CN113987916A - Non-linear fitting optimal array distribution method based on shadow multiplying power - Google Patents

Non-linear fitting optimal array distribution method based on shadow multiplying power Download PDF

Info

Publication number
CN113987916A
CN113987916A CN202111125125.3A CN202111125125A CN113987916A CN 113987916 A CN113987916 A CN 113987916A CN 202111125125 A CN202111125125 A CN 202111125125A CN 113987916 A CN113987916 A CN 113987916A
Authority
CN
China
Prior art keywords
data
matrix
optimal
data matrix
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111125125.3A
Other languages
Chinese (zh)
Inventor
万露
苏欣
周洲
张敬昂
王元龙
陶岳来
臧藏
施蒋娟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Sinogy Electric Engineering Co Ltd
Original Assignee
China Sinogy Electric Engineering Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Sinogy Electric Engineering Co Ltd filed Critical China Sinogy Electric Engineering Co Ltd
Priority to CN202111125125.3A priority Critical patent/CN113987916A/en
Publication of CN113987916A publication Critical patent/CN113987916A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • 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
    • 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
    • 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
    • Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
    • Y02E40/70—Smart grids as climate change mitigation technology in the energy generation sector
    • 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
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Geometry (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a non-linear fitting optimal array distribution method based on shadow multiplying power, which comprises the following steps of: acquiring a matrix data sample and forming a data matrix according to the data sample; carrying out data preprocessing on the data matrix; extracting characteristic quantity and extending implicit rules of the preprocessed data matrix; inputting the expanded data matrix into a pre-constructed neural network for training; and outputting the optimal array characteristic parameters by the trained neural network. And (3) depending on environmental data, performing a certain amount of event simulation, mining and expanding data characteristic quantity aiming at discrete distributed event data, and performing nonlinear fitting through a trained and tested Kohonen neural network clustering model to output optimal characteristic parameters. The optimal photovoltaic design electricity cost is realized, and the optimal inclination angle, the optimal array spacing, the land control index and the capital investment mode are coordinated and unified.

