CN106777494B - Method for calculating sensitivity of reliability influence factors of power system - Google Patents

Method for calculating sensitivity of reliability influence factors of power system Download PDF

Info

Publication number
CN106777494B
CN106777494B CN201611031272.3A CN201611031272A CN106777494B CN 106777494 B CN106777494 B CN 106777494B CN 201611031272 A CN201611031272 A CN 201611031272A CN 106777494 B CN106777494 B CN 106777494B
Authority
CN
China
Prior art keywords
reliability
sensitivity
neural network
calculating
index
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
CN201611031272.3A
Other languages
Chinese (zh)
Other versions
CN106777494A (en
Inventor
张建功
胡庆辉
田洪迅
王宏刚
万涛
李浩松
李金�
康泰峰
李欣
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.)
Beijing State Grid Information Telecommnication Group Accenture Information Technology Co ltd
State Grid Corp of China SGCC
State Grid Information and Telecommunication Co Ltd
Original Assignee
Beijing State Grid Information Telecommnication Group Accenture Information Technology Co ltd
State Grid Corp of China SGCC
State Grid Information and Telecommunication 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 Beijing State Grid Information Telecommnication Group Accenture Information Technology Co ltd, State Grid Corp of China SGCC, State Grid Information and Telecommunication Co Ltd filed Critical Beijing State Grid Information Telecommnication Group Accenture Information Technology Co ltd
Priority to CN201611031272.3A priority Critical patent/CN106777494B/en
Publication of CN106777494A publication Critical patent/CN106777494A/en
Application granted granted Critical
Publication of CN106777494B publication Critical patent/CN106777494B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • 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
    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent

Abstract

The invention provides a method for calculating sensitivity of influencing factors of reliability of a power system, which comprises the following steps: selecting N links in a neural network of the reliability index, wherein the links meet the following conditions: selecting a neuron in each layer of the neural network to form a link, wherein the selected neurons are directly connected layer by layer; and respectively calculating the sensitivity index of a certain influence factor in each link, and calculating the average value of the sensitivity indexes of all the links as the sensitivity of the influence factor to the reliability index. The invention can effectively distinguish the influence factors into the beneficial factors and the adverse factors through the simulation experiment of the actual data, and quantitatively sorts the sensitivity importance degrees of the influence factors with the same property according to the sensitivity contribution to the indexes.

Description

Method for calculating sensitivity of reliability influence factors of power system
Technical Field
The invention relates to the field of power information management, in particular to a method for calculating sensitivity of influence factors of power system reliability.
Background
At present, the reliability of a power system is closely related to factors such as an operation environment, a system operation planning mode and the like, and a complex dynamic high-dimensional nonlinear relationship is often presented between influencing factors and reliability indexes. According to the traditional method for deducing the system reliability based on the element reliability, as historical statistics and a relatively stable system structure are needed to ensure that the random variable correlation and the selection model structure are correctly described, at present, under the continuous stimulation of the requirements of continuous release of an electric power market, new energy grid connection and the like, the power distribution network system structure of each region is rapidly changed, and the traditional reliability prediction method based on statistical reasoning is difficult to effectively implement and apply.
The data-driven modeling approach provides a new idea for establishing relationships between rapidly changing input data and outputs. Currently, data-driven reliability index modeling and numerical prediction methods have been developed. In 2008 'power system and protection', published 'short-term evaluation of power system operation reliability considering weather prediction', a power system short-term reliability prediction model considering weather prediction is proposed by Hope, Chenglin, Sunyuan and Roc, a support vector machine is adopted to predict air temperature and air speed, and the contribution degree of failure probability of each part of the system to system load shedding probability, expected power shortage value and expected power shortage value is deduced. The probability of load loss, the frequency of load loss and the power deficiency are deduced from the sensitivity analysis of reliability evaluation of a large power system published in the power grid technology 2005 in Zhao Yuan, Zhou Mian, Xie Yuan and the likeSensitivity of the equal-magnitude power system reliability index to element availability, unavailability, failure rate, and repair rate is desired. A.m.l.dasilva, a.c.r.
Figure DEST_PATH_GDA0001207655940000021
And a component reliability parameter checking method is provided based on a concept of sensitivity of component reliability parameters to system reliability indexes and a sequential Monte Carlo simulation algorithm in Dallamo PMAPS international conference of L.C. Nascimento 2014. The 'Reliability and maintenance improvement of conservation With Aging Infrastructure', published by ieee transactions on Power Delivery in 2014, h.ge and s.asgarpoor, proposes the sensitivity of the substation availability relative to the availability of its constituent elements, and uses the specification value of the sensitivity to characterize the contribution of each element to the Reliability of the substation. X.Zhu in A New method of Analytical Formula reduction and sensing Analysis of EENS in Bulk Power System Reliability Assessment, Power supply is expected for the Reliability index of the composite Power System, and the sensitivity of the element failure rate and the repair rate is deduced. An element failure rate and repair time optimization method based on element maintenance cost sensitivity analysis is proposed in the Application of sensitive analysis for Improving Reliability indexes of a Radial Distribution System by R.arya, S.Choube, L.arya, and R.Shrivastava. On the basis, the invention provides a power distribution network reliability associated factor sensitivity calculation method based on a BP neural network.
Disclosure of Invention
The present invention provides a method that overcomes, or at least partially solves, the above problems.
According to one aspect of the invention, a method for calculating the sensitivity of the influence factors on the reliability of the power system is provided, which comprises the following steps: step 1, selecting N links in a neural network with reliability indexes, wherein the links meet the following conditions: selecting a neuron in each layer of the neural network to form a link, wherein the selected neurons are directly connected layer by layer; and 2, respectively calculating the sensitivity indexes of a certain influence factor in each link, and calculating the average value of the sensitivity indexes of all the links as the sensitivity of the influence factor to the reliability index.
The method can effectively identify the size and relative importance of the sensitivity of each relevant factor of the reliability index under a given neural network model, and provides a valuable numerical analysis tool for the management of system reliability and system construction under a big data environment. The invention can effectively distinguish the influence factors into the beneficial factors and the adverse factors through the simulation experiment of the actual data, and quantitatively sorts the sensitivity importance degrees of the influence factors with the same property according to the sensitivity contribution to the indexes.
Drawings
FIG. 1 is a schematic diagram of an overall flow chart according to an embodiment of the invention;
FIG. 2 is a schematic diagram of a neural network of reliability indicators according to an embodiment of the present invention;
fig. 3 is a general flow chart diagram according to an embodiment of the invention.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
In fig. 1, a method for calculating the sensitivity of the influence factors on the reliability of the power system is shown in an embodiment of the present invention. In general, the method comprises the following steps: step 1, selecting N links in a neural network with reliability indexes, wherein the links meet the following conditions: selecting a neuron in each layer of the neural network to form a link, wherein the selected neurons are directly connected layer by layer; and 2, respectively calculating the sensitivity indexes of a certain influence factor in each link, and calculating the average value of the sensitivity indexes of all the links as the sensitivity of the influence factor to the reliability index.
In another embodiment of the invention, the data network is a BP neural network.
In another embodiment of the present invention, a method for calculating sensitivity of an influencing factor of reliability of an electric power system, where the reliability index can be: the electricity of the user is cut off for hours every year.
In another embodiment of the present invention, the influencing factors can be: the ratio of insulated feeders, the proportion of junctor lines, the number ratio of sections to feeders, the average number of users per section or the annual average feeder capacity-carrying ratio.
In fig. 2, in another embodiment of the present invention, a method for calculating sensitivity of influencing factors of power system reliability includes, as shown, hidden layers, the number of neurons, and transfer functions included in each network of the neural network of the reliability index.
In another embodiment of the present invention, a method for calculating sensitivity of an influencing factor of reliability of an electrical power system, in step 2, the sensitivity index of an influencing factor in each link is calculated respectively, and the method is implemented by the following steps. By means of the continuous derivative rule, an analytic relational expression between the reliability index and a single influence factor under the neural network with the given reliability index is deduced as follows: similar to the symbolic meaning shown in a three-layer neural network shown in fig. 2, for other neural networks including an input layer (layer 0), M hidden layers and an output layer, the input-output general relationship model of each layer is:
Figure DEST_PATH_GDA0001207655940000041
Figure DEST_PATH_GDA0001207655940000051
o(k+1)(j)＝h(k+1)(n(k+1)(i)) (2)
wherein o is(k+1)(j)＝I(j)，
Figure DEST_PATH_GDA0001207655940000052
h(k)() Representing the transfer function of the k-th layer.
According to the chain derivation rule of partial derivatives, starting from the last layer, i.e. M +1 layer, the partial derivative calculation is performed according to formula (2), that is to say, the following can be obtained:
Figure DEST_PATH_GDA0001207655940000053
wherein I is a reliability index, F is an influence factor, M is the number of hidden layers in the neural network of the reliability index, and I is the neuron number in each layer.
In another embodiment of the present invention, a method for calculating sensitivity of an influencing factor of reliability of an electric power system further includes, before step 1: carrying out primary amplification on small samples of each influence factor data of reliability by using a sample amplification technology to obtain a primary amplification matrix AE(ii) a Based on a primary amplification matrix AEAnd carrying out single-output neural network training.
Fig. 3 is a diagram illustrating a sensitivity calculation method according to another embodiment of the present invention, taking a user average power outage hour (TOH) indicator as an example, showing a sensitivity calculation method for an influencing factor of power system reliability. The method comprises the following steps: firstly, aiming at the reliability index TOH, three different BP neural network reliability models are selected, and the hidden layer, the number of neurons and the transfer function contained in each selected network are shown in the following table 1.
Figure DEST_PATH_GDA0001207655940000061
Secondly, training each network by using a Matlab neural network tool box and adopting a Levenberg Marquardtocation algorithm, and adopting default settings of software packages for other parameters; the definitions of the influencing factors and the reliability indicators and the related raw statistics are summarized in the following table.
Figure DEST_PATH_GDA0001207655940000062
And (III) after the original data is expanded by 100 times by utilizing a nuclear density fitting technology, controlling the training precision of each BP nerve to be that the mean square error of the sample prediction and the actual value is less than 0.019.
And (IV) according to a given artificial neural network reliability model and the definition of the neuron link, finding out the neuron link which meets the index TOH of the condition and can transfer the feeder proportion. Calculating the sensitivity index of the neuron link
Figure DEST_PATH_GDA0001207655940000063
And the sensitivity average value of all links is taken as an influence factor to transfer the sensitivity of the feeder proportion to the index TOH.
And (V) aiming at the reliability index TOH, the steps in the previous paragraph are sequentially and independently repeated, and the sensitivity of the influence factors of the reliability index TOH on the reliability index TOH, such as 'insulated feeder ratio', 'tie line ratio', 'number ratio of segments to feeders', 'average number of users per segment' and 'year-average feeder capacity-load ratio', can be obtained.
And (4) for each reliability index, sequentially and independently repeating the steps (three) to (five), so that the sensitivity of each reliability index to the influence factors of the reliability index can be obtained.
In another embodiment of the present invention, a method for calculating sensitivity of an influencing factor of reliability of an electric power system further includes, before step 1: carrying out primary amplification on small samples of each influence factor data of reliability by using a sample amplification technology to obtain a primary amplification matrix AE(ii) a Based on a primary amplification matrix AEAnd carrying out single-output neural network training.
In another embodiment of the present invention, a method for calculating sensitivity of influencing factors of reliability of an electrical power system, the training of each neural network includes: and training by using a Matlab neural network tool box and adopting a Levenberg Marquardtocation algorithm, wherein other parameters are set by default by adopting a software package.
In another embodiment of the present invention, a method for calculating sensitivity of influencing factors of reliability of an electric power system, where the step 1 uses a sample augmentation technique to perform primary augmentation on small samples of each influencing factor data of reliability, includes: the small samples were augmented using nuclear density fitting techniques.
In another embodiment of the invention, the method for calculating the sensitivity of the influencing factors of the reliability of the power system controls the training precision to be that the mean square error of the predicted and actual values of the samples is less than 0.019 for each BP nerve.
Finally, the method of the present application is only a preferred embodiment and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A method for calculating sensitivity of influencing factors of reliability of a power system is characterized by comprising the following steps:
step 1, selecting N links in a neural network with reliability indexes, wherein the links meet the following conditions: selecting a neuron in each layer of the neural network to form a link, wherein the selected neurons are directly connected layer by layer;
step 2, respectively calculating the sensitivity indexes of a certain influence factor in each link, and calculating the sensitivity index average value of all links as the sensitivity of the influence factor to the reliability index;
the calculating the sensitivity index of an influencing factor in each link respectively includes:
Figure FDA0002549706000000011
wherein
Figure FDA0002549706000000012
Is a sensitivity index, I is a reliability index, F is an influence factor, M is the number of hidden layers in the neural network of the reliability index, and I is a nerve in each layerThe element number.
2. The method of claim 1, in which the neural network is a BP neural network.
3. The method of claim 1, wherein the reliability indicator can be: the electricity of the user is cut off for hours every year.
4. The method of claim 1, wherein the influencing factors can be: the ratio of insulated feeders, the proportion of junctor lines, the number ratio of sections to feeders, the average number of users per section or the annual average feeder capacity-carrying ratio.
5. The method of claim 1, wherein step 1 is preceded by: carrying out primary amplification on small samples of each influence factor data of reliability by using a sample amplification technology to obtain a primary amplification matrix AE(ii) a Based on a primary amplification matrix AEAnd carrying out single-output neural network training.
6. The method of claim 5, wherein the neural network training comprises: and training by using a Matlab neural network tool box and adopting a Levenberg Marquardt optimization algorithm, wherein other parameters are set by default by adopting a software package.
7. The method of claim 5, wherein the step 1 of performing a preliminary amplification on each small sample of reliability factor data by using a sample amplification technique comprises: the small samples were augmented using nuclear density fitting techniques.
8. The method of claim 5, wherein for each BP nerve, the training precision is controlled such that the mean square error of the predicted and actual values of the samples is less than 0.019.
CN201611031272.3A 2016-11-17 2016-11-17 Method for calculating sensitivity of reliability influence factors of power system Active CN106777494B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611031272.3A CN106777494B (en) 2016-11-17 2016-11-17 Method for calculating sensitivity of reliability influence factors of power system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611031272.3A CN106777494B (en) 2016-11-17 2016-11-17 Method for calculating sensitivity of reliability influence factors of power system

