CN108333504A - A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm - Google Patents

A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm Download PDF

Info

Publication number
CN108333504A
CN108333504A CN201810113553.6A CN201810113553A CN108333504A CN 108333504 A CN108333504 A CN 108333504A CN 201810113553 A CN201810113553 A CN 201810113553A CN 108333504 A CN108333504 A CN 108333504A
Authority
CN
China
Prior art keywords
layer
node
parameter
output
high voltage
Prior art date
Application number
CN201810113553.6A
Other languages
Chinese (zh)
Inventor
廖敏夫
段雄英
黄智慧
邹积岩
张帆
Original Assignee
大连理工大学
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 大连理工大学 filed Critical 大连理工大学
Priority to CN201810113553.6A priority Critical patent/CN108333504A/en
Publication of CN108333504A publication Critical patent/CN108333504A/en

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/327Testing of circuit interrupters, switches or circuit-breakers

Abstract

The invention belongs to intelligent switch appliance field, it is related to a kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference and closes performance prediction algorithm.Steps are as follows:The first step:The foundation of extra high voltage line failure rate model;Second step:Each parameter is determined by hybrid algorithm, determines premise parameter and consequent parameter;Third walks:Different line failure rates require the performance parameter requirement of lower phase selection breaker;It is analyzed according to the performance parameter of the complete extra high voltage line failure rate model of foundation, the breaker under requiring circuit different faults rate.The beneficial effects of the invention are as follows the model systems proposed to provide a large amount of parallel and distributed process, and extensive attribute especially learns attribute, and optimizes network using back-propagation algorithm.Language message can be used, itself can also be adapted to using numerical data, to realize that better extra high voltage line phase selection closes failure rate estimated performance, and has the characteristics that the calculating time is short.

Description

A kind of closing property of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference It can prediction algorithm

Technical field

The invention belongs to intelligent switch appliance fields, are related to a kind of extra-high voltage based on Adaptive Neuro-fuzzy Inference Circuit phase selection closes performance prediction algorithm.

Background technology

One of the principal element for influencing UHV transmission line Insulation Coordination is the switching overvoltage of transmission line of electricity.Due to spy High-pressure system connects low reactance, new transformer fe core material, the measures such as breaker of not restriking using neutral point, will cut off non-loaded line Road, excision no-load transformer and off-the-line overvoltage are inhibited or are reduced one by one, so that nonloaded line is closed a floodgate and overlapped Lock rises to principal contradiction, and as the deciding factor of selection extra-high voltage system operatio impact dielectric level.

Controlled switching technology closes phase combined floodgate by controlling breaker in target, can inhibit nonloaded line from principle Closing overvoltage.The determination of optimum phase is one of the key problem of controlled switching technology, and influences the master of best switching-on phase It includes prebreakdown and the mechanical scatter of breaker to want factor, thus by phase control techniques be applied to extra-high voltage system in when need examine Consider influence of the breaker performance parameter to failure rate, solves the Quantitatively mapping relationship of breaker performance parameter and line failure rate.

Manufacturer for phased breaker and user, it is to have to assess control effect before device fabrication and use very much Meaning.In order to improve manufacture efficiency, reduce energy consumption, raising product quality, manufacturer needs according to different user to failure rate Requirement adjust switch performance parameter in time.User can also be under the premise of known switching characteristic parameter, and forecast assessment is phased The spot effect of breaker, and then select suitable product.

Invention content

The technical problem to be solved by the present invention is to the relationships for extra high voltage line phase selection switch technology and line failure rate Extra high voltage line failure rate model is established, to realize breaker performance parameter, insulator chain operation wave withstanding voltage and circuit event The Quantitatively mapping relationship of barrier rate.

Due to can not directly obtain line failure rate f with close coefficient k, 3 σ of mechanical scatter and insulator chain operation wave it is resistance to By voltage uIt is resistance toBetween analytical expression, the present invention proposes a kind of mathematical model that can characterize relation above --- adaptive god Through fuzzy inference system (adaptive-network-based fuzzy inference system, ANFIS).Based on ANFIS It establishes one three and inputs then Sugeno fuzzy models of requiring to report his or her problems within a prescribed time and in a prescribed place, pass through rule and membership function that gridding method generates fuzzy system. Using in MATLAB softwares Mathworks and anfisedit complete the foundation of all models, coefficient k, machine are closed with breaker Tool 3 σ of dispersibility and insulator chain operation wave withstanding voltage uIt is resistance toIt is real using line failure rate f as output parameter as input parameter The Quantitatively mapping relationship of existing breaker performance parameter, insulator chain operation wave withstanding voltage and line failure rate.

