CN116388232A - Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium - Google Patents
Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116388232A CN116388232A CN202310652031.4A CN202310652031A CN116388232A CN 116388232 A CN116388232 A CN 116388232A CN 202310652031 A CN202310652031 A CN 202310652031A CN 116388232 A CN116388232 A CN 116388232A
- Authority
- CN
- China
- Prior art keywords
- inertia control
- neural network
- wind power
- deep neural
- comprehensive
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 238000013528 artificial neural network Methods 0.000 claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 241001529936 Murinae Species 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 33
- 241000699666 Mus <mouse, genus> Species 0.000 claims description 31
- 238000003062 neural network model Methods 0.000 claims description 26
- 230000008569 process Effects 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 23
- 210000002569 neuron Anatomy 0.000 claims description 19
- 230000008859 change Effects 0.000 claims description 18
- 230000008602 contraction Effects 0.000 claims description 15
- 230000000694 effects Effects 0.000 claims description 15
- 241000699670 Mus sp. Species 0.000 claims description 11
- 238000007665 sagging Methods 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 238000011161 development Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 101100113065 Mus musculus Cfi gene Proteins 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 abstract description 2
- 238000011217 control strategy Methods 0.000 description 10
- 230000004044 response Effects 0.000 description 9
- 230000006399 behavior Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000011084 recovery Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000010248 power generation Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000001360 synchronised effect Effects 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000035699 permeability Effects 0.000 description 2
- 206010001488 Aggression Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000016571 aggressive behavior Effects 0.000 description 1
- 208000012761 aggressive behavior Diseases 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- 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
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- 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
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- 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
- H02J3/24—Arrangements for preventing or reducing oscillations of power in networks
- H02J3/241—The oscillation concerning frequency
-
- 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
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Power Engineering (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a wind power frequency modulation integrated inertia control method, a system, electronic equipment and a storage medium, wherein the method comprises the following steps: establishing a comprehensive inertia control model; obtaining optimal comprehensive inertia control parameters by using a murine optimization algorithm and generating a data set; the automatic encoder extracts the characteristics of the data set; and the deep neural network learns data characteristics to generate comprehensive inertia control parameters corresponding to different wind power scenes and inputs the comprehensive inertia control parameters into the comprehensive inertia control model to obtain a wind power frequency modulation comprehensive inertia control scheme. The invention can rapidly and accurately provide the optimal wind power frequency modulation comprehensive inertia control scheme under the current system when facing the situation that the power system is unbalanced, considers the influence of different wind speeds, wind power duty ratios and load disturbance variables on the frequency characteristic of the system, can meet the requirements of different scenes in the power system, has good universality, effectively improves the quality and the efficiency of wind power frequency modulation, and provides an important scheme for the stable operation and decision-making work of the power system.
Description
Technical Field
The invention relates to the technical field of wind power frequency modulation, in particular to a wind power frequency modulation integrated inertia control method, a system, electronic equipment and a storage medium based on a shrinkage automatic encoder and a deep neural network.
Background
With the large-scale development of renewable energy sources, the inertia level of modern power systems gradually decreases, resulting in a serious decrease in inertial support and frequency regulation capabilities of the systems when disturbance events occur. Wind power generation sets, as a primary representative of renewable energy power generation, must be more effectively involved in power system frequency regulation.
Wind power has been widely used because of its advantages of cleanliness, environmental protection, mature technology, low cost, etc. With the continuous development of wind power generation technology, the wind power permeability of each country is continuously increased. According to the related data, the installed capacity of the Chinese wind power reaches 12.79 percent by the year 2020, and the Chinese wind power has a faster growth rate. The wind turbine generator is generally connected to a power grid through a converter, so that the coupling relation between the rotational speed of the wind turbine generator and the frequency of the power grid is lost; in the maximum power tracking control mode, the wind turbine generator cannot provide a power supporting function when the frequency of the power grid changes. Therefore, as the permeability of wind power is continuously increased, the problems of system inertia reduction, insufficient primary frequency modulation capability and the like are more obvious, and adverse factors are brought to the frequency of a power grid.
Therefore, the wind turbine should have inertia response characteristics and primary frequency modulation capability similar to those of synchronous turbines so as to realize friendly grid connection of wind power. To cope with this problem, the control method adopted at present mainly includes virtual inertia control and droop control.
The virtual inertia control takes the change rate of the system frequency as input to participate in frequency modulation, and is mostly used for supporting transient frequency. The virtual inertia control is used for simulating the inertia characteristic of the synchronous machine, so that when the frequency of the wind driven generator is changed, the active reference instruction is adjusted according to the frequency change rate, and the kinetic energy of the rotor is released or absorbed, so that the aim of participating in the frequency modulation of the system is fulfilled. The principle of virtual inertial control is shown in fig. 2, which can provide a fast active output response when the system is disturbed. In the initial stage of the system frequency response, the frequency deviation is small, and the absolute value of the frequency change rate is large. Thus, a wind turbine with virtual inertial control may provide faster and stronger power support at an initial stage. However, in the end of frequency modulation stage, as the absolute value of the frequency change rate becomes smaller, the output power support is further reduced, which is unfavorable for the recovery of the system frequency. Therefore, the virtual inertia control method is mostly used for supporting transient frequency, can not continuously participate in frequency modulation, and needs to avoid the problem of frequency secondary drop in the rotational speed recovery stage.
