CN106485705A - Power equipment infrared image abnormality recognition method based on support matrix machine - Google Patents

Power equipment infrared image abnormality recognition method based on support matrix machine Download PDF

Info

Publication number
CN106485705A
CN106485705A CN201610876909.2A CN201610876909A CN106485705A CN 106485705 A CN106485705 A CN 106485705A CN 201610876909 A CN201610876909 A CN 201610876909A CN 106485705 A CN106485705 A CN 106485705A
Authority
CN
China
Prior art keywords
infrared image
power equipment
hyperplane
normal
iteration
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201610876909.2A
Other languages
Chinese (zh)
Inventor
曹晖
刘尚
张盼盼
于雅洁
闫大鹏
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.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
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 Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN201610876909.2A priority Critical patent/CN106485705A/en
Publication of CN106485705A publication Critical patent/CN106485705A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G06T7/0006—Industrial image inspection using a design-rule based approach
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10048—Infrared image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Closed-Circuit Television Systems (AREA)
  • Image Analysis (AREA)

Abstract

The present invention discloses a kind of method diagnosing its running status abnormal conditions based on the infrared temperature image of power equipment, comprises the following steps：N is opened little greatly identical power equipment infrared image and carries out gray scale normalization, and add label as training sample by its failure condition；Build support matrix machine, using normalization sample matrix and label as input information, updated by alternating direction multiplier method iteration and calculate Optimal Separating Hyperplane parameter；Optimal Separating Hyperplane model will be substituted into after forecast sample gray scale normalization, be calculated its running status value, predicted the outcome；The present invention can accurately judge power equipment running status abnormal conditions, remind manager to make in time and process it is ensured that power equipment safety reliably runs accordingly.