Technical scheme of the present invention:

A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm, and step is such as Under:

The first step:The foundation of extra high voltage line failure rate model

1st layer:Three input variables are blurred, three input variables are to close coefficient k, 3 σ of mechanical scatter and insulation Substring operates wave withstanding voltage uIt is resistance to, the input of this layer is the sample data set of actual load, exports being subordinate to for each input variable Degree, output form are:

Wherein:

O11Indicate the output of the 1st layer of the 1st node;

O12Indicate the output of the 1st layer of the 2nd node;

O13Indicate the output of the 1st layer of the 3rd node;

O14Indicate the output of the 1st layer of the 4th node;

O15Indicate the output of the 1st layer of the 5th node;

O16Indicate the output of the 1st layer of the 6th node;

And PiIt is three corresponding Fuzzy Linguistic Variables of input variable;

Function mu is the membership function for the condition that meets, and membership function uses Gauss (Gauss) function, expression formula point It is not:

Wherein:Fuzzy membership function { ai, bi, ci, di, ei, giBe the i-th node layer parameter set;ai, bi, ci, di, ei, giReferred to as premise parameter,The membership function of coefficient k is closed for input variable;For the person in servitude of 3 σ of mechanical scatter Category degree function;μPi(z) it is that insulator chain operates wave withstanding voltage uIt is resistance toMembership function;

2nd layer:Realize that the rule intensity of the 2nd layer of i-th of node of the 1st layer of product calculation respectively exported is ωi, the 2nd layer I-th of node, which exports, is:

3rd layer:To the rule intensity ω of the 2nd layer of the 1st node1With the rule intensity ω of the 2nd layer of the 2nd node2Returned One change is handled, the output O of i-th of node of third layer3iFor:

4th layer:4th layer of each node i is the adaptive node with node function, node function fiIt is linear Function, fi=pix+qiy+siz+ti, the 4th layer of i-th of node output O4iFor:

Wherein:Fuzzy parameter pi、qi、siAnd tiFor consequent parameter, i.e. Takagi-Sugeno linear equations parameter, pass through The least-squares estimation algorithm of ANFIS determines.

5th layer:This layer is the stationary nodes for carrying ∑, and function is i-th of node output O for calculating the 4th layer4i's Summation, using total output O as system5, i.e. line failure rate f:

Second step:Premise parameter a is determined by hybrid algorithmi, bi, ci, di, ei, giWith consequent parameter pi, qi, si, ti;Tool Steps are as follows for body:

1) empirical value of three input variables is substituted into extra high voltage line failure rate model, it is in each iteration, first solid Determine premise parameter, the 1st layer of output continues to transmit along network forward direction, until the 4th layer, then use least-squares estimation algorithm pair Consequent parameter carries out calculating adjustment, obtains the consequent parameter for meeting the requirements value, and the 4th layer of output continuation is forward propagated to along network Up to total output O5

2) by known output empirical value and obtained total output O5Subtract each other to obtain error amount, error amount will reversely be passed along network It broadcasts, fixed consequent parameter adjusts l layers of premise parameter using gradient descent method;

3) first two steps are repeated, until error amount meets the requirements value, the premise parameter a that will be obtainedi, bi, ci, di, ei, giAnd knot By parameter pi, qi, si, tiIt substitutes into formula (6), obtains complete extra high voltage line failure rate model;

4) it closes coefficient k and 3 σ of mechanical scatter is substituted into complete extra high voltage line failure rate model, calculate total defeated Go out O5, i.e. failure rate f.

Third walks:Different line failure rates require the performance parameter requirement of lower phase selection breaker

According to the complete extra high voltage line failure rate model of foundation, to the breaker under the requirement of circuit different faults rate Performance parameter is analyzed.

1) insulator chain operation wave withstanding voltage u is setIt is resistance toFor certain value, breaker is closed into coefficient k in a certain section N parts are divided into, 3 σ of mechanical scatter of breaker is divided into M parts, obtains the input matrix of N × M.

2) using the matrix as input, backwards calculation is carried out based on complete extra high voltage line failure rate model, is worked as Corresponding breaker closes 3 σ of coefficient k and mechanical scatter when line failure rate f variations.