The sagging control simulates a primary frequency modulation power-frequency static characteristic curve of the synchronous generator, and is also called proportional control and slope control. It can respond directly to the system frequency offset and can provide continuous support for the frequency offset. The principle of sagging control is shown in fig. 3, wherein the input signal is a system frequency deviation, and the output signal is an output electric power increment of the wind driven generator. The output electric power increment of the wind driven generator linearly changes along with the frequency deviation of the system. The input used for droop control is a frequency deviation signal that can provide stronger power support in the vicinity of the lowest frequency point than virtual inertia control, but the response speed is slower than virtual inertia control, and there is a defect of insufficient power support in the initial frequency modulation.
An Automatic Encoder (AE) is shown in fig. 5, and is composed of an encoder and a decoder. The encoder, decoder has an hidden layer, the input is converted by the encoder into potential representations in the hidden layer, and then the internal representations are converted by the decoder into outputs, which are equivalent to being as close as possible to the input for reconstruction of the input.
The automatic encoder achieves the purpose of feature extraction by minimizing the reconstruction error between the output vector and the input vector, and continuously adjusts the network weight and the threshold value by using a gradient descent algorithm, so as to reduce the reconstruction error. Learning of an automatic encoder may simply preserve the original input data information and may not ensure that an efficient representation of the characteristic information is obtained.
In order to further optimize the frequency modulation effect of wind power participating in frequency modulation, at present, a control strategy for changing a droop coefficient is proposed to improve the self-adaptive capacity of the control strategy, but the solving process is complicated, and the combination of virtual inertia control and droop control is not comprehensively considered. According to the running area of the wind driven generator, a coordinated control strategy of inertia, overspeed and pitch angle control is proposed, but key factors such as wind power duty ratio are not considered. It has also been proposed to optimize the virtual inertia coefficient and the droop control coefficient by using a particle swarm algorithm, and perform wind speed adaptability analysis under three scenes, which proves that the method can significantly optimize the frequency modulation effect, but the method does not significantly shorten the frequency modulation time, and has few scenes and lacks universality. The current comprehensive inertia control frequency modulation effect depends on control parameters to a great extent, and parameters change in different scenes, so that the optimal control of the frequency of the wind turbine generator is difficult to realize.
The existing wind power frequency modulation control strategy also has a gradual inertia control strategy. For example, patent application document with publication number of CN114268129A discloses a wind power frequency modulation gradual inertia control method based on a deep neural network, and patent application document with publication number of CN114696340A discloses a wind power frequency modulation gradual inertia control method based on deep learning; the gradual inertia control strategy provides a step-type power increment after load fluctuation, can effectively prevent the system frequency from decreasing, and ensures the safety of the power grid frequency. The gradual inertia control strategy is usually formulated before frequency modulation, and has no strong correlation with the power grid frequency, the frequency change rate and the rotor rotating speed of the wind driven generator in the whole frequency modulation process, so that the control parameters of the gradual inertia control are correspondingly adjusted for different frequency modulation scenes. In addition, in the stage of recovering wind power, the gradual inertia control is easy to generate the phenomenon of secondary frequency drop of the power grid and influence the frequency modulation effect because of a frequency modulation process of rapid recovery after step power amplification.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a wind power frequency modulation integrated inertia control method, a system, electronic equipment and a storage medium based on a shrinkage automatic encoder and a deep neural network, so that the purposes of quickly and efficiently obtaining optimal integrated inertia control parameters for frequency modulation under corresponding scenes to obtain optimal control effects, avoiding secondary drop of power grid frequency and improving stability of a power system are achieved when the scenes of different wind speeds, different wind power duty ratios and different load disturbance amounts are combined.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a wind power frequency modulation integrated inertia control method based on a shrinkage automatic encoder and a deep neural network comprises the following steps:
step 1: introducing frequency deviation and frequency change rate into an active control link of the wind driven generator simultaneously, and establishing a comprehensive inertia control model;
step 2: obtaining optimal comprehensive inertia control parameters by using a murine optimization algorithm, and forming a comprehensive inertia control parameter data set by the optimal comprehensive inertia control parameters;
step 3: extracting features of the integrated inertial control parameter dataset using a contraction automatic encoder;
step 4: learning the characteristics of the comprehensive inertia control parameter data set by using the deep neural network to obtain a trained deep neural network model;
step 5: when different wind power scenes are met, the wind speed, the wind power duty ratio and the load disturbance quantity are input into a trained deep neural network model, the appropriate comprehensive inertia control parameters corresponding to the different wind power scenes are obtained, the comprehensive inertia control parameters output by the deep neural network model are input into a comprehensive inertia control model, and the comprehensive inertia control model outputs signals to adjust the frequency of a power grid.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, in step 1, the calculation formula of the additional active reference value of the integrated inertial control model is as follows:
in the method, in the process of the invention,additional active reference value for integrated inertial control model,/->For inertial control parameters->For droop control parameters, +.>Representing the system frequency>Representing the running time of the wind driven generator;
by varying inertial control parametersAnd droop control parameter->Value change of (2) additional active reference value +.>。
Further, in step 2, the integrated inertial control parameters include inertial control parametersAnd droop control parameter->The mouse group optimization algorithm specifically comprises the following steps:
step 2.1: initializing a mouse population, defining a comprehensive inertia control parameter as a mouse, and defining a global optimal solution as a prey;
Step 2.3: calculating exploration parametersAAnd development parametersCThe formula is as follows:
in the method, in the process of the invention,Nfor the maximum number of iterations to be performed,nfor the current number of iterations,Ris [1,5 ]]Random numbers in the range of the random numbers,is [0,1 ]]Random numbers within a range;
step 2.4: the mouse chases the prey and calculates the position of the preyThe formula is as follows:
in the method, in the process of the invention,is->Only the mouse is at the firstnThe position of the iteration, ++>Is the firstnOptimal mice in the sub-iterative population, +.>Is for the firstiThe location of the prey in the case of mice only;
step 2.5: the formula of the mouse attack prey is as follows:
in the method, in the process of the invention,is->The mouse is at the%>Positions in the second iteration;
step 2.6: checking whether the boundary limit of the search space is exceeded, and correcting the position of the mouse if the boundary limit is exceeded;
step 2.7: judging whether the termination condition is met, if yes, exiting, outputting the optimal comprehensive inertia control parameter, otherwise, returning to the step 2.2.
