, so that the purpose of representing the frequency modulation response in a certain working condition range by using one model is achieved. If the transfer functions of the two systems are Q 1 And Q 2 The gap between the two systems can be expressed as:
in the formula, G (Q) i ) (i =1,2) represents Q i Corresponding system diagram, orthogonal projection matrixIs defined as:
wherein i =1,2,N i ,D i ∈RH ∞ Can be composed of Q i Normalized right co-prime decomposition, i.e. Q i =N i D i -1 And satisfy(I is an identity matrix, representing a conjugate), the gap metric formula of the two systems obtained from (9) and (10) is:
wherein, P is any Hilbert matrix, g is more than or equal to 0 and less than or equal to 1, which means that the difference of the dynamic characteristics of the two systems is larger, and the closer the g value is to 0, the smaller the difference of the dynamic characteristics of the two systems is. The closer the g value is to 1, the greater the difference in dynamic characteristics.
Further, by using the second step in this embodiment, transfer function models under different operating conditions are obtained, and fitting graphs thereof are shown in fig. 2 to 5, as can be seen from fig. 2 to 5, the transfer function fitting degree is good, and then according to a clearance measurement method, clearance values between the models are obtained and are close to 1, so that data of all operating conditions are merged. Meanwhile, the clearance value also indicates that the transfer function model needs to be modeled respectively aiming at models of different working conditions, and when the working conditions are increased, the workload becomes huge. As can be seen from fig. 5, the transfer function model is not strong for characterizing complex frequency response events.
Preferably, the step 300 comprises:
calculating the distance between the gap value between the transfer function models under different working conditions and 0 to obtain a first distance;
calculating the distance between the gap value between the transfer function models under different working conditions and 1 to obtain a second distance;
judging whether the first distance is smaller than or equal to the second distance, if so, combining the frequency modulation data of the two working conditions according to time sequence to obtain combined data; and if not, selecting the frequency modulation data of any working condition to perform time sequence combination with the existing frequency modulation data to obtain the combined data.
Preferably, the step 300 further comprises:
constructing an initial neural network;
training the initial neural network according to the merged data based on a Levenberg-Markquark algorithm;
and testing the trained neural network by using the frequency modulation data of each working condition, and determining the nonlinear autoregressive neural network model according to a test result.
Specifically, the third step in this embodiment is to merge corresponding data according to the gap value of the transfer function model under each working condition, and construct a nonlinear autoregressive neural network model, where the specific method is as follows:
after the gap value between the models is obtained in the second step, if the gap value between the two working condition transfer function models is close to 1, the difference of the dynamic characteristics is large, and the frequency modulation data of the two working conditions are combined according to time sequence to obtain a data set D u If the gap value is close to 0, the two models have very similar dynamic characteristics, and the frequency modulation data of any working condition and the existing D are selected u Performing time sequence combination to obtain new D u . Data set D completed by merging u Training the neural network, and introducing a nonlinear autoregressive neural network as follows:
the output of the nonlinear autoregressive neural network depends not only on the current input but also on past inputs and outputs, which makes it have a certain memory function. The wind power plant frequency modulation process is dynamic, and the frequency modulation data is also a time sequence, so that the nonlinear autoregressive neural network model of the wind power plant frequency modulation process is as follows:
ΔP(t)=h(ΔP(t-1),ΔP(t-1),…,ΔP(t-d),Δf(t),Δf(t-1),…,Δf(t-d)) (12)
and h (-) is a nonlinear function of the network structure about the delta P (t) and the delta f (t). d is the delay number. The network structure is shown in fig. 6.
The invention trains the nonlinear autoregressive neural network by adopting the Levenberg-Markquark algorithm, which inherits the advantages of the Newton method and the gradient descent method and has high and stable convergence speed. The parameter updating rule is as follows:
in the formula, q n For the network weight of the nth iteration, e (-) is the error vector. H and J are respectively a Jacobian matrix and a Hessian matrix, and the algorithm gives an approximation of the Hessian matrix by using the Jacobian matrix:
H≈ηI+J T J (14)
wherein η and I are the learning coefficient and the identity matrix, respectively.
And (3) in the training process of the Levenberg-Mark algorithm, based on the adjustment eta value, searching the minimum error through iteration, reducing the eta value when the error value is smaller than the last iteration error, and otherwise, increasing the eta value. When eta is close to 0, the algorithm has local convergence similar to a gradient descent method, and when eta is large, the algorithm has global convergence similar to a Gaussian Newton method.
Further, the third step of the scheme is utilized to combine the data of the corresponding working conditions, train the nonlinear autoregressive neural network, and test the trained network model by using the data of the different working conditions to obtain fig. 7 to 10, and it can be known from fig. 7 to 10 that compared with the transfer function model, the nonlinear autoregressive neural network model can simultaneously represent various working conditions, and has good representation capability for the nonlinear stronger frequency modulation response.