A large amount of parallel and distributed process is provided the beneficial effects of the invention are as follows the model system proposed, extensive attribute, Especially learn attribute, and optimizes network using back-propagation algorithm.ANFIS frameworks can not only use language message, also Itself can be adapted to using numerical data, to realize that better extra high voltage line phase selection closes failure rate estimated performance, and And have the characteristics that the calculating time is short.

Description of the drawings

Fig. 1 is that three inputs are required to report his or her problems within a prescribed time and in a prescribed place the ANFIS structures of then Sugeno fuzzy models.

Fig. 2 is the relationship that line failure rate and breaker close coefficient, mechanical scatter.

Specific embodiment

Below in conjunction with technical solution and attached drawing, the implementation process of this patent is illustrated.

As shown in Figure 1 and Figure 2, a kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes Performance prediction algorithm, steps are as follows:

The first step:The foundation of extra high voltage line failure rate model

1st layer:Three input variables are blurred, three input variables close coefficient k, 3 σ of mechanical scatter and insulator String operation wave withstanding voltage uIt is resistance to, the input of this layer is the sample data set of actual load, is exported as the degree of membership of each input variable, Output form is:

Wherein:O11Indicate the output of the 1st layer of the 1st node;O12Indicate the output of the 1st layer of the 2nd node;O13Indicate the 1st The output of the 3rd node of layer;O14Indicate the output of the 1st layer of the 4th node;O15Indicate the output of the 1st layer of the 5th node;O16Table Show the output of the 1st layer of the 6th node;φki、φσiAnd PiBe close coefficient k, 3 σ of mechanical scatter and insulator chain operation wave it is resistance to By voltage uIt is resistance toThe corresponding Fuzzy Linguistic Variable of these three nodes;

Function mu is the membership function for the condition that meets, and membership function uses Gauss (Gauss) function, expression formula point It is not:

Wherein:Fuzzy membership function { ai, bi, ci, di, ei, giBe the i-th node layer parameter set;ai, bi, ci, di, ei, giReferred to as premise parameter,The membership function of coefficient k is closed for input variable;For the person in servitude of 3 σ of mechanical scatter Category degree function;μPi(z) it is that insulator chain operates wave withstanding voltage uIt is resistance toMembership function;

2nd layer:Realize that the rule intensity of the 2nd layer of i-th of node of the 1st layer of product calculation respectively exported is ωi, the 2nd layer I-th of node, which exports, is:

3rd layer:To the rule intensity ω of the 2nd layer of the 1st node1With the rule intensity ω of the 2nd layer of the 2nd node2Returned One change is handled, the output O of i-th of node of third layer3iFor:

4th layer:4th layer of each node i is the adaptive node with node function, node function fiIt is linear Function, fi=pix+qiy+siz+ti, the 4th layer of i-th of node output O4iFor:

Wherein:Fuzzy parameter pi、qi、siAnd tiFor consequent parameter, i.e. Takagi-Sugeno linear equations parameter, pass through The least-squares estimation algorithm of ANFIS determines.

5th layer:This layer is the stationary nodes for carrying ∑, and function is i-th of node output O for calculating the 4th layer4i's Summation, using total output O as system5, i.e. line failure rate f:

Second step:Premise parameter a is determined by hybrid algorithmi, bi, ci, di, ei, giWith consequent parameter pi, qi, si, ti;Tool Steps are as follows for body:

1) empirical value of three input variables is substituted into extra high voltage line failure rate model, it is in each iteration, first solid Determine premise parameter, the 1st layer of output continues to transmit along network forward direction, until the 4th layer, then use least-squares estimation algorithm pair Consequent parameter carries out calculating adjustment, obtains determining that the consequent parameter for meeting the requirements value, the 4th layer of output continue to pass along network forward direction It is multicast to up to total output O5

2) by known output empirical value and obtained total output O5Subtract each other to obtain error amount, error amount will reversely be passed along network It broadcasts, fixed consequent parameter adjusts l layers of premise parameter using gradient descent method;

3) first two steps are repeated, until error amount meets the requirements value, the premise parameter a that will be obtainedi, bi, ci, di, ei, giAnd knot By parameter pi, qi, si, tiIt substitutes into formula (6), obtains complete extra high voltage line failure rate model;

4) it closes coefficient k and 3 σ of mechanical scatter is substituted into complete extra high voltage line failure rate model, calculate total defeated Go out O5, i.e. failure rate f.