Description

Power equipment infrared image abnormality recognition method based on support matrix machine
Technical field
The present invention relates to power equipment infrared image anomalous identification technical field is and in particular to electricity based on support matrix machine Just whether power equipment infrared image abnormality recognition method, judge its running status based on the infrared temperature image of power equipment Often.
Background technology
The developing rapidly of continuous expansion with China's electrical network scale and intelligent grid, efficient, accurate power equipment Abnormality diagnostic method demand is further urgent, and the abnormity diagnosis of electrical equipment are all study hotspots all the time；Traditional electrically sets Standby abnormity diagnosis are typically termly to be measured the operational factor of electrical equipment using instrument, instrument etc. by maintenance person, to parameter Draw diagnostic result after being analyzed calculating, this method not only complex operation, real-time results cannot be drawn, and expend a large amount of Time and human cost；In recent years, infrared thermal imaging equipment is applied increasingly popular in power system, and infrared thermal imaging technique exists There is in electrical equipment abnormity diagnosis noncontact, do not shut down, do not have a power failure, but the diagnosis of most of electrical equipment is by patrolling Inspection Personnel Dependence micro-judgment abnormal it may occur that careless omission or erroneous judgement and waste substantial amounts of manpower and materials, time-consuming efficiency is low, The problems such as error rate height, monitoring time length.
Develop towards intelligent direction with transformer station, the abnormity diagnosis technology of power equipment is also sent out towards automation direction Exhibition, image recognition technology is applied in the infrared image anomalous identification of power equipment.The infrared image of power equipment is known extremely Image Zhu Yaoshi be divided into abnormal and normal two classes, in numerous sorting algorithms, support vector machine have preferable classification effect Really, but its input be necessary for vector, and the infrared image of power equipment be matrix form.Matrix is launched into by traditional method Vector, then be trained predicting with support vector machine, but after matrix is launched into vector, destroy the entirety knot of image Structure, produces certain impact to predicting the outcome；Separately have feature extraction, extract from infrared image some feature composition characteristics to Amount is trained predicting, but feature extraction is difficult to extract preferable feature it is impossible to reduce the complete information of infrared image.Therefore Propose a kind of support matrix machine, directly input matrix image and be trained predicting, retain the complete letter of power equipment infrared image Breath, has the meaning of any to the abnormality diagnostic accuracy rate of infrared image improving power equipment.
Content of the invention
In order to solve some shortcomings in conventional electric power equipment infrared image anomalous identification, the present invention proposes a kind of being based on and props up Hold the power equipment infrared image abnormality recognition method of matrix machine, support vector machine are expanded to support matrix machine by the method, can It is trained predicting to directly input matrix picture；Similar with support vector machine, structural classification hyperplane equation, and with replacing Direction multiplier method come to construct extension Lagrangian formulation be iterated update hyperplane equation parameter, finally give classification super flat Face equation, new infrared image matrix is substituted into parameter and is calculated predictive value, thus it is abnormal to realize power equipment infrared image The Accurate Diagnosis of situation.
Technical scheme comprises the following steps that：
Step 1：Collection n opens the infrared image { X of little greatly identical power equipment1,X2,...,XnAs training sample；
Step 2：Training sample infrared image is carried out as the following formula gray scale normalization：
Wherein XiRepresent i-th infrared image (its size is p × q), Xi(R) represent that the redness of i-th infrared image is divided Amount, Xi(G) green component of i-th infrared image, X are representedi(B) represent the blue component of i-th infrared image；Then basis The abnormal conditions of every power equipment infrared image add label yi∈ { -1,1 }, wherein yi=-1 expression the i-th width infrared image Power equipment ruuning situation is normal, yiThe power equipment ruuning situation of=1 expression the i-th width infrared image is abnormal, thus sets up Training sample data collection
Step 3：One hyperplane W of method construct using support matrix machineTX+b, infrared image is pressed abnormal conditions Classified, wherein W is the normal of hyperplane, b is the deviation of hyperplane, WTTransposition for W；
The normal W of hyperplane and the concrete solution procedure of deviation b are as follows：
(1) construct following solving equation：
And S-W=0
Wherein tr (WTW) represent the mark seeking matrix, C and τ is proportionality coefficient, S is that waiting of hyperplane normal is worth, | | S | |* For nuclear norm, [u]+Represent loss function；
(2) solve the constraint equation corresponding extension Lagrangian formulation L constructing with step (1)1：
Wherein ρ is penalty factor, and Λ is diagonal matrix；
(3) use alternating direction multiplier method iteration to update the normal W solving hyperplane and deviation b, comprise the following steps that：
1) input training sample data collectionThe iteration initialization value of hyperplane normalFor zero moment Battle array, iteration initialization parameterBased on diagonal be 1 matrix, iteration initialization parameter t(1)=1, iteration initialization is joined Number c(0)=0, penalty factor ρ ＞ 0, controlling elements η ∈ (0,1), iterationses k=1；
2) update kth time iterative parameter Λ(k)With kth time iterative characteristic value c(k)