The invention also provides a data-driven wind power plant frequency modulation response characteristic modeling system, which comprises:
the data processing module is used for analyzing and preprocessing the measured data of the frequency modulation response characteristic of the wind power plant based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
the transfer function building module is used for building a transfer function model for each working condition according to the processed data and measuring a gap value among the models by using a gap measurement method so as to determine a working condition area represented by the nonlinear autoregressive neural network model;
and the neural network modeling module is used for merging the frequency modulation data of the working condition according to the clearance value and training the nonlinear autoregressive neural network model according to the merged data to obtain the trained nonlinear autoregressive neural network model.
Preferably, the data processing module specifically includes:
the working condition division unit is used for dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
the peak value calculating unit is used for acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
a maximum value calculation unit for calculating the maximum value of the upper peak value set and the minimum value of the lower peak value set;
an upper and lower bound calculation unit for calculating an upper and lower bound of an error band based on the data points in the data set;
the first judgment unit is used for judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if the judgment results are yes, determining the adjustment time according to the sampling interval;
the second judgment unit is used for judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step lower disturbance or not, and if yes, determining the overshoot according to the maximum value of the upper peak value set;
a time determination unit for determining a rise time or a fall time from the data set.
Preferably, the transfer function constructing module specifically includes:
the initial model building unit is used for building an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
the order setting unit is used for setting the numerator order and the denominator order of the transfer function;
and the model determining unit is used for carrying out model identification on the initial transfer function model and adjusting the numerator order and the denominator order to obtain the optimal transfer function model.
Preferably, the transfer function constructing module specifically includes:
the formula determining unit is used for determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
a gap calculation unit for calculating the gap value according to the gap measurement formula; the clearance value is used to determine the operating region.
The invention has the following beneficial effects:
(1) The invention provides a frequency step response characteristic index solving algorithm based on data, which is used for analyzing the frequency modulation response characteristic of a wind power plant and processing data required by modeling.
(2) According to the method, transfer function models of the wind power plant frequency modulation response characteristics under different working conditions are constructed, and a model clearance measurement method is used for measuring the dynamic similarity between the models to determine the working condition area.
(3) According to the method, an NARX neural network model is constructed to represent the frequency modulation response characteristics of the wind power plant under multiple working conditions. The model overcomes the defect that the characterization working condition range capability and the nonlinear strong response characterization capability of a transfer function model are different, and has a good application prospect in the field of wind power plant frequency modulation response characteristic modeling through simulation verification.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the foregoing, the description is not to be taken in a limiting sense.
Claims (2)
1. A data-driven wind power plant frequency modulation response characteristic modeling method is characterized by comprising the following steps:
analyzing and preprocessing the measured data of the wind power plant frequency modulation response characteristic based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
establishing a transfer function model for each working condition according to the processed data, and measuring the clearance value among the models by utilizing a clearance measurement method to determine a working condition area represented by a nonlinear autoregressive neural network model;
merging the frequency modulation data of the working condition according to the gap value, and training the nonlinear autoregressive neural network model according to the merged data to obtain a trained nonlinear autoregressive neural network model;
the method for analyzing and preprocessing the actually measured data of the wind power plant frequency modulation response characteristics based on the step response dynamic performance index solving algorithm to obtain the processed data under each working condition comprises the following steps:
dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant, and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
calculating the maximum value of the upper peak set and the minimum value of the lower peak set;
calculating an upper error band bound and a lower error band bound based on data points in the dataset;
judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if yes, determining the adjustment time according to the sampling interval;
judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step lower disturbance or not, and if yes, determining overshoot according to the maximum value of the upper peak value set;
determining a rise time or a fall time from the data set;
the establishing of the transfer function model for each working condition according to the processed data comprises the following steps:
respectively constructing an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
setting the numerator order and denominator order of the transfer function;
carrying out model identification on the initial function model, and adjusting the numerator order and the denominator order to obtain the optimal transfer function model;
the method for measuring the clearance value among the models by utilizing the clearance measurement method to determine the working condition area characterized by the nonlinear autoregressive neural network model comprises the following steps:
determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
calculating the gap value according to the gap measurement formula; the clearance value is used for determining the working condition area;
the merging of the frequency modulation data of the working conditions according to the gap value comprises the following steps:
calculating the distance between the gap value between the transfer function models under different working conditions and 0 to obtain a first distance;
calculating the distance between the gap value between the transfer function models under different working conditions and 1 to obtain a second distance;
judging whether the first distance is smaller than or equal to the second distance, if so, combining the frequency modulation data of the two working conditions according to time sequence to obtain combined data; if not, selecting the frequency modulation data of any working condition to carry out time sequence combination with the existing frequency modulation data to obtain combined data;
the training of the nonlinear autoregressive neural network model according to the combined data to obtain the trained nonlinear autoregressive neural network model comprises the following steps:
constructing an initial neural network;
training the initial neural network according to the merged data based on a Levenberg-Markquark algorithm;
and testing the trained neural network by using the frequency modulation data of each working condition, and determining the nonlinear autoregressive neural network model according to the test result.