Third walks:Different line failure rates require the performance parameter requirement of lower phase selection breaker

According to the complete extra high voltage line failure rate model of foundation, to the breaker under the requirement of circuit different faults rate Performance parameter is analyzed.

1) insulator chain operation wave withstanding voltage u is setIt is resistance toFor certain value, breaker is closed into coefficient k in a certain section N parts are divided into, 3 σ of mechanical scatter of breaker is divided into M parts, obtains the input matrix of N × M.

2) using the matrix as input, backwards calculation is carried out based on complete extra high voltage line failure rate model, is worked as Corresponding breaker closes 3 σ of coefficient k and mechanical scatter when line failure rate f variations.

By taking 1000kV UHV transmission lines as an example, under being required using the ANFIS model analysis different faults rates of foundation The performance parameter of breaker obtains as shown in Fig. 2 as a result, obtaining as drawn a conclusion:

(1) when closing maximum coefficient k, 3 σ minimums of mechanical scatter, line failure rate is preferably minimized;

(2) when closing coefficient k and rising to 0.7, the downward trend of failure rate slows down, this value is considered as closing The minimum value of coefficient k.

Claims (1)

1. a kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm, feature exists In steps are as follows:
The first step:The foundation of extra high voltage line failure rate model
1st layer:Three input variables are blurred, three input variables are to close coefficient k, 3 σ of mechanical scatter and insulator chain Operate wave withstanding voltage uIt is resistance to, the input of this layer is the sample data set of actual load, is exported as the degree of membership of each input variable, defeated Going out form is:
Wherein:
O11Indicate the output of the 1st layer of the 1st node;
O12Indicate the output of the 1st layer of the 2nd node;
O13Indicate the output of the 1st layer of the 3rd node;
O14Indicate the output of the 1st layer of the 4th node;
O15Indicate the output of the 1st layer of the 5th node;
O16Indicate the output of the 1st layer of the 6th node;
And PiIt is three corresponding Fuzzy Linguistic Variables of input variable;
Function mu is the membership function for the condition that meets, and membership function is respectively using Gauss (Gauss) function, expression formula:
Wherein:Fuzzy membership function { ai, bi, ci, di, ei, giBe the i-th node layer parameter set;ai, bi, ci, di, ei, giClaim Premised on parameter,The membership function of coefficient k is closed for input variable;For the degree of membership of 3 σ of mechanical scatter Function;μPi(z) it is that insulator chain operates wave withstanding voltage uIt is resistance toMembership function;
2nd layer:Realize that the rule intensity of the 2nd layer of i-th of node of the 1st layer of product calculation respectively exported is ωi, i-th of the 2nd layer Node exports:
3rd layer:To the rule intensity ω of the 2nd layer of the 1st node1With the rule intensity ω of the 2nd layer of the 2nd node2It is normalized Processing, the output O of i-th of node of third layer3iFor:
4th layer:4th layer of each node i is the adaptive node with node function, node function fiFor linear function, fi=pix+qiy+siz+ti, the 4th layer of i-th of node output O4iFor:
Wherein:Fuzzy parameter pi、qi、siAnd tiFor consequent parameter, i.e. Takagi-Sugeno linear equations parameter, pass through ANFIS's Least-squares estimation algorithm determines;
5th layer:This layer is the stationary nodes for carrying ∑, and function is i-th of node output O for calculating the 4th layer4iIt is total With using total output O as system5, i.e. line failure rate f:
Second step:Premise parameter a is determined by hybrid algorithmi, bi, ci, di, ei, giWith consequent parameter pi, qi, si, ti;Specific step It is rapid as follows:
1) empirical value of three input variables is substituted into extra high voltage line failure rate model, in each iteration, before first fixing Parameter is put forward, the 1st layer of output continues to transmit along network forward direction, until the 4th layer, then use least-squares estimation algorithm to conclusion Parameter carries out calculating adjustment, obtains the consequent parameter for meeting the requirements value, and the 4th layer of output continues to reach along network forward-propagating total Export O5
2) by known output empirical value and obtained total output O5Subtract each other to obtain error amount, error amount will along network backpropagation, Gu Determine consequent parameter, l layers of premise parameter is adjusted using gradient descent method;
3) first two steps are repeated, until error amount meets the requirements value, the premise parameter a that will be obtainedi, bi, ci, di, ei, giJoin with conclusion Number pi, qi, si, tiIt substitutes into formula (6), obtains complete extra high voltage line failure rate model;
4) it closes coefficient k and 3 σ of mechanical scatter is substituted into complete extra high voltage line failure rate model, calculate total output O5, That is failure rate f;
Third walks:Different line failure rates require the performance parameter requirement of lower phase selection breaker
According to the complete extra high voltage line failure rate model of foundation, the performance of the breaker under requiring circuit different faults rate Parameter is analyzed;
1) insulator chain operation wave withstanding voltage u is setIt is resistance toFor certain value, breaker is closed into coefficient k decile in a certain section It it is N parts, 3 σ of mechanical scatter of breaker is divided into M parts, obtains the input matrix of N × M;
2) using the matrix as input, backwards calculation is carried out based on complete extra high voltage line failure rate model, obtains working as circuit Corresponding breaker closes 3 σ of coefficient k and mechanical scatter when failure rate f variations.
CN201810113553.6A 2018-02-05 2018-02-05 A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm CN108333504A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810113553.6A CN108333504A (en) 2018-02-05 2018-02-05 A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810113553.6A CN108333504A (en) 2018-02-05 2018-02-05 A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm

Publications (1)

Publication Number Publication Date
CN108333504A true CN108333504A (en) 2018-07-27

Family

ID=62926988

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810113553.6A CN108333504A (en) 2018-02-05 2018-02-05 A kind of extra high voltage line phase selection based on Adaptive Neuro-fuzzy Inference closes performance prediction algorithm

Country Status (1)

Country Link
CN (1) CN108333504A (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150188415A1 (en) * 2013-12-30 2015-07-02 King Abdulaziz City For Science And Technology Photovoltaic systems with maximum power point tracking controller

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150188415A1 (en) * 2013-12-30 2015-07-02 King Abdulaziz City For Science And Technology Photovoltaic systems with maximum power point tracking controller

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
周沛洪 等: "取消750 kV线路断路器合闸电阻的研究", 《电网与水力发电进展》 *
林圣 等: "基于ANFIS的特高压输电线路故障分类识别方法", 《西南交通大学学报》 *
谢将剑 等: "基于永磁操动机构的同步关合关键技术的研究", 《高压电器》 *

Similar Documents

Publication Publication Date Title
Tang et al. An evidential reasoning approach to transformer condition assessments
Liao et al. A fuzzy-expert system for classifying power quality disturbances
Thangaraj et al. Optimal coordination of over-current relays using modified differential evolution algorithms
Bansal Optimization methods for electric power systems: An overview
Kisi et al. Forecasting daily lake levels using artificial intelligence approaches
He et al. A comparative study of artificial neural network, adaptive neuro fuzzy inference system and support vector machine for forecasting river flow in the semiarid mountain region
Hariri et al. A fuzzy logic based power system stabilizer with learning ability
Luk et al. A study of optimal model lag and spatial inputs to artificial neural network for rainfall forecasting
Firat et al. Hydrological time‐series modelling using an adaptive neuro‐fuzzy inference system
Chang et al. Adaptive neuro-fuzzy inference system for prediction of water level in reservoir
Guney et al. Comparison of Mamdani and Sugeno fuzzy inference system models for resonant frequency calculation of rectangular microstrip antennas
Elmas et al. Modelling of a nonlinear switched reluctance drive based on artificial neural networks
Saffar et al. A new fuzzy optimal reconfiguration of distribution systems for loss reduction and load balancing using ant colony search-based algorithm
Ata et al. An adaptive neuro-fuzzy inference system approach for prediction of tip speed ratio in wind turbines
CN103778575B (en) Transformer state evaluation method and system
Esmin et al. Hybrid evolutionary algorithm based on PSO and GA mutation
CN102289590B (en) Method for estimating operating state of SF6 high-voltage circuit breaker and intelligent system
Bisht et al. Discharge modelling using adaptive neuro-fuzzy inference system
De Oliveira et al. Artificial immune systems applied to the reconfiguration of electrical power distribution networks for energy loss minimization
Mathur et al. A comprehensive analysis of intelligent controllers for load frequency control
Jamal et al. Power Flow Control of Power Systems Using UPFC Based on Adaptive Neuro Fuzzy
Lee et al. An enhanced Lagrangian neural network for the ELD problems with piecewise quadratic cost functions and nonlinear constraints
Yazdanbakhsh et al. A systematic review of complex fuzzy sets and logic
CN104794534B (en) A kind of power grid security Tendency Prediction method based on improvement deep learning model
Ghorbani et al. A comparative study of artificial neural network (MLP, RBF) and support vector machine models for river flow prediction

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