Further, step 3 includes:
step 3.1: normalizing the integrated inertial control parameter dataset;
step 3.2: dividing the normalized data set into a training set and a testing set;
step 3.3: constructing a shrinkage automatic encoder model, and selecting the number of hidden layers and the number of neurons;
step 3.4: performing layer-by-layer greedy unsupervised pre-training, inputting a training set, and adopting a dropout technology to randomly select part of neurons to disappear from a network of the shrinkage automatic encoder;
step 3.5: performing supervised refinement training, and performing fine adjustment on the network weight and the bias until the iteration times reach a set value to obtain the network weight and the bias with optimal training contraction automatic encoder effect;
step 3.6: and extracting the characteristics of the test set by using a trained shrink automatic encoder.
Further, in step 3.3, the loss function of the puncturing automatic encoder model is:
in the method, in the process of the invention,representing the jacobian matrix of the puncturing automatic encoder,Wfor a weight matrix between the input layer and the hidden layer,Bfor the bias matrix between the input layer and the hidden layer,J() The jacobian matrix is represented by a matrix,Xfor the input normalized integrated inertial control parameter matrix,Ythe extracted feature matrix is obtained; />For shrinking regularization coefficient, ++>Jacobian matrix for implicit layer output with respect to input samples,>is the square of the jacobian matrix Frobenius norm, < >>The value of the input normalized comprehensive inertia control parameter is obtained;
in the method, in the process of the invention,representation encoder NojReduced-dimension samples of the implicit layer neuron outputs, < > and the implicit layer neurons outputs>Indicate->Individual neuronsIs a sample of the input of (a).
Further, step 4 specifically includes:
step 4.1: constructing a deep neural network model, and selecting the number of hidden layers and the number of neurons of each hidden layer;
step 4.2: optimizing an input weight matrix, an implicit layer feature vector and an output weight matrix of the deep neural network model by using an Adam optimization algorithm, and selectively adding regularization items to the implicit layer;
step 4.3: taking the features extracted by the contraction automatic encoder as the input of the deep neural network, and learning the features extracted by the contraction automatic encoder by the deep neural network to obtain a trained deep neural network model;
step 4.4: taking the mean square error as an evaluation index for training the deep neural network model, judging whether the mean square error is not reduced any more, if so, completing training the deep neural network model, otherwise, returning to the step 4.1;
step 4.5: and saving the trained deep neural network model.
The invention also provides a wind power frequency modulation integrated inertia control system based on the shrinkage automatic encoder and the depth neural network, which comprises an integrated inertia control module, a data set generation module, the shrinkage automatic encoder and the depth neural network;
the integrated inertia control module is used for obtaining an additional active reference value of the integrated inertia control model according to the frequency deviation, the frequency change rate, the inertia control parameter and the sagging control parameter, and fusing the additional active reference value with output power of the wind driven generator in a maximum power point tracking mode to output reference power so as to adjust the frequency of a power grid;
the data set generation module obtains optimal comprehensive inertia control parameters by using a mouse group optimization algorithm and generates a comprehensive inertia control parameter data set;
the contraction automatic encoder is used for extracting the characteristics of the comprehensive inertia control parameter data set;
the deep neural network is used for obtaining inertia control parameters and sagging control parameters in different wind power scenes according to different wind power scenes, and inputting the inertia control parameters and sagging control parameters into the comprehensive inertia control model.
The invention also provides an electronic device, comprising: the wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network is realized when the processor executes the computer program.
The invention also provides a computer readable storage medium storing a computer program, wherein the computer program enables a computer to execute the wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network.