Publications (2)

Publication Number Publication Date
CN106777494A CN106777494A (en) 2017-05-31
CN106777494B true CN106777494B (en) 2020-11-03

Family

ID=58971542

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611031272.3A Active CN106777494B (en) 2016-11-17 2016-11-17 Method for calculating sensitivity of reliability influence factors of power system

Country Status (1)

Country Link
CN (1) CN106777494B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110414086B (en) * 2019-07-10 2023-01-17 北京华安中泰检测技术有限公司 Sensitivity-based comprehensive stress acceleration factor calculation method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101635457A (en) * 2009-05-14 2010-01-27 国家电网公司 Electric network parameter estimation method based on parameter sensitivity of state estimation residual error
CN103384069A (en) * 2013-06-13 2013-11-06 广东电网公司电力科学研究院 Method and system for adjusting sensitivity of offshore wind plant current collecting system
CN105932775A (en) * 2016-05-23 2016-09-07 天津大学 Analysis method for influence from information system to operational reliability of micro-grid

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101635457A (en) * 2009-05-14 2010-01-27 国家电网公司 Electric network parameter estimation method based on parameter sensitivity of state estimation residual error
CN103384069A (en) * 2013-06-13 2013-11-06 广东电网公司电力科学研究院 Method and system for adjusting sensitivity of offshore wind plant current collecting system
CN105932775A (en) * 2016-05-23 2016-09-07 天津大学 Analysis method for influence from information system to operational reliability of micro-grid

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
发输电互联系统风险评估的解析模型研究;周竞;《中国优秀硕士学位论文全文数据库（工程科技II辑）》;20130615(第6期);全文 *
基于BP神经网络的城网供电可靠性预测方法;宋云亭 等;《电网技术》;20081031;第32卷(第20期);摘要，正文第2小节 *

Also Published As

Publication number Publication date
CN106777494A (en) 2017-05-31

Similar Documents

Publication Publication Date Title
CN110311376B (en) Dynamic safety assessment comprehensive model and space-time visualization method for power system
CN102722759B (en) Method for predicting power supply reliability of power grid based on BP neural network
CN103198235B (en) Based on the wind power prediction value Pre-Evaluation method of the longitudinal moment probability distribution of wind power
CN106329516A (en) Typical scene recognition based dynamic reconstruction method of power distribution network
CN104037943B (en) A kind of voltage monitoring method and system that improve grid voltage quality
CN112348339A (en) Power distribution network planning method based on big data analysis
CN105740975A (en) Data association relationship-based equipment defect assessment and prediction method
CN101888087A (en) Method for realizing distributed super-short-term area load forecasting in distribution network terminal
CN107633320B (en) Power grid line importance degree evaluation method based on meteorological prediction and risk evaluation
CN103150635B (en) Power equipment O&M method
CN104077651B (en) Maintenance scheduling for power systems optimization method
CN109782124B (en) Main distribution integrated fault positioning method and system based on gradient descent algorithm
CN104408562A (en) Photovoltaic system generating efficiency comprehensive evaluation method based on BP (back propagation) neural network
CN103971175B (en) Short-term load prediction method of multistage substations
CN102982393A (en) Online prediction method of electric transmission line dynamic capacity
CN111680841B (en) Short-term load prediction method, system and terminal equipment based on principal component analysis
LU500551B1 (en) Virtual load dominant parameter identification method based on incremental learning
CN104299031A (en) Ultra-short-term load prediction method of BP neural network
CN113191687B (en) Elastic power distribution network panoramic information visualization method and system
CN110601204A (en) Photovoltaic grid-connected system probability power flow analysis method based on random variable state time sequence simulation
CN104299043A (en) Ultra-short-term load prediction method of extreme learning machine
CN107145968A (en) Photovoltaic apparatus life cycle cost Forecasting Methodology and system based on BP neural network
Ma et al. Economic operation evaluation of active distribution network based on fuzzy Borda method
CN110910026B (en) Cross-provincial power transmission line loss intelligent management and decision method and system
CN108694479A (en) Consider the distribution network reliability prediction technique that weather influences time between overhaul

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
CB03 Change of inventor or designer information
CB03 Change of inventor or designer information

Inventor after: Zhang Jiangong

Inventor after: Hu Qinghui

Inventor after: Tian Hongxun

Inventor after: Wang Honggang

Inventor after: Wan Tao

Inventor after: Li Haosong

Inventor after: Li Jin

Inventor after: Kang Taifeng

Inventor after: Li Xin

Inventor before: Zhang Jiangong

Inventor before: Qing Hui

Inventor before: Tian Hongxun

Inventor before: Wang Honggang

Inventor before: Wan Tao

Inventor before: Li Haosong

Inventor before: Li Jin

Inventor before: Kang Taifeng

Inventor before: Li Xin

GR01 Patent grant
GR01 Patent grant