Description

Non-linear fitting optimal array distribution method based on shadow multiplying power
Technical Field
The invention relates to array layout design, in particular to a non-linear fitting optimal array layout method based on shadow multiplying power.
Background
In the traditional arrangement design, the maximum non-shielding design is carried out on the basis of a 9: 00-15: 00 time period of winter solstice days, and the optimal inclination angle and the optimal arrangement distance are determined to realize the maximum power generation design. Inspired by the theory of zero sum game, after the conditions of the optimal inclination angle and the optimal arrangement spacing are met, the optimal LCOE cost is also limited by factors such as land control indexes, capital investment modes and the like. It is clear how to maximize the best LCOE cost, the traditional linear interpolation scheme cannot meet the requirement of fine design.
Disclosure of Invention
In view of the existing defects, the invention provides a non-linear fitting optimal arrangement method based on shadow multiplying power, which can realize optimal photovoltaic design electricity cost, and enable the optimal inclination angle, the optimal arrangement distance, the land control index and the capital investment mode to be cooperatively unified.
In order to achieve the above purpose, the embodiment of the invention adopts the following technical scheme:
a non-linear fitting optimal arraying method based on shadow multiplying power comprises the following steps:
acquiring a matrix data sample and forming a data matrix according to the data sample;
carrying out data preprocessing on the data matrix;
extracting characteristic quantity and extending implicit rules of the preprocessed data matrix;
inputting the expanded data matrix into a pre-constructed neural network for training;
and outputting the optimal array characteristic parameters by the trained neural network.
In accordance with one aspect of the present invention, the obtaining the arrayed data samples and forming the data matrix from the data samples comprises: acquiring maximum unconcealed day data of winter solstice and spring equinox, group string inclined plane length data, group string length data, inclination angle data, latitude data, bracket interval data, south-north and east-west gradient data and generating capacity data, and establishing a data matrix taking the inclination angle, the arrangement interval and the generating capacity as characteristics.
According to one aspect of the invention, the data preprocessing of the data matrix comprises: and carrying out Weber-Fisher law normalization processing, wavelet decomposition and filter setting and reconstruction on the data matrix.
In accordance with one aspect of the invention, the Weber-Fechner law
Figure BDA0003278600760000021
Indicating that the generating capacity is in direct proportion to the logarithm of the inclination angle and the distance of a certain interval; wherein
Figure BDA0003278600760000022
For evaluating the parameter coefficient, the reference quantity is the theoretically calculated inclination angle, distance and generated energy.
In accordance with one aspect of the invention, the wavelet decomposition and its filter setup and reconstruction performs a function reduction on the data matrix and de-noise processing of the noisy data and forms noise samples.
According to an aspect of the invention, the performing feature quantity extraction and implicit rule extension on the preprocessed data matrix comprises: and performing GAN network learning training on the noise sample and the preprocessed data matrix aiming at the characteristic quantity and implicit rule extension.
In accordance with one aspect of the present invention, the GAN network learning training comprises: the noise sample is input into the generator to generate generated data, the generated data and the preprocessed data matrix are input into the discriminator to be discriminated, and the discrimination result is returned to the generator through the loss function LG and returned to the discriminator through the loss function LD, so that the data matrix is expanded.
According to one aspect of the invention, the inputting the expanded data matrix into the pre-constructed neural network for training comprises: and inputting the expanded data matrix into a Kohonen neural network clustering model for training and testing.
According to one aspect of the invention, inputting the expanded data matrix into the Kohonen neural network clustering model for training and testing comprises: by connecting weights omegaijImplementing input layer feature vectors alphai(i ═ 1, 2, …, m) and competition layer matrix βj(j is 1, 2, …, n) and the weight ω of input layer matrix neuron i and output neuron jijCalculating Euclidean values
Figure BDA0003278600760000023
Minimum Euclidean value djThe corresponding input layer matrix neuron i is a winning neuron, the best matching of the competition layer matrix neuron j is realized by adjusting the winning neuron and the adjacent weight, the similar aggregation is realized by gradual iteration, and a winning connotation matrix gamma is established by the best matching neuron and the adjacent neurons thereofk(k 1, 2, …, n), iterative optimization of the neuron weight coefficients τ is achieved within the context of an inclusion matrix, i.e.
Figure BDA0003278600760000024
ωij＝ωij+ε(αi-ωij)
Wherein posτ，posjTo win over the neuron tau, j positions of the connotation matrix,
Figure BDA0003278600760000025
respectively, a defined threshold and a learning rate.