The beneficial effects of the invention are as follows:
(1) The effect of the integrated inertial control depends largely on the control parameters, and the optimal parameters in different scenarios are variable; compared with the traditional time domain simulation, the method greatly saves the on-line decision time of frequency control, and has higher precision for adjusting the optimal comprehensive inertia control frequency;
(2) The method provided by the invention considers the influence of different wind speeds, wind power duty ratios and load disturbance variables on the frequency characteristic of the system, can meet the requirements of different scenes in the power system, and has good universality; simulation results verify the effectiveness of the method in different scenes;
(3) According to the method provided by the invention, the CAE-based feature learning is added before the network training, so that effective data features can be learned, hidden information of data is mined, DNN (digital network) is better fitted, and the acquisition speed of network parameters is improved; according to the frequency curve comparison of different parameters, the comprehensive inertia control of the method provided by the invention achieves the optimal frequency adjustment effect;
(4) Compared with the gradual inertia control which does not consider the defects of the frequency and the frequency change rate of the power grid, the input quantity of the comprehensive inertia control strategy is the frequency deviation quantity and the frequency change rate, and the key factors of the frequency and the frequency change rate of the power grid are considered, so that the real-time frequency information of the power grid can be considered, and the comprehensive inertia control strategy has stronger adaptability and better safety compared with the gradual inertia control strategy for different frequency modulation scenes; and because the gradual inertia control has a frequency modulation process of quickly recovering after step power increasing, the phenomenon of secondary frequency drop of the power grid is easy to occur, the frequency modulation effect is influenced, the phenomenon is better avoided by the comprehensive inertia control, and the power grid frequency can be quickly regulated to a safe level.
Drawings
FIG. 1 is a flow chart of the method proposed by the present invention;
FIG. 2 is a schematic diagram of virtual inertial control;
FIG. 3 is a droop control schematic;
FIG. 4 is a schematic diagram of integrated inertial control;
FIG. 5 is a block diagram of an automatic encoder;
FIG. 6 is a block diagram of a deep neural network;
FIG. 7 is an IEEE 9 node test system;
FIG. 8 is a predicted outcome of optimal droop control parameters for wind power frequency modulation integrated inertial control based on CAE and DNN;
FIG. 9 is a predicted result of optimal inertial control parameters for wind power frequency modulation integrated inertial control based on CAE and DNN;
FIG. 10 is a graph of frequency response for different parameters;
fig. 11 is a graph of frequency response at different load disturbances.
Detailed Description
The invention will now be described in further detail with reference to the accompanying drawings.
In an embodiment, the invention provides a wind power frequency modulation integrated inertia control method based on a shrinkage automatic encoder and a deep neural network, and a flow chart of the method is shown in fig. 1, and the method comprises the following steps:
step 1: introducing frequency deviation and frequency change rate into an active control link of the wind driven generator simultaneously, and establishing a comprehensive inertia control model; a schematic diagram of the integrated inertial control model is shown in fig. 4;
the calculation formula of the additional active reference value of the integrated inertial control model is as follows:
in the method, in the process of the invention,additional active reference value for integrated inertial control model,/->For inertial control parameters->For droop control parameters, +.>Representing the system frequency>Representing the running time of the wind driven generator;
by varying inertial control parametersAnd droop control parameter->Value change of (2) additional active reference value +.>Thereby influencing the inertial supporting capacity of the unit and the recovery time of the rotating speed after frequency modulation; the high pass filter only allows transient components of the frequency to pass, and the low pass filter is used to avoid interference of noise during measurement.
The integrated inertial control combines the virtual inertial control and droop control, can provide powerful power support in the initial stage of the system frequency response and the range near the lowest frequency point, and has inertial control parameters in practical applicationAnd droop control parameter->The setting of the integrated inertial control parameters is complex, the external influence is large, and the integrated inertial control parameters need to be scientifically set, so that the integrated lifting effect after the integrated inertial control parameters are combined is realized. The following steps are therefore used to obtain the integrated inertial control parameters adapted to different wind power scenarios.
When the wind speed is low and the actual rotation speed of the wind driven generator is insufficient to reach the rated rotation speed, the wind driven generator normally operates in a maximum power point tracking (maximum power point tracking, MPPT) mode to capture wind energy to the maximum extent and output electric energy, and the rotor rotation speed of the wind driven generator isOutput power of wind driven generator in maximum power point tracking modeP MPPT Additional active reference value +.>Subtracting to obtain reference power->To adjust the grid frequency;
step 2: obtaining optimal comprehensive inertia control parameters by using a murine optimization algorithm, wherein the comprehensive inertia control parameters comprise inertia control parametersAnd droop control parameter->And the optimal comprehensive inertia control parameters form a comprehensive inertia control parameter data set;
the mouse group optimization (Rat Swarm Optimizer, RSO) algorithm is a novel intelligent group optimization algorithm inspired by the group behaviors of mice chasing and attacking hunting, and has the advantages of simple design, less parameters, small calculation amount and excellent performance. Mice are a social collar animal whose behaviour is in many cases very aggressive, which may lead to the death of some other animals. This aggressive behavior of mice is the primary source of inspiration for this algorithm, mainly involving chase and attack on prey. By carrying out mathematical modeling on the pursuit and the attack behaviors of mice, an RSO algorithm is designed.