2. A data-driven wind power plant frequency modulation response characteristic modeling system is characterized by comprising:
the data processing module is used for analyzing and preprocessing the measured data of the frequency modulation response characteristic of the wind power plant based on a step response dynamic performance index solving algorithm to obtain processed data under each working condition;
the transfer function building module is used for building a transfer function model for each working condition according to the processed data and measuring a gap value among the models by using a gap measurement method so as to determine a working condition area represented by the nonlinear autoregressive neural network model;
the neural network modeling module is used for merging the frequency modulation data of the working condition according to the gap value and training the nonlinear autoregressive neural network model according to the merged data to obtain a trained nonlinear autoregressive neural network model;
the data processing module specifically comprises:
the working condition division unit is used for dividing the actually measured data of the wind power plant frequency modulation response characteristic according to the initial active power during the frequency modulation of the wind power plant to obtain a plurality of working conditions;
the peak value calculating unit is used for acquiring a power curve of a data set of the measured data of the frequency modulation response characteristic of the wind power plant and calculating an upper peak value set and a lower peak value set of each data point of the power curve;
a maximum value calculation unit for calculating the maximum value of the upper peak value set and the minimum value of the lower peak value set;
an upper and lower bound calculation unit for calculating an upper and lower bound of an error band based on the data points in the data set;
the first judgment unit is used for judging whether the upper bound of the error band is greater than or equal to the maximum value of the upper peak value set or not and whether the lower bound of the error band is greater than or equal to the minimum value of the lower peak value set or not, and if the judgment results are yes, determining the adjustment time according to the sampling interval;
the second judgment unit is used for judging whether the measured data of the wind power plant frequency modulation response characteristic is frequency step lower disturbance or not, and if yes, determining the overshoot according to the maximum value of the upper peak value set;
a time determination unit for determining a rise time or a fall time from the data set;
the transfer function building module specifically includes:
the initial model building unit is used for building an initial function model according to the frequency variation as input and the power variation as output in the processed data under each working condition;
the order setting unit is used for setting the numerator order and the denominator order of the transfer function;
the model determining unit is used for carrying out model identification on the initial function model and adjusting the numerator order and the denominator order to obtain the optimal transfer function model;
the formula determining unit is used for determining a gap measurement formula according to orthogonal projection matrixes of different transfer function models;
a gap calculation unit for calculating the gap value according to the gap measurement formula; the clearance value is used to determine the operating region.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210229805.8A CN114583767B (en) | 2022-03-10 | 2022-03-10 | Data-driven wind power plant frequency modulation response characteristic modeling method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210229805.8A CN114583767B (en) | 2022-03-10 | 2022-03-10 | Data-driven wind power plant frequency modulation response characteristic modeling method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114583767A CN114583767A (en) | 2022-06-03 |
CN114583767B true CN114583767B (en) | 2023-03-17 |
Family
ID=81773321
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210229805.8A Active CN114583767B (en) | 2022-03-10 | 2022-03-10 | Data-driven wind power plant frequency modulation response characteristic modeling method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114583767B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115498660B (en) * | 2022-11-17 | 2023-04-07 | 华北电力大学 | Wind power plant frequency modulation dynamic modeling method and device and electronic equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109004687A (en) * | 2018-08-03 | 2018-12-14 | 山东大学 | The intelligent inertia response control mehtod and system of wind power plant participation power grid frequency modulation |
CN110556842A (en) * | 2019-09-16 | 2019-12-10 | 湖南大学 | Control method of direct-drive wind power plant inductive weak grid-connected subsynchronous oscillation suppression device |
CN110765703A (en) * | 2019-11-04 | 2020-02-07 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Wind power plant aggregation characteristic modeling method |
AU2020102245A4 (en) * | 2019-01-08 | 2020-10-29 | Nanjing Institute Of Technology | A grid hybrid rolling dispatching method considering congestion and energy storage tou price |
CN113837432A (en) * | 2021-08-12 | 2021-12-24 | 华北电力大学 | Power system frequency prediction method driven by physics-data combination |