According to an aspect of the present invention, the outputting the optimal constellation feature parameters by the trained neural network comprises: and outputting the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameter by the trained neural network through nonlinear fitting.
The implementation of the invention has the advantages that: the invention relates to a non-linear fitting optimal arraying method based on shadow multiplying power, which comprises the following steps of: acquiring a matrix data sample and forming a data matrix according to the data sample; carrying out data preprocessing on the data matrix; extracting characteristic quantity and extending implicit rules of the preprocessed data matrix; inputting the expanded data matrix into a pre-constructed neural network for training; and outputting the optimal array characteristic parameters by the trained neural network. And (3) depending on environmental data, performing a certain amount of event simulation, mining and expanding data characteristic quantity aiming at discrete distributed event data, and performing nonlinear fitting through a trained and tested Kohonen neural network clustering model to output optimal characteristic parameters. The optimal photovoltaic design electricity cost is realized, and the optimal inclination angle, the optimal array spacing, the land control index and the capital investment mode are coordinated and unified.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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 that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart of a non-linear fitting optimal arraying method based on shadow multiplying power according to the present invention;
FIG. 2 is a block diagram of implicit rule extended learning training of the GAN network according to the present invention;
fig. 3 is a topological structure diagram of a Kohonen neural network of a non-linear fitting optimal arraying method based on shadow multiplying power according to the present invention.
FIG. 4 is a schematic structural diagram of an optimal arraying system based on shadow multiplying power according to 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 derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The first embodiment is as follows: as shown in fig. 1, fig. 2 and fig. 3, a non-linear fitting optimal arrangement method based on shadow magnification includes the following steps:
s1: acquiring a matrix data sample and forming a data matrix according to the data sample;
in practical application, the obtaining the arrayed data samples and forming the data matrix according to the data samples includes: acquiring maximum unconcealed day data of winter solstice and spring equinox, group string inclined plane length data, group string length data, inclination angle data, latitude data, bracket interval data, south-north and east-west gradient data and generating capacity data, and establishing a data matrix taking the inclination angle, the arrangement interval and the generating capacity as characteristics.
In practical applications, the data sample of this embodiment can be listed as follows:
Figure BDA0003278600760000041
in practical application, winter solstice day and spring minutes day data output of a flat ground array interval D, a mountain land declination angle delta, a sun hour angle omega, a sun altitude angle alpha and a sun azimuth angle beta is realized by adjusting an array inclination angle, flat ground shadows are realized, the distance between the front row and the rear row of a square array is realized, only the south and north direction has a slope, and the distance between the front row and the rear row and the left row and the right row of the east and the north row and the south row and the north row and the east and the west row and the south and the north row have the slope are output, wherein multiple groups of data acquisition is realized by adjusting the array inclination angle, the south and north, the east and west direction slopes, and the generated energy data acquisition is completed by PVsyst software.
S2: carrying out data preprocessing on the data matrix;
in practical applications, the data preprocessing on the data matrix includes: and carrying out Weber-Fisher law normalization processing, wavelet decomposition and filter setting and reconstruction on the data matrix.
In practical application, the Weber-Fisher law
Figure BDA0003278600760000042
Indicating that the generating capacity is in direct proportion to the logarithm of the inclination angle and the distance of a certain interval; wherein
Figure BDA0003278600760000043
For evaluating the parameter coefficient, the reference quantity is the theoretically calculated inclination angle, distance and generated energy.
In practical application, the wavelet decomposition and the filter setting and reconstruction thereof perform function reduction and noise data denoising processing on a data matrix and form a noise sample.
S3: extracting characteristic quantity and extending implicit rules of the preprocessed data matrix;
in practical application, the extracting feature quantity and the extending implicit rule of the preprocessed data matrix include: and performing GAN network learning training on the noise sample and the preprocessed data matrix aiming at the characteristic quantity and implicit rule extension.
In practical application, the GAN network learning training includes: the noise sample is input into the generator to generate generated data, the generated data and the preprocessed data matrix are input into the discriminator to be discriminated, and the discrimination result is returned to the generator through the loss function LG and returned to the discriminator through the loss function LD, so that the data matrix is expanded.