The mouse group optimization algorithm specifically comprises the following steps:
step 2.1: initializing a mouse population, defining a comprehensive inertia control parameter as a mouse, and defining a global optimal solution as a prey;
Step 2.3: calculating exploration parametersAAnd development parametersCThe formula is as follows:
in the method, in the process of the invention,Nfor the maximum number of iterations to be performed,nfor the current number of iterations,Ris [1,5 ]]Random numbers in the range of the random numbers,is [0,1 ]]Random numbers within a range;
step 2.4: to mathematically define this behavior, it is assumed that the optimal individual in the population knows where the prey is located. Thus, other mice can chase the prey by the optimal individual, thereby updating the current location to obtain a better search location. The behavior of a mouse chase by a prey is described as follows:
in the method, in the process of the invention,is->Only the mouse is at the firstnThe position of the iteration, ++>Is the firstnOptimal mice in the sub-iterative population, +.>Is for the firstiThe location of the prey in the case of mice only;Afor the purpose of the exploration parameters,Cis a development parameter, and the two realize the dynamic balance of the algorithm between local search and global search;
step 2.5: the formula of the mouse attack prey is as follows:
in the method, in the process of the invention,is->The mouse is at the%>Positions in the second iteration;
step 2.6: checking whether the boundary limit of the search space is exceeded, and correcting the position of the mouse if the boundary limit is exceeded;
step 2.7: judging whether the termination condition is met, if yes, exiting, outputting the optimal comprehensive inertia control parameter, otherwise, returning to the step 2.2.
Step 3: extracting features of the integrated inertial control parameter dataset using a contraction automatic encoder; the method specifically comprises the following steps:
step 3.1: normalizing the integrated inertial control parameter dataset;
step 3.2: taking 80% of normalized data set as training set and the rest 20% of normalized data set as test set;
step 3.3: constructing a shrinkage automatic encoder model, and selecting the number of hidden layers and the number of neurons;
the puncturing automatic encoder (contractive autoencoders, CAE) is based on an automatic encoder, and by adding a puncturing regularization term to the loss function, the encoder is forced to learn a feature extraction function with stronger puncturing effect, thereby increasing the robustness when small disturbance appears around the training data set.
The loss function of the shrink automatic encoder model is:
in the method, in the process of the invention,representing the jacobian matrix of the puncturing automatic encoder,Wfor a weight matrix between the input layer and the hidden layer,Bfor the bias matrix between the input layer and the hidden layer,J() The jacobian matrix is represented by a matrix,Xfor the input normalized integrated inertial control parameter matrix,Ythe extracted feature matrix is obtained; />For shrinking regularization coefficient, ++>Jacobian matrix for implicit layer output with respect to input samples,>is the square of the jacobian matrix Frobenius norm, < >>The value of the input normalized comprehensive inertia control parameter is obtained;
in the method, in the process of the invention,representation encoder NojReduced-dimension samples of the implicit layer neuron outputs, < > and the implicit layer neurons outputs>Indicate->Input samples of neurons.
From the loss function, CAE extracts abstract features of the sample by balancing the reconstruction error with the shrinkage regularization term. Shrinking regularization terms makes the CAE-learned function small for the input gradient, while reconstruction errors force CAE to retain complete information. Under the combined action of the two, the gradient of the characteristic extraction function relative to the input is mostly smaller, and only a small part of the gradient is larger. Thus, when the input has small disturbances, the small gradients attenuate the disturbances, thereby improving the CAE's robustness to the small disturbances of the input. The CAE pre-training is a process of continuously learning data characteristics and extracting the characteristics, and provides reasonable initial parameters for the deep neural network.
Step 3.4: the greedy layer by layer is not supervised and pre-trained, a training set is input, a dropout technology is adopted, part of neurons are randomly selected to disappear from a network of the shrinkage automatic encoder, and interaction between specific nodes is reduced;
step 3.5: performing supervised refinement training, namely performing fine adjustment on the network weight and the bias until the iteration times reach a set value to obtain the network weight and the bias with the optimal training contraction automatic encoder effect;
step 3.6: and extracting the characteristics of the test set by using a trained shrink automatic encoder.
Step 4: learning the characteristics of the comprehensive inertia control parameter data set by using the deep neural network to obtain a trained deep neural network model;
the deep neural network (deep neural network, DNN) is a neural network with many hidden layers. The neural network layers inside the DNN can be divided into: an input layer, a hidden layer, and an output layer. Typically the first layer is the input layer, the last layer is the output layer, and the middle layers are all hidden layers. The layers are fully connected, i.e. any neuron in the former layer must be connected with any neuron in the latter layer, and the basic structure is shown in fig. 6.
The step 4 specifically comprises the following steps:
step 4.1: constructing a deep neural network model, and selecting the number of hidden layers and the number of neurons of each hidden layer;
step 4.2: optimizing an input weight matrix, an implicit layer feature vector and an output weight matrix of the deep neural network by using an Adam optimization algorithm, and selectively adding regularization items to the implicit layer;
step 4.3: taking the features extracted by the contraction automatic encoder as the input of the deep neural network, and learning the features extracted by the contraction automatic encoder by the deep neural network to obtain a trained deep neural network model;
step 4.4: taking the mean square error as an evaluation index for training the deep neural network model, judging whether the mean square error is not reduced any more, if so, completing training the deep neural network model, otherwise, returning to the step 4.1;
step 4.5: and storing the trained deep neural network model, and directly using the trained deep neural network model to generate an optimal wind power frequency modulation comprehensive inertia control scheme when a load disturbance event is encountered next time.
Step 5: when different wind power scenes are met, the wind speed, the wind power duty ratio and the load disturbance quantity are input into a trained deep neural network model, the comprehensive inertia control parameters corresponding to the different wind power scenes are obtained, the comprehensive inertia control parameters are input into a comprehensive inertia control model, and the comprehensive inertia control model outputs signals to adjust the frequency of a power grid.
The effect of this example is demonstrated by experimental results below.
The IEEE 9 node system is taken as an example test system, in which a wind turbine model is connected on line L3, as shown in FIG. 7. The studies involved were all found in wind turbines operating at less than rated power simulations. In order to meet the requirements of using the method in different scenes, the influence of wind speed, wind power duty ratio and load disturbance quantity on the frequency characteristic of the system is considered. The integrated inertia control is parameter-set from these three aspects, respectively. In the calculation example, the wind speed is set to be 4m/s to 10m/s, each condition is increased by 1m/s, and 7 conditions are added; the wind power ratio is set from 5% to 60%, each condition is increased by 5%, and the total number is 12; the load disturbance quantity is set from 1.005 to 1.25, and each case is increased by 0.005, and the total number of the cases is 50, and the total number of the cases is 4200. The optimal parameter prediction result of wind power frequency modulation integrated inertia control based on CAE and DNN is shown in figure 8. And adding an effect comparison graph and a frequency curve comparison graph under different load capacity interferences with other frequency modulation control methods in a typical scene. Typical scenes in the system are 6m/s of wind speed, 30% of wind power occupation ratio and 0.05 p.u. of load disturbance quantity. The comparison of the frequency response curves under different parameters and the frequency response curves under different load interferences of the same typical scene is shown in fig. 9.
In another embodiment, the invention provides a wind power frequency modulation integrated inertial control system based on a shrinkage automatic encoder and a depth neural network, which corresponds to the wind power frequency modulation integrated inertial control method based on the shrinkage automatic encoder and the depth neural network provided in the first embodiment, and comprises an integrated inertial control module, a data set generation module, a shrinkage automatic encoder and a depth neural network;
the integrated inertia control module is used for obtaining an additional active reference value of the integrated inertia control model according to the frequency deviation, the frequency change rate, the inertia control parameter and the sagging control parameter, and outputting power in a maximum power point tracking mode of the wind driven generatorP MPPT Fusing and outputting reference power to adjust the frequency of the power grid;
the data set generation module obtains optimal comprehensive inertia control parameters by using a mouse group optimization algorithm and generates a comprehensive inertia control parameter data set;
the contraction automatic encoder is used for extracting the characteristics of the comprehensive inertia control parameter data set;
the deep neural network is used for obtaining inertia control parameters and sagging control parameters in different wind power scenes according to different wind power scenes, and inputting the inertia control parameters and sagging control parameters into the comprehensive inertia control model.
Specifically, the workflow of each module in the system is consistent with each step of the wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network described in the embodiment one, so the description is not repeated.
In another embodiment, the present invention provides an electronic device, including: the wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network is realized according to the first embodiment when the processor executes the computer program.
In another embodiment, the invention provides a computer readable storage medium storing a computer program, the computer program causing a computer to execute the wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network according to the first embodiment.
In the embodiments disclosed herein, a computer storage medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a computer storage medium would include one or more wire-based electrical connections, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above examples, and all technical solutions belonging to the concept of the present invention belong to the protection scope of the present invention. It should be noted that modifications and adaptations to the invention without departing from the principles thereof are intended to be within the scope of the invention as set forth in the following claims.
Claims (9)
1. The wind power frequency modulation integrated inertia control method based on the shrinkage automatic encoder and the deep neural network is characterized by comprising the following steps of:
step 1: introducing frequency deviation and frequency change rate into an active control link of the wind driven generator simultaneously, and establishing a comprehensive inertia control model;
step 2: obtaining optimal comprehensive inertia control parameters by using a murine optimization algorithm, and forming a comprehensive inertia control parameter data set by the optimal comprehensive inertia control parameters;
step 3: extracting features of the integrated inertial control parameter dataset using a contraction automatic encoder;
step 4: learning the characteristics of the comprehensive inertia control parameter data set by using the deep neural network to obtain a trained deep neural network model;
step 5: and inputting the wind speed, the wind power duty ratio and the load disturbance quantity into a trained deep neural network model to obtain comprehensive inertia control parameters corresponding to different wind power scenes, inputting the comprehensive inertia control parameters output by the deep neural network model into a comprehensive inertia control model, and outputting signals by the comprehensive inertia control model to adjust the frequency of a power grid.