WhereinW(k)For the normal of kth time iteration hyperplane, b(k)For The deviation of kth time iteration hyperplane, symbol DτRepresent that after entering school singular value decomposition to matrix, middle diagonal matrix deducts τ (i.e.Sτ[ΣA]=diag ([σ1(A)-τ]+,...,[σr(A)-τ]+))；
3) compare kth time iterative characteristic value c(k)With k-1 iteration adjustment eigenvalue η c(k-1)If, c(k)＜ η c(k-1)
Then
Otherwise
Wherein t(k+1)For+1 iteration factor of kth,For+1 iteration hyperplane normal of kth equivalence iterative value, For+1 iterative parameter iterative value of kth；
4) k=k+1, repeat step 2) -3) n time, finally give normal W and deviation b of hyperplane；
Step 4：Predict the abnormal conditions of new power equipment infrared image, with step 2, m is opened forecast sample and carries out ash Degree normalization obtains forecast sample data set
Step 5：Normal W according to the calculated hyperplane of step 3 and deviation b estimate output data ynew, such as following formula Shown：
WhereinFor mapping function,Represent that the power equipment of the i-th width infrared image runs Situation is normal,Then represent that the power equipment ruuning situation of the i-th width infrared image is abnormal.
The infrared image of power equipment is carried out normalizing gray processing by the present invention first, solves instruction with alternating direction multiplier method Practice the optimal classification hyperplane equation parameter of sample, finally by this Optimal Separating Hyperplane model, input new infrared image matrix, Output predictive value, thus realize the Accurate Diagnosis of power equipment infrared image abnormal conditions.
Described power equipment infrared image abnormality recognition method adopts FLIR T420+AT89S52 single-chip microcomputer to form electric power Unit exception monitor, carries out anomalous identification to infrared image.
Brief description
Fig. 1 is forecast sample infrared image, and wherein Fig. 1 a is abnormal transformator infrared image, and Fig. 1 b is that abnormal plug-in strip is infrared Image.
Specific embodiment
It is that the present invention will be described in more detail for example with reference to power equipment infrared image abnormity diagnosis.
The power equipment abnormality diagnostic method based on infrared image for the present invention, step is as follows：
Step 1：Infrared image { the X of 1000 little greatly power equipments for 300*300 of collection1,X2,...,X1000Conduct Training sample.
Step 2：Training sample infrared image is carried out as the following formula gray scale normalization：
Wherein XiRepresent i-th infrared image (its size is 100 × 100), Xi(R) represent the redness of i-th infrared image Component, Xi(G) green component of i-th infrared image, X are representedi(B) represent the blue component of i-th infrared image.Then root Abnormal conditions according to every power equipment infrared image add label yi∈ { -1,1 }, wherein yi=-1 expression the i-th width infrared image Power equipment ruuning situation normal, yiThe power equipment ruuning situation of=1 expression the i-th width infrared image is abnormal, thus builds Vertical training sample data collection
Step 3：One hyperplane W of method construct using support matrix machineTX+b, infrared image is pressed abnormal conditions Classified, wherein W is the normal of hyperplane, b is the deviation of hyperplane, WTTransposition for W.
The normal W of hyperplane and the concrete solution procedure of deviation b are as follows：
(1) construct following solving equation：
And S-W=0
Wherein tr (WTW) represent the mark seeking matrix, C and τ is proportionality coefficient, S is that waiting of hyperplane normal is worth, | | S | |* For nuclear norm, [u]+Represent loss function.
(2) solve the constraint equation corresponding extension Lagrangian formulation L constructing with step (1)1：
Wherein ρ is penalty factor, and Λ is diagonal matrix.
(3) use alternating direction multiplier method iteration to update the normal W solving hyperplane and deviation b, comprise the following steps that：
1) input training sample data collectionThe iteration initialization value of hyperplane normalFor zero moment Battle array, iteration initialization parameterBased on be diagonally 1 matrix, iteration initialization parameter t(1)=1, iteration initialization parameter c(0)=0, penalty factor ρ=10, controlling elements η=0.99, iterationses k=1.
2) update kth time iterative parameter Λ(k)With kth time iterative characteristic value c(k)
WhereinW(k)For the normal of kth time iteration hyperplane, b(k)For The deviation of kth time iteration hyperplane, symbol DτRepresent that after entering school singular value decomposition to matrix, middle diagonal matrix deducts τ (i.e.Sτ[ΣA]=diag ([σ1(A)-τ]+,...,[σr(A)-τ]+))
3) compare kth time iterative characteristic value c(k)With k-1 iteration adjustment eigenvalue η c(k-1)If, c(k)＜ η c(k-1)
Then
Otherwise
Wherein t(k+1)For+1 iteration factor of kth,For+1 iteration hyperplane normal of kth equivalence iterative value, For+1 iterative parameter iterative value of kth.
4) k=k+1, repeat step 2) -3) 500 times, finally give normal W and deviation b of hyperplane.
Step 4：Predict the abnormal conditions of new power equipment infrared image, as shown in figure 1, wherein Fig. 1 a is forecast sample Transformator infrared image, Fig. 1 b is forecast sample plug-in strip infrared image, with step 2, forecast sample is carried out gray scale normalization and obtains To forecast sample collection
Step 5：Normal W according to the calculated hyperplane of step 3 and deviation b estimate output data ynew, such as following formula Shown：
WhereinFor mapping function,Represent that the power equipment of the i-th width infrared image runs Situation is normal,Then represent that the power equipment ruuning situation of the i-th width infrared image is abnormal.
Finally giveShow plug-in strip in transformator and Fig. 1 b in Fig. 1 a, the infrared figure of two in figure power equipments As being in abnormality, it is consistent with practical situation.
Described power equipment infrared image abnormality recognition method adopts FLIR T420+AT89S52 single-chip microcomputer to form electric power Unit exception monitor, carries out anomalous identification to infrared image.

Claims (2)

1. a kind of power equipment infrared image abnormality recognition method based on support matrix machine it is characterised in that：Step is as follows：
Step 1：Collection n opens the infrared image { X of little greatly identical power equipment1,X2,...,XnAs training sample；
Step 2：Training sample infrared image is carried out as the following formula gray scale normalization：
Wherein XiRepresent i-th infrared image, its size is p × q, Xi(R) red component of i-th infrared image, X are representedi (G) green component of i-th infrared image, X are representedi(B) represent the blue component of i-th infrared image.Then according to every The abnormal conditions of power equipment infrared image add label yi∈ { -1,1 }, wherein yiThe electric power of=- 1 expression the i-th width infrared image Machine operation is normal, yiThe power equipment ruuning situation of=1 expression the i-th width infrared image is abnormal, thus sets up training Sample data set
Step 3：One hyperplane W of method construct using support matrix machineTX+b, infrared image is carried out point by abnormal conditions Class, wherein W are the normal of hyperplane, and b is the deviation of hyperplane, WTTransposition for W；
The normal W of hyperplane and the concrete solution procedure of deviation b are as follows：
(1) construct following solving equation：
And S-W=0
Wherein tr (WTW) represent the mark seeking matrix, C and τ is proportionality coefficient, S is that waiting of hyperplane normal is worth, | | S | |*For core Norm, [u]+Represent loss function；
(2) solve the constraint equation corresponding extension Lagrangian formulation L constructing with step (1)1：
Wherein ρ is penalty factor, and Λ is diagonal matrix；
(3) use alternating direction multiplier method iteration to update the normal W solving hyperplane and deviation b, comprise the following steps that：
1) input training sample data collectionThe iteration initialization value of hyperplane normalFor null matrix, iteration Initiation parameterBased on diagonal be 1 matrix, iteration initialization parameter t(1)=1, iteration initialization parameter c(0)= 0, penalty factor ρ ＞ 0, controlling elements η ∈ (0,1), iterationses k=1；
2) update kth time iterative parameter Λ(k)With kth time iterative characteristic value c(k)
WhereinW(k)For the normal of kth time iteration hyperplane, b(k)For kth time The deviation of iteration hyperplane, symbol DτRepresent that after entering school singular value decomposition to matrix, middle diagonal matrix deducts τ, that is,Sτ[ΣA]=diag ([σ1(A)-τ]+,...,[σr(A)-τ]+)；
3) compare kth time iterative characteristic value c(k)With k-1 iteration adjustment eigenvalue η c(k-1)If, c(k)＜ η c(k-1)Then
Otherwise
Wherein t(k+1)For+1 iteration factor of kth,For+1 iteration hyperplane normal of kth equivalence iterative value,For kth + 1 iterative parameter iterative value；
4) k=k+1, repeat step 2) -3) n time, finally give normal W and deviation b of hyperplane；
Step 4：Predict the abnormal conditions of new power equipment infrared image, with step 2, m is opened forecast sample carry out gray scale and return One change obtains forecast sample data set
Step 5：Normal W according to the calculated hyperplane of step 3 and deviation b estimate output data ynew, as following formula institute Show：
WhereinFor mapping function,Represent the power equipment ruuning situation of the i-th width infrared image Normally,Then represent that the power equipment ruuning situation of the i-th width infrared image is abnormal.
2. a kind of power equipment infrared image abnormality recognition method based on support matrix machine according to claim 1, its It is characterised by：Described power equipment infrared image abnormality recognition method adopts FLIRT420+AT89S52 single-chip microcomputer to form electric power Unit exception monitor, carries out anomalous identification to infrared image.
CN201610876909.2A 2016-10-08 2016-10-08 Power equipment infrared image abnormality recognition method based on support matrix machine Pending CN106485705A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610876909.2A CN106485705A (en) 2016-10-08 2016-10-08 Power equipment infrared image abnormality recognition method based on support matrix machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610876909.2A CN106485705A (en) 2016-10-08 2016-10-08 Power equipment infrared image abnormality recognition method based on support matrix machine

Publications (1)

Publication Number Publication Date
CN106485705A true CN106485705A (en) 2017-03-08

Family

ID=58268557

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610876909.2A Pending CN106485705A (en) 2016-10-08 2016-10-08 Power equipment infrared image abnormality recognition method based on support matrix machine

Country Status (1)

Country Link
CN (1) CN106485705A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107958292A (en) * 2017-10-19 2018-04-24 山东科技大学 Transformer based on cost sensitive learning obscures careful reasoning method for diagnosing faults
CN109493292A (en) * 2018-10-29 2019-03-19 平高集团有限公司 Enhancing treating method and apparatus based on power equipment infrared measurement of temperature image
CN110261703A (en) * 2019-07-08 2019-09-20 厦门理工学院 A kind of transformer fault method for early warning, terminal device and storage medium
CN113158230A (en) * 2021-03-16 2021-07-23 陕西数盾慧安数据科技有限公司 Online classification method based on differential privacy
WO2021243724A1 (en) * 2020-06-05 2021-12-09 北京嘀嘀无限科技发展有限公司 Image processing method, electronic device, vehicle traveling data recorder, and server