S4: inputting the expanded data matrix into a pre-constructed neural network for training;
in practical application, the training by inputting the expanded data matrix into a pre-constructed neural network includes: and inputting the expanded data matrix into a Kohonen neural network clustering model for training and testing.
In practical application, the training and testing of the expanded data matrix input into the Kohonen neural network clustering model comprises: by connecting weights omegaijImplementing input layer feature vectors alphai(i ═ 1, 2, …, m) and competition layer matrix βj(j＝1，2，…，n) is fully connected, and the weight omega of the input layer matrix neuron i and the output neuron jijCalculating Euclidean values
Figure BDA0003278600760000051
Minimum Euclidean value djThe corresponding input layer matrix neuron i is a winning neuron, the best matching of the competition layer matrix neuron j is realized by adjusting the winning neuron and the adjacent weight, the similar aggregation is realized by gradual iteration, and a winning connotation matrix gamma is established by the best matching neuron and the adjacent neurons thereofk(k 1, 2, …, n), iterative optimization of the neuron weight coefficients τ is achieved within the context of an inclusion matrix, i.e.
Figure BDA0003278600760000052
ωij＝ωij+ε(αi-ωij)
Wherein posτ，posjTo win over the neuron tau, j positions of the connotation matrix,
Figure BDA0003278600760000053
respectively, a defined threshold and a learning rate.
S5: and outputting the optimal array characteristic parameters by the trained neural network.
In practical application, the outputting of the optimal array characteristic parameter by the trained neural network includes: and outputting the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameter by the trained neural network through nonlinear fitting.
In practical application, the method realizes the establishment of a non-linear fitting clustering model of shadow multiplying power by inputting inclination angle, spacing and power generation data samples, using a Weber-Fisher law, an association clustering theory and a Kohonen neural network, and outputting an optimal inclination angle, an optimal array spacing and a maximum power generation characteristic parameter.
In practical application, a certain amount of small sample database is formed by a large amount of simulation experiment data acquisition, the Weber-Fisher law, wavelet decomposition and filter denoising and reconstruction are combined, the Kohonen neural network model of a large data sample set is trained and tested by resisting the internal recessive rule expansion of a generation network, and finally the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameters are output. The scheme of the invention realizes the nonlinear optimization of shadow multiplying power optimization.
Example two: as shown in fig. 1, fig. 2, fig. 3 and fig. 4, an optimal arraying system based on shadow magnification includes:
the acquisition module is used for acquiring the array data samples;
the matrix module is used for forming a data matrix according to the data samples;
in practical application, the acquisition module is connected with the matrix module, the matrix module is connected with the preprocessing module, the preprocessing module is connected with the extraction extension module, and the extraction extension module is connected with the neural network.
In practical application, obtaining the arrayed data samples and forming the data matrix according to the data samples includes: acquiring maximum unconcealed day data of winter solstice and spring equinox, group string inclined plane length data, group string length data, inclination angle data, latitude data, bracket interval data, south-north and east-west gradient data and generating capacity data, and establishing a data matrix taking the inclination angle, the arrangement interval and the generating capacity as characteristics.
In practical applications, the data sample of this embodiment can be listed as follows:
Figure BDA0003278600760000061
in practical application, winter solstice day and spring minutes day data output of a flat ground array interval D, a mountain land declination angle delta, a sun hour angle omega, a sun altitude angle alpha and a sun azimuth angle beta is realized by adjusting an array inclination angle, flat ground shadows are realized, the distance between the front row and the rear row of a square array is realized, only the south and north direction has a slope, and the distance between the front row and the rear row and the left row and the right row of the east and the north row and the south row and the north row and the east and the west row and the south and the north row have the slope are output, wherein multiple groups of data acquisition is realized by adjusting the array inclination angle, the south and north, the east and west direction slopes, and the generated energy data acquisition is completed by PVsyst software.
The preprocessing module is used for preprocessing data of the data matrix;
in practical application, the data preprocessing of the data matrix comprises: and carrying out Weber-Fisher law normalization processing, wavelet decomposition and filter setting and reconstruction on the data matrix.
In practical application, the Weber-Fisher law shows that the generating capacity is in direct proportion to the logarithm of the inclination angle and the distance of a certain interval; wherein, for evaluating parameter coefficients, the reference quantity is the inclination angle, the distance and the generating capacity which are calculated theoretically.