2. The wind power frequency modulation integrated inertial control method based on a shrinkage automatic encoder and a deep neural network according to claim 1, wherein in step 1, the calculation formula of the additional active reference value of the integrated inertial control model is as follows:
in the method, in the process of the invention,additional active reference value for integrated inertial control model,/->For inertial control parameters->For droop control parameters, +.>Representing the system frequency>Representing the running time of the wind driven generator;
3. Shrink-based automatic coding as claimed in claim 2The wind power frequency modulation integrated inertia control method of the wind power frequency modulation and depth neural network is characterized in that in the step 2, the integrated inertia control parameters comprise inertia control parametersAnd droop control parameter->The mouse group optimization algorithm specifically comprises the following steps:
step 2.1: initializing a mouse population, defining a comprehensive inertia control parameter as a mouse, and defining a global optimal solution as a prey;
Step 2.3: calculating exploration parametersAAnd development parametersCThe formula is as follows:
in the method, in the process of the invention,Nfor the maximum number of iterations to be performed,nfor the current number of iterations,Ris [1,5 ]]Random numbers in the range of the random numbers,is [0,1 ]]Random numbers within a range;
step 2.4: the mouse chases the prey and calculates the position of the preyThe formula is as follows:
in the method, in the process of the invention,is->Only the mouse is at the firstnThe position of the iteration, ++>Is the firstnOptimal mice in the sub-iterative population, +.>Is for the firstiThe location of the prey in the case of mice only;
step 2.5: the formula of the mouse attack prey is as follows:
in the method, in the process of the invention,is->The mouse is at the%>Positions in the second iteration;
step 2.6: checking whether the boundary limit of the search space is exceeded, and correcting the position of the mouse if the boundary limit is exceeded;
step 2.7: judging whether the termination condition is met, if yes, exiting, outputting the optimal comprehensive inertia control parameter, otherwise, returning to the step 2.2.
4. The wind power frequency modulation integrated inertial control method based on a shrinkage automatic encoder and a deep neural network according to claim 1, wherein step 3 comprises:
step 3.1: normalizing the integrated inertial control parameter dataset;
step 3.2: dividing the normalized data set into a training set and a testing set;
step 3.3: constructing a shrinkage automatic encoder model, and selecting the number of hidden layers and the number of neurons;
step 3.4: performing layer-by-layer greedy unsupervised pre-training, inputting a training set, and adopting a dropout technology to randomly select part of neurons to disappear from a network of the shrinkage automatic encoder;
step 3.5: performing supervised refinement training, and performing fine adjustment on the network weight and the bias until the iteration times reach a set value to obtain the network weight and the bias with optimal training contraction automatic encoder effect;
step 3.6: and extracting the characteristics of the test set by using a trained shrink automatic encoder.
5. The method for wind power frequency modulation integrated inertial control based on a shrinkage automatic encoder and a depth neural network according to claim 4, wherein in step 3.3, the loss function of the shrinkage automatic encoder model is:
in the method, in the process of the invention,representing the jacobian matrix of the puncturing automatic encoder,Wfor a weight matrix between the input layer and the hidden layer,Bfor the bias matrix between the input layer and the hidden layer,J() The jacobian matrix is represented by a matrix,Xfor the input normalized integrated inertial control parameter matrix,Ythe extracted feature matrix is obtained; />For shrinking regularization coefficient, ++>Jacobian matrix for implicit layer output with respect to input samples,>is the square of the jacobian matrix Frobenius norm, < >>The value of the input normalized comprehensive inertia control parameter is obtained;
6. The wind power frequency modulation integrated inertial control method based on the shrinkage automatic encoder and the deep neural network according to claim 1, wherein the step 4 specifically comprises:
step 4.1: constructing a deep neural network model, and selecting the number of hidden layers and the number of neurons of each hidden layer;
step 4.2: optimizing an input weight matrix, an implicit layer feature vector and an output weight matrix of the deep neural network model by using an Adam optimization algorithm, and selectively adding regularization items to the implicit layer;
step 4.3: taking the features extracted by the contraction automatic encoder as the input of the deep neural network, and learning the features extracted by the contraction automatic encoder by the deep neural network to obtain a trained deep neural network model;
step 4.4: taking the mean square error as an evaluation index for training the deep neural network model, judging whether the mean square error is not reduced any more, if so, completing training the deep neural network model, otherwise, returning to the step 4.1;
step 4.5: and saving the trained deep neural network model.
7. The wind power frequency modulation integrated inertial control system based on the shrinkage automatic encoder and the deep neural network is characterized by comprising an integrated inertial control module, a data set generation module, the shrinkage automatic encoder and the deep neural network;
the integrated inertia control module is used for obtaining an additional active reference value of the integrated inertia control model according to the frequency deviation, the frequency change rate, the inertia control parameter and the sagging control parameter, and fusing the additional active reference value with output power of the wind driven generator in a maximum power point tracking mode to output reference power so as to adjust the frequency of a power grid;
the data set generation module obtains optimal comprehensive inertia control parameters by using a mouse group optimization algorithm and generates a comprehensive inertia control parameter data set;
the contraction automatic encoder is used for extracting the characteristics of the comprehensive inertia control parameter data set;
the deep neural network is used for obtaining inertia control parameters and sagging control parameters corresponding to different wind power scenes according to different wind power scenes, and inputting the inertia control parameters and sagging control parameters into the comprehensive inertia control model.
8. An electronic device, comprising: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the wind power frequency modulation integrated inertial control method based on a systolic automatic encoder and a deep neural network as claimed in any one of claims 1 to 6 when the computer program is executed.