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103824092A (en) * 2014-03-04 2014-05-28 国家电网公司 Image classification method for monitoring state of electric transmission and transformation equipment on line
CN105139034A (en) * 2015-08-27 2015-12-09 北京市遥感信息研究所 Spectrum filtering based ship detection method
CN105300528A (en) * 2015-10-12 2016-02-03 国家电网公司 Infrared image diagnosis method and infrared image diagnosis system for transformer station equipment
CN105869408A (en) * 2016-04-27 2016-08-17 长安大学 Hub temperature anomaly detection and early warning method and system

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103824092A (en) * 2014-03-04 2014-05-28 国家电网公司 Image classification method for monitoring state of electric transmission and transformation equipment on line
CN105139034A (en) * 2015-08-27 2015-12-09 北京市遥感信息研究所 Spectrum filtering based ship detection method
CN105300528A (en) * 2015-10-12 2016-02-03 国家电网公司 Infrared image diagnosis method and infrared image diagnosis system for transformer station equipment
CN105869408A (en) * 2016-04-27 2016-08-17 长安大学 Hub temperature anomaly detection and early warning method and system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LUO LUO 等: "Support Matrix Machines", 《PROCEEDING OF THE 32ND INTERNATIONAL CONFERENCE ON MACHINE LEARNING》 *
解迎刚 等: "《远程教育中情感计算技术》", 31 May 2011 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107958292A (en) * 2017-10-19 2018-04-24 山东科技大学 Transformer based on cost sensitive learning obscures careful reasoning method for diagnosing faults
CN109493292A (en) * 2018-10-29 2019-03-19 平高集团有限公司 Enhancing treating method and apparatus based on power equipment infrared measurement of temperature image
CN110261703A (en) * 2019-07-08 2019-09-20 厦门理工学院 A kind of transformer fault method for early warning, terminal device and storage medium
WO2021243724A1 (en) * 2020-06-05 2021-12-09 北京嘀嘀无限科技发展有限公司 Image processing method, electronic device, vehicle traveling data recorder, and server
CN113158230A (en) * 2021-03-16 2021-07-23 陕西数盾慧安数据科技有限公司 Online classification method based on differential privacy
CN113158230B (en) * 2021-03-16 2024-02-09 陕西数盾慧安数据科技有限公司 Online classification method based on differential privacy

Similar Documents

Publication Publication Date Title
CN106485705A (en) Power equipment infrared image abnormality recognition method based on support matrix machine
CN110598736B (en) Power equipment infrared image fault positioning, identifying and predicting method
CN104595170B (en) A kind of air compressor machine monitoring and diagnosis system and method for self-adaptive kernel gauss hybrid models
CN103901072B (en) Method for diagnosing equipment overheating defects by utilizing infrared spectrum analysis
CN105404867B (en) A kind of substation isolating-switch state identification method of view-based access control model
CN107480440A (en) A kind of method for predicting residual useful life for modeling of being degenerated at random based on two benches
CN111751002B (en) Intelligent charged equipment fault diagnosis method based on infrared thermal imaging
CN104167818B (en) Intelligent remote inspection system in linkage with GIS substation integrated automation system and method
CN108921230A (en) Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network
CN103678869A (en) Prediction and estimation method of flight parameter missing data
CN105262441B (en) Infrared image-based photovoltaic array fault grading method
CN104199441B (en) Blast furnace multi-state fault separating method based on sparse contribution plot and system
CN108445435A (en) A kind of electric energy meter calibrating apparatus error online evaluation method
CN109409444A (en) A kind of method of discrimination of the polynary electric network fault type based on prior probability
CN105205503A (en) Crowdsourcing-active-learning-based method for detecting abnormal picture
CN109738723A (en) A kind of electric energy meter three-phase automatic identification method
CN116929815A (en) Equipment working state monitoring system and method based on Internet of things
CN116502160A (en) Automatic electric quantity data acquisition system
CN111143835B (en) Non-invasive protection method for business logic of electric power metering system based on machine learning
CN116824517B (en) Substation operation and maintenance safety control system based on visualization
CN114879628A (en) Multi-mode industrial process fault diagnosis method based on local maximum mean difference resistance
CN107358297A (en) A kind of fast indexing method of state-detection data
CN109509169A (en) The pressing plate condition checkout gear and method of learning distance metric based on nearest neighbour classification
CN109242008B (en) Compound fault identification method under incomplete sample class condition
CN106774190A (en) Wet cooling gas turbine group condenser vacuum exception auto-check system and method

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20170308

WD01 Invention patent application deemed withdrawn after publication