In practical applications, wavelet decomposition and filter setting and reconstruction thereof perform function reduction and denoising of noisy data on a data matrix and form noise samples.
The extraction extension module is used for extracting characteristic quantity and extending implicit rules of the preprocessed data matrix;
in practical application, the extraction extension module comprises a GAN network learning unit, and the GAN network learning unit comprises a generator and an arbiter.
In practical application, the generator receives the noise sample and generates data to be sent to the discriminator, the discriminator outputs a discrimination result, and the discrimination result is returned to the generator through the loss function LG and returned to the discriminator through the loss function LD.
In practical application, the characteristic quantity extraction and implicit rule extension of the preprocessed data matrix comprise: and performing GAN network learning training on the noise sample and the preprocessed data matrix aiming at the characteristic quantity and implicit rule extension.
In practical application, the GAN network learning training comprises: the noise sample is input into the generator to generate generated data, the generated data and the preprocessed data matrix are input into the discriminator to be discriminated, and the discrimination result is returned to the generator through the loss function LG and returned to the discriminator through the loss function LD, so that the data matrix is expanded.
And the neural network is trained by using the expanded data matrix and is used for outputting the optimal array characteristic parameters. In practical application, inputting the expanded data matrix into a pre-constructed neural network for training comprises the following steps: and inputting the expanded data matrix into a Kohonen neural network clustering model for training and testing.
In practical application, the training and testing of the expanded data matrix input into the Kohonen neural network clustering model comprises the following steps: by connecting weights omegaijImplementing input layer feature vectors alphai(i ═ 1, 2, …, m) and competition layer matrix βj(j is 1, 2, …, n) and the weight ω of input layer matrix neuron i and output neuron jijCalculating Euclidean values
Figure BDA0003278600760000081
Minimum Euclidean value djThe corresponding input layer matrix neuron i is a winning neuron, the best matching of the competition layer matrix neuron j is realized by adjusting the winning neuron and the adjacent weight, the similar aggregation is realized by gradual iteration, and a winning connotation matrix gamma is established by the best matching neuron and the adjacent neurons thereofk(k 1, 2, …, n), iterative optimization of the neuron weight coefficients τ is achieved within the context of an inclusion matrix, i.e.
Figure BDA0003278600760000082
ωij＝ωij+ε(αi-ωij)
Wherein posτ，posjTo win over the neuron tau, j positions of the connotation matrix,
Figure BDA0003278600760000083
respectively, a defined threshold and a learning rate.
In practical application, the trained neural network outputs the optimal array characteristic parameters, including: and outputting the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameter by the trained neural network through nonlinear fitting.
In practical application, the method realizes the establishment of a non-linear fitting clustering model of shadow multiplying power by inputting inclination angle, spacing and power generation data samples, using a Weber-Fisher law, an association clustering theory and a Kohonen neural network, and outputting an optimal inclination angle, an optimal array spacing and a maximum power generation characteristic parameter.
In practical application, a certain amount of small sample database is formed by a large amount of simulation experiment data acquisition, the Weber-Fisher law, wavelet decomposition and filter denoising and reconstruction are combined, the Kohonen neural network model of a large data sample set is trained and tested by resisting the internal recessive rule expansion of a generation network, and finally the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameters are output. The scheme of the invention realizes the nonlinear optimization of shadow multiplying power optimization.
The implementation of the invention has the advantages that: the invention relates to a non-linear fitting optimal arraying method based on shadow multiplying power, which comprises the following steps of: acquiring a matrix data sample and forming a data matrix according to the data sample; carrying out data preprocessing on the data matrix; extracting characteristic quantity and extending implicit rules of the preprocessed data matrix; inputting the expanded data matrix into a pre-constructed neural network for training; and outputting the optimal array characteristic parameters by the trained neural network. And (3) depending on environmental data, performing a certain amount of event simulation, mining and expanding data characteristic quantity aiming at discrete distributed event data, and performing nonlinear fitting through a trained and tested Kohonen neural network clustering model to output optimal characteristic parameters. The optimal photovoltaic design electricity cost is realized, and the optimal inclination angle, the optimal array spacing, the land control index and the capital investment mode are coordinated and unified.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention disclosed herein are intended to be covered by the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.