CN114065460A (en) * | 2020-08-03 | 2022-02-18 | 北京国电智深控制技术有限公司 | Model processing method, storage medium and electronic device in thermal power generation system |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108242819B (en) * | 2016-12-26 | 2021-01-22 | 北京金风科创风电设备有限公司 | Measurement and control device, system and method for wind power plant |
CN110426953B (en) * | 2019-07-18 | 2022-06-10 | 国网山东省电力公司电力科学研究院 | AGC performance evaluation method based on thermal power generating unit power generation model |
CN111525593A (en) * | 2020-04-10 | 2020-08-11 | 中国电力科学研究院有限公司 | Wind power plant primary frequency modulation parameter fitting method and system |
CN111900743B (en) * | 2020-07-28 | 2021-11-16 | 南京东博智慧能源研究院有限公司 | Wind power frequency modulation potential prediction error distribution estimation method |
CN112636366B (en) * | 2020-12-01 | 2023-05-16 | 国家电网有限公司 | Wind power plant dynamic frequency control method based on control process data fitting |
CN112560352A (en) * | 2020-12-24 | 2021-03-26 | 华北电力大学 | System frequency response model modeling method based on AM-LSTM neural network |
CN113708389B (en) * | 2021-09-10 | 2023-08-04 | 国网湖南省电力有限公司 | Wind farm primary frequency modulation model parameter identification method and system based on actual power response |
CN113964884A (en) * | 2021-11-17 | 2022-01-21 | 国家电网有限公司华东分部 | Power grid active frequency regulation and control method based on deep reinforcement learning |
CN113988481B (en) * | 2021-12-23 | 2022-05-03 | 南京鼐威欣信息技术有限公司 | Wind power prediction method based on dynamic matrix prediction control |
-
2022
- 2022-03-10 CN CN202210229805.8A patent/CN114583767B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109004687A (en) * | 2018-08-03 | 2018-12-14 | 山东大学 | The intelligent inertia response control mehtod and system of wind power plant participation power grid frequency modulation |
AU2020102245A4 (en) * | 2019-01-08 | 2020-10-29 | Nanjing Institute Of Technology | A grid hybrid rolling dispatching method considering congestion and energy storage tou price |
CN110556842A (en) * | 2019-09-16 | 2019-12-10 | 湖南大学 | Control method of direct-drive wind power plant inductive weak grid-connected subsynchronous oscillation suppression device |
CN110765703A (en) * | 2019-11-04 | 2020-02-07 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Wind power plant aggregation characteristic modeling method |
CN114065460A (en) * | 2020-08-03 | 2022-02-18 | 北京国电智深控制技术有限公司 | Model processing method, storage medium and electronic device in thermal power generation system |
CN113837432A (en) * | 2021-08-12 | 2021-12-24 | 华北电力大学 | Power system frequency prediction method driven by physics-data combination |
Also Published As
Publication number | Publication date |
---|---|
CN114583767A (en) | 2022-06-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20200327435A1 (en) | Systems and methods for sequential power system model parameter estimation | |
Lindemann et al. | Anomaly detection and prediction in discrete manufacturing based on cooperative LSTM networks | |
CN108920863B (en) | Method for establishing energy consumption estimation model of robot servo system | |
CN100507460C (en) | Dynamic soft measuring and form establishing method base pulse response formwork and parameter optumization | |
CN108155648A (en) | Method for estimating state based on the infinite Extended Kalman filter of adaptive H | |
CN105158723A (en) | Error evaluation system and method for digital electric energy metering system | |
CN110442911B (en) | High-dimensional complex system uncertainty analysis method based on statistical machine learning | |
CN116150897A (en) | Machine tool spindle performance evaluation method and system based on digital twin | |
CN110795780A (en) | XGboost algorithm-based cable-stayed bridge finite element correction method | |
CN114583767B (en) | Data-driven wind power plant frequency modulation response characteristic modeling method and system | |
CN116451514A (en) | Bridge structure state evaluation method and device and electronic equipment | |
CN113281229B (en) | Multi-model self-adaptive atmosphere PM based on small samples 2.5 Concentration prediction method | |
CN114357870A (en) | Metering equipment operation performance prediction analysis method based on local weighted partial least squares | |
CN103279030A (en) | Bayesian framework-based dynamic soft measurement modeling method and device | |
CN110474323B (en) | Method for measuring inertia time constant of power system | |
CN109684749B (en) | Photovoltaic power station equivalent modeling method considering operating characteristics | |
CN106709598B (en) | Voltage stability prediction and judgment method based on single-class samples | |
CN116085245A (en) | Online compressor performance prediction method and system based on OS-ELM | |
Adedotun et al. | Modelling and forecasting climate time series with state-space model | |
CN115561697A (en) | Intelligent ammeter error analysis method | |
CN114970698A (en) | Metering equipment operation performance prediction method based on improved LWPLS | |
CN113627041A (en) | Time-varying dynamic load online modeling method and device based on multi-source data fusion | |
CN108053093A (en) | A kind of k- neighbour's method for diagnosing faults based on the conversion of average influence Value Data | |
Deng et al. | A fractional hilbert transform order optimization algorithm based DE for bearing health monitoring | |
CN113837432A (en) | Power system frequency prediction method driven by physics-data combination |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |