CN110676852B - Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics - Google Patents

Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics Download PDF

Info

Publication number
CN110676852B
CN110676852B CN201910791388.4A CN201910791388A CN110676852B CN 110676852 B CN110676852 B CN 110676852B CN 201910791388 A CN201910791388 A CN 201910791388A CN 110676852 B CN110676852 B CN 110676852B
Authority
CN
China
Prior art keywords
extreme learning
learning machine
power
node
layer
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910791388.4A
Other languages
Chinese (zh)
Other versions
CN110676852A (en
Inventor
余娟
高倩
杨知方
代伟
雷星雨
余红欣
王洪彬
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.)
Chongqing University
Electric Power Research Institute of State Grid Chongqing Electric Power Co Ltd
State Grid Corp of China SGCC
Original Assignee
Chongqing University
Electric Power Research Institute of State Grid Chongqing Electric Power Co Ltd
State Grid Corp of China SGCC
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 Chongqing University, Electric Power Research Institute of State Grid Chongqing Electric Power Co Ltd, State Grid Corp of China SGCC filed Critical Chongqing University
Priority to CN201910791388.4A priority Critical patent/CN110676852B/en
Publication of CN110676852A publication Critical patent/CN110676852A/en
Application granted granted Critical
Publication of CN110676852B publication Critical patent/CN110676852B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/04—Circuit arrangements for ac mains or ac distribution networks for connecting networks of the same frequency but supplied from different sources
    • H02J3/06—Controlling transfer of power between connected networks; Controlling sharing of load between connected 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

Abstract

The invention discloses a method for calculating fast probability load flow of an improved extreme learning machine by considering load flow characteristics, which mainly comprises the following steps: 1) and acquiring basic data of the power network. 2) Based on basic data of the power network, establishing a mapping relation f: pi，Qi→Ui，θi.3) And decomposing the mapping relation and establishing an extreme learning machine neural network. 4) And (4) optimizing hidden layer parameters of the extreme learning machine neural network so as to establish an improved extreme learning machine neural network. 5) And inputting the basic data of the power network into the neural network of the improved extreme learning machine, and calculating to obtain the probability load flow of the power network. The method replaces the time-consuming solving process of a large-scale high-dimensional complex nonlinear power flow equation in the PPF calculation by a high-precision simulation method, thereby considering the actual engineering requirements of the PPF calculation on precision and speed.

Description

Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics
Technical Field
The invention relates to the field of electric power systems and automation thereof, in particular to a fast probability load flow calculation method of an improved extreme learning machine considering load flow characteristics.
Background
The increasing source and load uncertainty makes the Probabilistic Power Flow (PPF) an important tool for Power system planning, operation and reliability evaluation. The calculation of PPF is mainly divided into simulation method, analytic method and approximation method. The simulation method solves a large number of high-dimensional complex nonlinear power flow samples to ensure the accuracy of state quantity statistical analysis, but the calculation speed is insufficient. The analytic method carries out convolution calculation on the relation between input random variables to obtain state quantity probability distribution, and the approximation method approximately describes the statistical characteristics of the state quantity by using the digital characteristics of the input random variables, all aims to avoid large-scale sampling, and has high calculation speed but is difficult to ensure the accuracy. The existing method cannot meet the requirements of high precision and high speed of PPF.
The neural network can learn the mapping relation between input and output under the line and perform quick calculation on the line, and is expected to replace the time-consuming solving process of a high-precision simulation method for a large number of samples, thereby realizing both high precision and high speed. The deep learning algorithm in the neural network has strong fitting capability, but has the problems of large network parameter adjustment scale, long time consumption, lack of guidance and the like. And the other Extreme Learning Machine (ELM) algorithm has less adjusting parameters and is efficient to implement. However, when the ELM is applied to the PPF calculation, the problems that the shallow structure has limited feature extraction capability, errors are increased due to random generation of hidden layer parameters and the like need to be overcome.
Disclosure of Invention
The present invention is directed to solving the problems of the prior art.
The technical scheme adopted for achieving the purpose of the invention is that the improved extreme learning machine rapid probabilistic power flow calculation method considering the power flow characteristics mainly comprises the following steps:
1) and acquiring basic data of the power network.
Furthermore, the basic data of the power network mainly comprises active power, reactive power, voltage amplitude, voltage phase angle, branch active power and reactive power of the power network nodes.
2) Based on basic data of the power network, establishing a mapping relation f: pi，Qi→Ui，θi. Wherein, Pi、QiRespectively representing the active power and the reactive power injected by the node i. U shapei、θiRespectively representing nodesiVoltage magnitude and voltage phase angle.
3) And decomposing the mapping relation and establishing an extreme learning machine neural network.
Further, the main steps for establishing the neural network of the extreme learning machine are as follows:
3.1) mapping the relationship f: pi，Qi→Ui，θiDecomposed into a mapping relation f1：Pi，Qi→Pij，QijAnd a mapping relation f2：Pij，Qij→Ui，θi。
3.2) based on the mapping f1：Pi，Qi→Pij，QijAnd establishing a first-stage extreme learning machine neural network. Wherein the input layer is a nodeiInjected active power PiReactive power Q injected from node ii. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. Branch active power P with output layer of line ijijAnd branch reactive power Q of line ijij。
Active power P injected by node iiReactive power Q injected from node iiAs follows:
Figure BDA0002179647980000021
in the formula, n represents the number of nodes in the power network, i, j represents the number of the nodes, j ∈ i represents that the node j needs to be connected with the node i, and j ═ i is included. Thetaij＝θi-θjRepresenting the voltage angle difference of node j and node i. Gij、BijRespectively, conductance and susceptance.
3.3) based on the mapping f2：Pij，Qij→Ui，θiAnd establishing a second-stage extreme learning machine neural network. Wherein, the input layer is the branch active power P of the line ijijAnd branch reactive power Q of line ijij. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. The output layer being a nodeiVoltage phase angle theta ofiAnd the voltage amplitude U of the node iiSquare of (2) Ti。
Figure BDA0002179647980000022
Wherein, the branch active power P of the line ijijAnd branch reactive power Q of line ijijAs follows:
Figure BDA0002179647980000023
Figure BDA0002179647980000024
4) and (4) optimizing hidden layer parameters of the extreme learning machine neural network so as to establish an improved extreme learning machine neural network.
Further, the main steps for optimizing hidden layer parameters of the neural network of the extreme learning machine are as follows:
4.1) in a supervised layer with error correction module, randomly generating parameters a and b which obey normal distribution and establishing optimized variables
Figure BDA0002179647980000025
Wherein, mua、μbRespectively representing the mean values of the parameter a and the parameter b,
Figure BDA0002179647980000031
representing the variance of parameter a and parameter b, respectively.
Figure BDA0002179647980000032
4.2) establishing a hidden layer output expression, namely:
Figure BDA0002179647980000033
in the formula, x represents hidden layer input data. g (-) represents the activation function. H ═ g (a · x + b) denotes the feature mapping matrix. a denotes hidden layer input weight. b represents the input deviation. β represents the output weight. λ is a penalty factor.
4.3) establishing an objective function, namely:
Figure BDA0002179647980000034
4.4) substituting the formula 4 into the formula 3, and calculating to obtain an optimized variable VoptNamely:
Figure BDA0002179647980000035
4.5) based on the optimization variable VoptAnd correcting the hidden layer input data to obtain the hidden layer output data.
5) And inputting the basic data of the power network into the neural network of the improved extreme learning machine, and calculating to obtain the probability load flow of the power network.
The technical effect of the present invention is undoubted. The method replaces the time-consuming solving process of a large-scale high-dimensional complex nonlinear power flow equation in the PPF calculation by a high-precision simulation method, thereby considering the actual engineering requirements of the PPF calculation on precision and speed. Therefore, the present invention has the following effects:
1) the fitting difficulty of the ELM to the PPF algorithm is reduced by decomposing and reducing the trend characteristics.
2) And the fitting capability of the ELM to the PPF calculation is improved by using the error correction mode.
3) The fitting effect of the ELM on the PPF calculation is improved by optimizing and solving the ELM hidden layer parameter optimal generation mode.
Drawings
FIG. 1 is a flow characteristic decomposition order-reducing diagram;
FIG. 2 illustrates an error correction mode;
fig. 3 is the overall structure of the FLM neural network.
Detailed Description
The present invention is further illustrated by the following examples, but it should not be construed that the scope of the above-described subject matter is limited to the following examples. Various substitutions and alterations can be made without departing from the technical idea of the invention and the scope of the invention is covered by the present invention according to the common technical knowledge and the conventional means in the field.
Example 1:
referring to fig. 1 to 3, the method for calculating the fast probability load flow of the improved extreme learning machine considering the load flow characteristics mainly comprises the following steps:
1) and acquiring basic data of the power network.
Furthermore, the basic data of the power network mainly comprises active power, reactive power, voltage amplitude, voltage phase angle, branch active power and reactive power of the power network nodes.
2) Based on basic data of the power network, establishing a mapping relation f: pi，Qi→Ui，θi. Wherein, Pi、QiRespectively representing the active power and the reactive power injected by the node i. U shapei、θiRepresenting the voltage magnitude and voltage phase angle, respectively, of node i.
3) And decomposing the mapping relation and establishing an extreme learning machine neural network.
Further, the main steps for establishing the neural network of the extreme learning machine are as follows:
3.1) mapping the relationship f: pi，Qi→Ui，θiDecomposed into a mapping relation f1：Pi，Qi→Pij，QijAnd a mapping relation f2：Pij，Qij→Ui，θi。
3.2) based on the mapping f1：Pi，Qi→Pij，QijAnd establishing a first-stage extreme learning machine neural network. Wherein the input layer is the active power P injected by the node iiReactive power Q injected from node ii. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. Branch active power P with output layer of line ijijAnd branch reactive power Q of line ijij。
Active power P injected by node iiReactive power Q injected from node iiAs follows:
Figure BDA0002179647980000041
in the formula, n represents the number of nodes in the power network, i, j represents the number of the nodes, j ∈ i represents that the node j needs to be connected with the node i, and j ═ i is included. Thetaij＝θi-θjRepresenting the voltage angle difference of node j and node i. Gij、BijRespectively, conductance and susceptance. U shapej、θjRepresenting the voltage magnitude and voltage phase angle, respectively, of node j.
3.3) based on the mapping f2：Pij，Qij→Ui，θiAnd establishing a second-stage extreme learning machine neural network. Wherein, the input layer is the branch active power P of the line ijijAnd branch reactive power Q of line ijij. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. The output layer being a nodeiVoltage phase angle theta ofiAnd the voltage amplitude U of the node iiSquare of (2) Ti。
Figure BDA0002179647980000051
Wherein, the branch active power P of the line ijijAnd branch reactive power Q of line ijijAs follows:
Figure BDA0002179647980000052
Figure BDA0002179647980000053
in the formula, TjIs the voltage amplitude UjSquare of (d).
4) And (4) optimizing hidden layer parameters of the extreme learning machine neural network so as to establish an improved extreme learning machine neural network.
Further, the main steps for optimizing hidden layer parameters of the neural network of the extreme learning machine are as follows:
4.1) in a supervised layer with error correction module, randomly generating parameters a and b which obey normal distribution and establishing optimized variables
Figure BDA0002179647980000054
Wherein, mua、μbRespectively representing the mean values of the parameter a and the parameter b,
Figure BDA0002179647980000055
representing the variance of parameter a and parameter b, respectively.
Figure BDA0002179647980000056
4.2) establishing a hidden layer output expression, namely:
Figure BDA0002179647980000057
in the formula, x represents hidden layer input data. g (-) represents the activation function. H ═ g (a · x + b) denotes the feature mapping matrix. a denotes hidden layer input weight. b represents the input deviation. β represents the output weight. λ is a penalty factor. The superscript T denotes transpose.
Figure BDA0002179647980000058
For iteratively modified hidden layer outputs, f (x) is the hidden layer output.
4.3) establishing an objective function, namely:
Figure BDA0002179647980000061
4.4) substituting the formula 4 into the formula 3, and calculating to obtain an optimized variable VoptNamely:
Figure BDA0002179647980000062
4.5) based on the optimization variable VoptAnd correcting the hidden layer input data to obtain the hidden layer output data.
5) And inputting the basic data of the power network into the neural network of the improved extreme learning machine, and calculating to obtain the probability load flow of the power network.
Example 2:
the fast probability load flow calculation method of the improved extreme learning machine considering the load flow characteristics mainly comprises the following steps:
1) and acquiring basic data of the power network.
2) Based on basic data of the power network, establishing a mapping relation f: pi，Qi→Ui，θi. Wherein, Pi、QiRespectively representing the active power and the reactive power injected by the node i. U shapei、θiRespectively representing nodesiVoltage magnitude and voltage phase angle.
3) And decomposing the mapping relation and establishing an extreme learning machine neural network.
4) And (4) optimizing hidden layer parameters of the extreme learning machine neural network so as to establish an improved extreme learning machine neural network.
5) And inputting the basic data of the power network into the neural network of the improved extreme learning machine, and calculating to obtain the probability load flow of the power network.
Example 3:
the method for calculating the fast probability load flow of the improved extreme learning machine considering the load flow characteristics mainly comprises the following steps of embodiment 2, wherein the method for establishing the neural network of the extreme learning machine comprises the following main steps:
1) and mapping the relation f: pi，Qi→Ui，θiDecomposed into a mapping relation f1：Pi，Qi→Pij，QijAnd a mapping relation f2：Pij，Qij→Ui，θi。
2) Based on the mapping relation f1：Pi，Qi→Pij，QijAnd establishing a first-stage extreme learning machine neural network. Wherein the input layer is the active power P injected by the node iiReactive power Q injected from node ii. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. Branch active power P with output layer of line ijijAnd branch reactive power Q of line ijij。
Active power P injected by node iiReactive power Q injected from node iiAs follows:
Figure BDA0002179647980000071
in the formula, n represents the number of nodes in the power network, i, j represents the number of the nodes, j ∈ i represents that the node j needs to be connected with the node i, and j ═ i is included. Thetaij＝θi-θjRepresenting the voltage angle difference of node j and node i. Gij、BijRespectively, conductance and susceptance.
3) Based on the mapping relation f2：Pij，Qij→Ui，θiAnd establishing a second-stage extreme learning machine neural network. Wherein, the input layer is the branch active power P of the line ijijAnd branch reactive power Q of line ijij. The hidden layers include an unsupervised layer with an extreme learning sparse auto-encoder, a supervised layer with extreme learning raw algorithms, and a supervised layer with an error correction module. The output layer being a nodeiVoltage phase angle theta ofiAnd the voltage amplitude U of the node iiSquare of (2) Ti。
Figure BDA0002179647980000072
Wherein, the branch active power P of the line ijijAnd branch reactive power Q of line ijijAs follows:
Figure BDA0002179647980000073
Figure BDA0002179647980000074
example 4:
the method for calculating the fast probability load flow of the improved extreme learning machine considering the load flow characteristics mainly comprises the following steps of embodiment 2,
the main steps for optimizing hidden layer parameters of the neural network of the extreme learning machine are as follows:
1) in a supervised layer with error correction module, parameters a and b obeying normal distribution are randomly generated and optimized variables are established
Figure BDA0002179647980000075
Wherein, mua、μbRespectively representing the mean values of the parameter a and the parameter b,
Figure BDA0002179647980000076
respectively represent ginsengThe variance of the number a and the parameter b.
Figure BDA0002179647980000077
2) Establishing a hidden layer output expression, namely:
Figure BDA0002179647980000081
in the formula, x represents hidden layer input data. g (-) represents the activation function. H ═ g (a · x + b) denotes the feature mapping matrix. a denotes hidden layer input weight. b represents the input deviation. β represents the output weight. λ is a penalty factor.
3) Establishing an objective function, namely:
Figure BDA0002179647980000082
4) substituting the formula 4 into the formula 3, and calculating to obtain an optimized variable VoptNamely:
Figure BDA0002179647980000083
5) based on an optimization variable VoptAnd correcting the hidden layer input data to obtain the hidden layer output data.
Example 5:
the experiment for verifying the improved extreme learning machine rapid probability load flow calculation method considering the load flow characteristics mainly comprises the following steps:
1) the effectiveness of the invention is verified by IEEE 57 calculation examples, the permeability is 45%, and the load fluctuation is 10%. The comparison algorithm comprises the following steps: m7: the unmodified ELM algorithm. M1: and (4) considering a power flow characteristic decomposition order reduction method on the basis of M7. M2: the method of error correction mode is further considered on the basis of M1. M3: the method for optimizing hidden layer parameters is further considered on the basis of M2. The accuracy and maximum out-of-limit comparison results of the branch active power, the branch reactive power, the node voltage phase angle and the node voltage amplitude obtained by the algorithms are shown in table 1. Simulation results show that the fitting effect of the ELM algorithm on probability load flow calculation is effectively improved.
TABLE 1 comparison of the improved results of the present invention in the IEEE 57 calculation example
Figure BDA0002179647980000091
2) Superiority verification and analysis of the invention
The superiority of the invention is verified by IEEE 57 and IEEE 2383 calculation examples, wherein the permeability of IEEE 2383 is 30%, and the load fluctuation is 10%. The comparison algorithm comprises the following steps: m0: the invention relates to a method for preparing a high-temperature-resistant ceramic material. M4: and (3) taking a traditional Newton-Raphson iteration method as a precision standard. M5: a commonly used deep learning algorithm SDA. M6: a common deep learning algorithm SAE. The accuracy comparison results of the training time, the testing time, the node voltage phase angle and the amplitude of each algorithm are shown in tables 2 and 3. Simulation results show that the test time and the precision performance are comprehensively considered, compared with other algorithms, the method can realize high-precision and high-speed PPF calculation, the training time is far shorter than that of a deep learning algorithm, the method is more suitable for the actual engineering requirements, and the superiority is verified.
Figure BDA0002179647980000092
In conclusion, the invention provides an improved ELM rapid PPF calculation method considering the trend characteristics. The PPF calculation fitting difficulty is reduced by using a load flow characteristic decomposition order reduction method, an error correction mode and a hidden layer parameter optimization model are provided to improve the ELM fitting capability, and the PPF calculation method with high precision and high speed is formed. Example researches show that compared with other algorithms, the method is more suitable for engineering requirements of power system probability load flow calculation.

Claims (3)

1. The fast probability load flow calculation method of the improved extreme learning machine considering the load flow characteristics is characterized by mainly comprising the following steps of:
1) acquiring basic data of a power network;
2) establishing a mapping relation f: P based on basic data of the power networki,Qi→Ui,θi(ii) a Wherein, Pi、QiRespectively representing active power and reactive power injected by a node i; u shapei、θiRespectively representing the voltage amplitude and the voltage phase angle of the node i;
3) decomposing the mapping relation and establishing an extreme learning machine neural network;
the method mainly comprises the following steps of:
3.1) mapping the relationship f to Pi,Qi→Ui,θiDecomposed into a mapping relation f1:Pi,Qi→Pij,QijAnd a mapping relation f2:Pij,Qij→Ui,θi；
3.2) based on the mapping f1:Pi,Qi→Pij,QijEstablishing a first-stage extreme learning machine neural network; wherein the input layer is the active power P injected by the node iiReactive power Q injected from node ii(ii) a The hidden layer comprises an unsupervised layer with an extreme learning machine sparse self-encoder, a supervised layer with an extreme learning machine original algorithm and a supervised layer with an error correction module; branch active power P with output layer of line ijijAnd branch reactive power Q of line ijij；
Active power P injected by node iiReactive power Q injected from node iiAs follows:
Figure FDA0002678232260000011
in the formula, n represents the number of nodes in the power network, i, j represents the number of the nodes, j belongs to i and represents that the node j needs to be connected with the node i, and j is equal to i; thetaij＝θi-θjRepresents the voltage phase angle difference of the node j and the node i; gij、BijRespectively representing conductance and electricityNano;
3.3) based on the mapping f2:Pij,Qij→Ui,θiEstablishing a second-stage extreme learning machine neural network; wherein, the input layer is the branch active power P of the line ijijAnd branch reactive power Q of line ijij(ii) a The hidden layer comprises an unsupervised layer with an extreme learning machine sparse self-encoder, a supervised layer with an extreme learning machine original algorithm and a supervised layer with an error correction module; the phase angle theta of the voltage with the output layer as a node iiAnd the voltage amplitude U of the node iiSquare of (2) Ti；
Figure FDA0002678232260000012
Wherein, the branch active power P of the line ijijAnd branch reactive power Q of line ijijAs follows:
Figure FDA0002678232260000021
4) optimizing hidden layer parameters of the neural network of the extreme learning machine so as to establish an improved neural network of the extreme learning machine;
5) and inputting the basic data of the power network into the neural network of the improved extreme learning machine, and calculating to obtain the probability load flow of the power network.
2. The improved extreme learning machine fast probabilistic power flow calculation method considering power flow characteristics as claimed in claim 1, wherein: the basic data of the power network mainly comprise active power, reactive power, voltage amplitude, voltage phase angle, branch active power and reactive power of power network nodes.
3. The method for improving the rapid probabilistic power flow calculation of the extreme learning machine considering the power flow characteristics as claimed in claim 1, wherein the main steps of optimizing hidden layer parameters of the neural network of the extreme learning machine are as follows:
1) in thatIn a supervised layer with an error correction module, parameters a and b which are subject to normal distribution are randomly generated, and an optimized variable is established
Figure FDA0002678232260000022
Wherein, mua、μbRespectively representing the mean values of the parameter a and the parameter b,
Figure FDA0002678232260000023
respectively representing the variances of the parameter a and the parameter b;
Figure FDA0002678232260000024
2) establishing a hidden layer output expression, namely:
Figure FDA0002678232260000025
in the formula, x represents hidden layer input data; g (-) represents an activation function; h ═ g (a · x + b) represents the feature mapping matrix; a represents hidden layer input weight; b represents an input deviation; β represents an output weight; λ is a penalty factor;
3) establishing an objective function, namely:
Figure FDA0002678232260000026
4) substituting the formula 4 into the formula 3, and calculating to obtain an optimized variable VoptNamely:
Figure FDA0002678232260000031
5) based on an optimization variable VoptAnd correcting the hidden layer input data to obtain the hidden layer output data.
CN201910791388.4A 2019-08-26 2019-08-26 Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics Active CN110676852B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910791388.4A CN110676852B (en) 2019-08-26 2019-08-26 Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910791388.4A CN110676852B (en) 2019-08-26 2019-08-26 Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics

Publications (2)

Publication Number Publication Date
CN110676852A CN110676852A (en) 2020-01-10
CN110676852B true CN110676852B (en) 2020-11-10

Family

ID=69075571

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910791388.4A Active CN110676852B (en) 2019-08-26 2019-08-26 Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics

Country Status (1)

Country Link
CN (1) CN110676852B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112100920A (en) * 2020-09-15 2020-12-18 东南大学 Power distribution network three-phase voltage calculation method, device, equipment and storage medium
CN113987940B (en) * 2021-10-28 2024-02-23 国网山东省电力公司烟台供电公司 ELM integrated online learning-based power system tide data driving regression method

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101252280A (en) * 2008-04-08 2008-08-27 昆明理工大学 On-line evaluating method of urban network max power supply capability
CN101373905A (en) * 2008-05-16 2009-02-25 清华大学 Visualization method for voltage stability domain of electric power system
CN107069708A (en) * 2017-03-14 2017-08-18 国网湖北省电力公司 A kind of power grids circuits strategy for security correction method based on extreme learning machine
CN107681682A (en) * 2017-10-25 2018-02-09 国家电网公司 A kind of ac and dc systemses equivalence method equivalent based on WARD
US9960599B1 (en) * 2017-06-06 2018-05-01 University Of Macau Thyristor controlled LC compensator for compensating dynamic reactive power
CN108336739A (en) * 2018-01-15 2018-07-27 重庆大学 A kind of Probabilistic Load Flow on-line calculation method based on RBF neural
CN108695859A (en) * 2018-06-04 2018-10-23 南京工程学院 A kind of capacitor group addressing constant volume method of access active distribution network
CN108876054A (en) * 2018-07-06 2018-11-23 国网河南省电力公司郑州供电公司 Short-Term Load Forecasting Method based on improved adaptive GA-IAGA optimization extreme learning machine
CN109086784A (en) * 2017-06-13 2018-12-25 昆山鲲鹏无人机科技有限公司 A kind of intelligent method for classifying based on introducing hiding information extreme learning machine algorithm
CN109117951A (en) * 2018-01-15 2019-01-01 重庆大学 Probabilistic Load Flow on-line calculation method based on BP neural network
CN109217295A (en) * 2018-09-20 2019-01-15 吉林大学 The trend sensitivity computing method and computer installation of prevention system overload
US10291029B2 (en) * 2016-08-18 2019-05-14 General Electric Technology Gmbh Enhanced island management application for power grid systems
CN109995031A (en) * 2019-05-05 2019-07-09 重庆大学 Probabilistic Load Flow deep learning calculation method based on physical model
CN110110434A (en) * 2019-05-05 2019-08-09 重庆大学 A kind of initial method that Probabilistic Load Flow deep neural network calculates

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9362749B2 (en) * 2012-04-20 2016-06-07 Battelle Memorial Institute Controller for thermostatically controlled loads

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101252280A (en) * 2008-04-08 2008-08-27 昆明理工大学 On-line evaluating method of urban network max power supply capability
CN101373905A (en) * 2008-05-16 2009-02-25 清华大学 Visualization method for voltage stability domain of electric power system
US10291029B2 (en) * 2016-08-18 2019-05-14 General Electric Technology Gmbh Enhanced island management application for power grid systems
CN107069708A (en) * 2017-03-14 2017-08-18 国网湖北省电力公司 A kind of power grids circuits strategy for security correction method based on extreme learning machine
US9960599B1 (en) * 2017-06-06 2018-05-01 University Of Macau Thyristor controlled LC compensator for compensating dynamic reactive power
CN109086784A (en) * 2017-06-13 2018-12-25 昆山鲲鹏无人机科技有限公司 A kind of intelligent method for classifying based on introducing hiding information extreme learning machine algorithm
CN107681682A (en) * 2017-10-25 2018-02-09 国家电网公司 A kind of ac and dc systemses equivalence method equivalent based on WARD
CN109117951A (en) * 2018-01-15 2019-01-01 重庆大学 Probabilistic Load Flow on-line calculation method based on BP neural network
CN108336739A (en) * 2018-01-15 2018-07-27 重庆大学 A kind of Probabilistic Load Flow on-line calculation method based on RBF neural
CN108695859A (en) * 2018-06-04 2018-10-23 南京工程学院 A kind of capacitor group addressing constant volume method of access active distribution network
CN108876054A (en) * 2018-07-06 2018-11-23 国网河南省电力公司郑州供电公司 Short-Term Load Forecasting Method based on improved adaptive GA-IAGA optimization extreme learning machine
CN109217295A (en) * 2018-09-20 2019-01-15 吉林大学 The trend sensitivity computing method and computer installation of prevention system overload
CN109995031A (en) * 2019-05-05 2019-07-09 重庆大学 Probabilistic Load Flow deep learning calculation method based on physical model
CN110110434A (en) * 2019-05-05 2019-08-09 重庆大学 A kind of initial method that Probabilistic Load Flow deep neural network calculates

Also Published As

Publication number Publication date
CN110676852A (en) 2020-01-10

Similar Documents

Publication Publication Date Title
CN111008502B (en) Fault prediction method for complex equipment driven by digital twin
CN109214575B (en) Ultrashort-term wind power prediction method based on small-wavelength short-term memory network
CN106548418B (en) Small interference stability evaluation method for power system
CN110676852B (en) Improved extreme learning machine rapid probability load flow calculation method considering load flow characteristics
CN108336739B (en) RBF neural network-based probability load flow online calculation method
CN109255173B (en) Structural failure probability interval calculation method considering interval uncertainty
Jin et al. Design and analysis of two FTRNN models with application to time-varying Sylvester equation
CN107528322B (en) Probability power flow analysis method based on Gauss-Hermite product-solving method for NATAF transformation
CN107272409B (en) Linear servo system vibration suppression method based on iterative learning
CN109726437B (en) Cabin door pneumatic load equivalent node force processing method
CN109599872A (en) Probabilistic optimal load flow calculation method based on storehouse noise reduction autocoder
Peng et al. Efficient multiple starting point optimization for automated analog circuit optimization via recycling simulation data
CN112257361A (en) Standard unit library construction method based on quadratic fitting model
Mokhlis et al. System identification of a dc servo motor using arx and armax models
Du et al. An effective gradient projection method for stochastic optimal control
CN116894180B (en) Product manufacturing quality prediction method based on different composition attention network
CN110707703B (en) Improved Nataf transformation-based efficient probabilistic power flow calculation method containing high-dimensional related uncertain sources
CN109950935B (en) Probability tide method for alternating current-direct current (AC-DC) series-parallel micro-grid in isolated island operation
CN107480381B (en) The method of response surface model is built based on simulated annealing and applies its system
CN111797564A (en) Method and system for obtaining correlation sample of high-dimensional distributed photovoltaic output
Spiegel et al. Iterative learning control with discrete‐time nonlinear nonminimum phase models via stable inversion
CN116316629A (en) Data-driven optimal power flow calculation method considering topology feature learning
CN111859657B (en) Fast iterative solution method for optimizing oil consumption of engine
CN113627670A (en) Ultra-short-term photovoltaic power prediction system based on deep Fourier network
CN111798037A (en) Data-driven optimal power flow calculation method based on stacked extreme learning machine frame

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