Claims (10)

1. A non-linear fitting optimal arraying method based on shadow multiplying power is characterized by comprising the following steps:
acquiring a matrix data sample and forming a data matrix according to the data sample;
carrying out data preprocessing on the data matrix;
extracting characteristic quantity and extending implicit rules of the preprocessed data matrix;
inputting the expanded data matrix into a pre-constructed neural network for training;
and outputting the optimal array characteristic parameters by the trained neural network.
2. The shadow magnification-based nonlinear fitting optimal arraying method according to claim 1, wherein the obtaining of arrayed data samples and the forming of a data matrix from the data samples comprise: acquiring maximum unconcealed day data of winter solstice and spring equinox, group string inclined plane length data, group string length data, inclination angle data, latitude data, bracket interval data, south-north and east-west gradient data and generating capacity data, and establishing a data matrix taking the inclination angle, the arrangement interval and the generating capacity as characteristics.
3. The non-linear fitting optimal arraying method based on shadow multiplying power as claimed in claim 1, wherein the data preprocessing on the data matrix comprises: and carrying out Weber-Fisher law normalization processing, wavelet decomposition and filter setting and reconstruction on the data matrix.
4. The shadow magnification-based nonlinear fitting optimal constellation method as claimed in claim 3, wherein the Weber-Fisher's law
Figure FDA0003278600750000011
Indicating that the generating capacity is in direct proportion to the logarithm of the inclination angle and the distance of a certain interval; wherein
Figure FDA0003278600750000012
For evaluating ginsengAnd the number coefficient and the reference quantity are the inclination angle, the distance and the generated energy which are calculated theoretically.
5. The shadow-magnification-based nonlinear-fitting optimal arraying method according to claim 4, wherein the wavelet decomposition and filter setting and reconstruction thereof performs function reduction and denoising processing of noise data on a data matrix and forms noise samples.
6. The non-linear fitting optimal arraying method based on shadow multiplying power as claimed in claim 5, wherein the performing feature quantity extraction and implicit rule extension on the preprocessed data matrix comprises: and performing GAN network learning training on the noise sample and the preprocessed data matrix aiming at the characteristic quantity and implicit rule extension.
7. The non-linear fit optimal arraying method based on shadow magnification of claim 6, wherein the GAN network learning training comprises: the noise sample is input into the generator to generate generated data, the generated data and the preprocessed data matrix are input into the discriminator to be discriminated, and the discrimination result is returned to the generator through the loss function LG and returned to the discriminator through the loss function LD, so that the data matrix is expanded.
8. The shadow magnification-based nonlinear fitting optimal arraying method according to claim 1, wherein the inputting the expanded data matrix into a pre-constructed neural network for training comprises: and inputting the expanded data matrix into a Kohonen neural network clustering model for training and testing.
9. The shadow magnification-based nonlinear fitting optimal arraying method according to claim 8, wherein the inputting of the expanded data matrix into a Kohonen neural network clustering model for training and testing comprises: by connecting weights omegaijImplementing input layer feature vectors alphai(i ═ 1, 2, …, m) and competition layer matrix βj(j＝1，2，…, n), from the weight ω of the input layer matrix neuron i and the output neuron jijCalculating Euclidean values
Figure FDA0003278600750000021
Minimum Euclidean value djThe corresponding input layer matrix neuron i is a winning neuron, the best matching of the competition layer matrix neuron j is realized by adjusting the winning neuron and the adjacent weight, the similar aggregation is realized by gradual iteration, and a winning connotation matrix gamma is established by the best matching neuron and the adjacent neurons thereofk(k 1, 2, …, n), iterative optimization of the neuron weight coefficients τ is achieved within the context of an inclusion matrix, i.e.
Figure FDA0003278600750000022
ωij＝ωij+ε(αj-ωij)
Wherein posτ，posjTo win over the neuron tau position of the connotation matrix,
Figure FDA0003278600750000023
respectively, a defined threshold and a learning rate.
10. The method of claim 2, wherein the outputting of the optimal constellation characteristic parameters by the trained neural network comprises: and outputting the optimal inclination angle, the optimal array spacing and the maximum generating capacity characteristic parameter by the trained neural network through nonlinear fitting.
CN202111125125.3A 2021-09-25 2021-09-25 Non-linear fitting optimal array distribution method based on shadow multiplying power Pending CN113987916A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111125125.3A CN113987916A (en) 2021-09-25 2021-09-25 Non-linear fitting optimal array distribution method based on shadow multiplying power