9. A computer-readable storage medium, characterized by: a computer program is stored which causes a computer to execute the wind power frequency modulation integrated inertia control method based on a shrinkage automatic encoder and a deep neural network as claimed in any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310652031.4A CN116388232B (en) | 2023-06-05 | 2023-06-05 | Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310652031.4A CN116388232B (en) | 2023-06-05 | 2023-06-05 | Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116388232A true CN116388232A (en) | 2023-07-04 |
CN116388232B CN116388232B (en) | 2023-08-25 |
Family
ID=86967904
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310652031.4A Active CN116388232B (en) | 2023-06-05 | 2023-06-05 | Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116388232B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117853827A (en) * | 2024-03-07 | 2024-04-09 | 安徽省大气探测技术保障中心 | Sampling pump working state operation monitoring system and method for atmospheric chamber gas monitoring |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111864813A (en) * | 2020-06-23 | 2020-10-30 | 国网辽宁省电力有限公司电力科学研究院 | Wind/thermal power combined frequency control method based on virtual weight coefficient |
CN112183641A (en) * | 2020-09-29 | 2021-01-05 | 国网四川省电力公司经济技术研究院 | Transient frequency stability assessment method and system integrating prediction-correction deep learning |
CN114336632A (en) * | 2021-12-28 | 2022-04-12 | 西安交通大学 | Method for correcting alternating current power flow based on model information assisted deep learning |
CN114696340A (en) * | 2022-02-11 | 2022-07-01 | 南京理工大学 | Wind power frequency modulation gradual inertia control method based on deep learning |
-
2023
- 2023-06-05 CN CN202310652031.4A patent/CN116388232B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111864813A (en) * | 2020-06-23 | 2020-10-30 | 国网辽宁省电力有限公司电力科学研究院 | Wind/thermal power combined frequency control method based on virtual weight coefficient |
CN112183641A (en) * | 2020-09-29 | 2021-01-05 | 国网四川省电力公司经济技术研究院 | Transient frequency stability assessment method and system integrating prediction-correction deep learning |
CN114336632A (en) * | 2021-12-28 | 2022-04-12 | 西安交通大学 | Method for correcting alternating current power flow based on model information assisted deep learning |
CN114696340A (en) * | 2022-02-11 | 2022-07-01 | 南京理工大学 | Wind power frequency modulation gradual inertia control method based on deep learning |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117853827A (en) * | 2024-03-07 | 2024-04-09 | 安徽省大气探测技术保障中心 | Sampling pump working state operation monitoring system and method for atmospheric chamber gas monitoring |
CN117853827B (en) * | 2024-03-07 | 2024-05-14 | 安徽省大气探测技术保障中心 | Sampling pump working state operation monitoring system and method for atmospheric chamber gas monitoring |
Also Published As
Publication number | Publication date |
---|---|
CN116388232B (en) | 2023-08-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wu et al. | A data mining approach combining $ K $-means clustering with bagging neural network for short-term wind power forecasting | |
CN116388232B (en) | Wind power frequency modulation integrated inertia control method, system, electronic equipment and storage medium | |
Zhang et al. | Load shedding scheme with deep reinforcement learning to improve short-term voltage stability | |
CN111478314A (en) | Transient stability assessment method for power system | |
WO2024113835A1 (en) | Method and apparatus for generating fan model | |
CN111898212A (en) | Impeller mechanical profile design optimization method based on BezierGAN and Bayesian optimization | |
Sundarrajan et al. | Open-loop control co-design of floating offshore wind turbines using linear parameter-varying models | |
CN114841090A (en) | Wind power plant grouping optimization control method, system, device and medium | |
CN114781692A (en) | Short-term power load prediction method and device and electronic equipment | |
Branlard et al. | A digital twin solution for floating offshore wind turbines validated using a full-scale prototype | |
Farag et al. | Design and implementation of a variable-structure adaptive fuzzy-logic yaw controller for large wind turbines | |
CN113240217A (en) | Photovoltaic power generation prediction method and device based on integrated prediction model | |
Wilkinson et al. | Inductive aerodynamics | |
KR102439311B1 (en) | Coordinated optimization method for optimiztion of wind farm using sparsified wake digraph and apparatus performing the same | |
CN115936303A (en) | Transient voltage safety analysis method based on machine learning model | |
CN114781244A (en) | Grouping and parameter optimization method in wind power plant | |
Bao et al. | Iterative modeling of wind turbine power curve based on least‐square B‐spline approximation | |
CN114139788A (en) | Method and device for predicting power of offshore wind turbine generator | |
CN114298300A (en) | Uncertainty prediction method, uncertainty prediction device, electronic device, and storage medium | |
CN113420492A (en) | Modeling method for frequency response model of wind-solar-fire coupling system based on GAN and GRU neural network | |
CN113569464A (en) | Wind turbine generator oscillation mode prediction method and device based on deep learning network and multi-task learning strategy | |
CN112290539A (en) | Method and system for predicting transient voltage stability margin of power system | |
CN110797906B (en) | Method and device for determining maximum installed capacity of wind power base and storage medium | |
CN113536508A (en) | Method and system for classifying manufacturing network nodes | |
CN113048012A (en) | Wind turbine generator yaw angle identification method and device based on Gaussian mixture model |
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 |