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111125125.3A CN113987916A (en) 2021-09-25 2021-09-25 Non-linear fitting optimal array distribution method based on shadow multiplying power

Publications (1)

Publication Number Publication Date
CN113987916A true CN113987916A (en) 2022-01-28

Family

ID=79736605

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111125125.3A Pending CN113987916A (en) 2021-09-25 2021-09-25 Non-linear fitting optimal array distribution method based on shadow multiplying power

Country Status (1)

Country Link
CN (1) CN113987916A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104281741A (en) * 2014-09-10 2015-01-14 长江勘测规划设计研究有限责任公司 Photovoltaic assembly inclination angle and array pitch cross feedback multi-factor comprehensive calculation method
CN107330517A (en) * 2017-06-14 2017-11-07 华北电力大学 One kind is based on S_Kohonen non-intrusion type resident load recognition methods
CN111458142A (en) * 2020-04-02 2020-07-28 苏州智传新自动化科技有限公司 Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network
CN113128903A (en) * 2021-05-06 2021-07-16 中国电子科技集团公司第五十四研究所 Satellite task demand configuration method and system
CN216053046U (en) * 2021-09-25 2022-03-15 中机国能电力工程有限公司 Optimal array distribution system based on shadow multiplying power

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104281741A (en) * 2014-09-10 2015-01-14 长江勘测规划设计研究有限责任公司 Photovoltaic assembly inclination angle and array pitch cross feedback multi-factor comprehensive calculation method
CN107330517A (en) * 2017-06-14 2017-11-07 华北电力大学 One kind is based on S_Kohonen non-intrusion type resident load recognition methods
CN111458142A (en) * 2020-04-02 2020-07-28 苏州智传新自动化科技有限公司 Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network
CN113128903A (en) * 2021-05-06 2021-07-16 中国电子科技集团公司第五十四研究所 Satellite task demand configuration method and system
CN216053046U (en) * 2021-09-25 2022-03-15 中机国能电力工程有限公司 Optimal array distribution system based on shadow multiplying power

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
丁凯宁: "《山地型光伏发电系统的优化设计及并网方案的研究》", 《中国优秀硕士学位论文全文数据库 工程科技II辑》, no. 3, 15 March 2021 (2021-03-15), pages 19 - 35 *
赵为光等: "《复式Kohonen和改进 BP网络的微网光伏发电预测》", 《黑龙江科技大学学报》, vol. 27, no. 3, 31 May 2017 (2017-05-31), pages 297 - 302 *

Similar Documents

Publication Publication Date Title
CN110059878B (en) Photovoltaic power generation power prediction model based on CNN LSTM and construction method thereof
Yang et al. A novel competitive swarm optimized RBF neural network model for short-term solar power generation forecasting
CN111310889B (en) Evaporation waveguide profile estimation method based on deep neural network
CN112766199A (en) Hyperspectral image classification method based on self-adaptive multi-scale feature extraction model
CN109117951B (en) BP neural network-based probability load flow online calculation method
CN111553587B (en) New energy scene generation method and system based on confrontation learning model
CN109492556B (en) Synthetic aperture radar target identification method for small sample residual error learning
CN110334870B (en) Photovoltaic power station short-term power prediction method based on gated cyclic unit network
CN111582555B (en) Photovoltaic power prediction method based on foundation cloud picture image characteristics
CN114707765B (en) Federal learning load prediction method based on dynamic weighted aggregation
CN108830411A (en) A kind of wind power forecasting method based on data processing
CN113935249B (en) Upper-layer ocean thermal structure inversion method based on compression and excitation network
CN112699838B (en) Hyperspectral mixed pixel nonlinear blind decomposition method based on spectral diagnosis characteristic weighting
CN117648863B (en) Regional water supply capacity acquisition method under special drought conditions based on water system communication
CN112052913A (en) Distributed photovoltaic power station power data virtual acquisition method
CN216053046U (en) Optimal array distribution system based on shadow multiplying power
CN117374956A (en) Short-term prediction method for photovoltaic power generation of comprehensive energy station
CN114792158A (en) Multi-wind-farm short-term power prediction method based on space-time fusion graph neural network
CN112418346A (en) Numerical weather forecast total radiation system error classification calculation method
CN108038518A (en) A kind of photovoltaic generation power based on meteorological data determines method and system
CN113987916A (en) Non-linear fitting optimal array distribution method based on shadow multiplying power
CN112734073A (en) Photovoltaic power generation short-term prediction method based on long and short-term memory network
CN115456286A (en) Short-term photovoltaic power prediction method
Zhang et al. Spectral normalization generative adversarial networks for photovoltaic power scenario generation
CN115545256A (en) Small-scale photovoltaic power prediction method based on multi-dimensional data feature learning

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