CN109617526A - A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM - Google Patents
A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM Download PDFInfo
- Publication number
- CN109617526A CN109617526A CN201811561729.0A CN201811561729A CN109617526A CN 109617526 A CN109617526 A CN 109617526A CN 201811561729 A CN201811561729 A CN 201811561729A CN 109617526 A CN109617526 A CN 109617526A
- Authority
- CN
- China
- Prior art keywords
- signal
- array
- classification
- fault diagnosis
- sample
- 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
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 34
- 238000004458 analytical method Methods 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000010248 power generation Methods 0.000 title claims abstract description 23
- 239000013598 vector Substances 0.000 claims abstract description 39
- 238000012549 training Methods 0.000 claims abstract description 32
- 230000008859 change Effects 0.000 claims abstract description 28
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 238000012706 support-vector machine Methods 0.000 claims abstract description 12
- 230000009466 transformation Effects 0.000 claims abstract description 10
- 238000005286 illumination Methods 0.000 claims abstract description 7
- 230000007257 malfunction Effects 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 4
- 230000005611 electricity Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 28
- 239000000306 component Substances 0.000 description 14
- 238000005457 optimization Methods 0.000 description 10
- 238000001514 detection method Methods 0.000 description 6
- 241001269238 Data Species 0.000 description 5
- 230000009977 dual effect Effects 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000002310 reflectometry Methods 0.000 description 3
- 238000001931 thermography Methods 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000011664 signaling Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- WSGCGMGMFSSTNK-UHFFFAOYSA-M 1-methyl-4-phenyl-1-propan-2-ylpiperidin-1-ium;iodide Chemical compound [I-].C1C[N+](C(C)C)(C)CCC1C1=CC=CC=C1 WSGCGMGMFSSTNK-UHFFFAOYSA-M 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02S—GENERATION OF ELECTRIC POWER BY CONVERSION OF INFRARED RADIATION, VISIBLE LIGHT OR ULTRAVIOLET LIGHT, e.g. USING PHOTOVOLTAIC [PV] MODULES
- H02S50/00—Monitoring or testing of PV systems, e.g. load balancing or fault identification
- H02S50/10—Testing of PV devices, e.g. of PV modules or single PV cells
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/50—Photovoltaic [PV] energy
Landscapes
- Testing Of Individual Semiconductor Devices (AREA)
Abstract
The present invention relates to a kind of method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM, the not electric current of equality of temperature illumination and voltage sample signal first under acquisition photovoltaic array different working condition；Then it is normalized to obtain current changing rate, voltage change ratio, power variation rate and conductance change rate；Then sliding is carried out to four obtained sample signals and takes window signal, and carry out the multiresolution analysis of wavelet transformation；Calculate 2 norms of the high-frequency signal of four window signal n-th layers decomposition, and then the feature vector that dimension is 4；Then multiple feature value vectors are obtained by the multiple groups sample signal of different working condition, and is divided into training data and test data；The parameter of support vector machines is finally set, and training fault diagnosis model, the accuracy of the classification of the model is detected with test data from training data.Ambient adaptability of the present invention is strong, there is stronger novelty, and trained model is able to achieve the fault diagnosis and classification of degree of precision.
Description
Technical field
The present invention relates to photovoltaic power generation fault diagnosises and sorting technique field, especially a kind of to be based on wavelet multiresolution analysis
With the method for the photovoltaic power generation array fault diagnosis and classification of SVM.
Background technique
The Development of Novel energy is the grand strategy of current the problems such as solving fossil energy exhaustion and environmental pollution, and solar energy
Because of its cleaning, the advantages such as energy abundance become the first choice for developing new energy.Photovoltaic power generation is the main side currently with solar energy
Formula, and the photovoltaic power generation array for acquiring solar energy is usually operated at family complicated and changeable as the core component of photovoltaic generating system
In external environment, vulnerable to the severe factor such as wind and frost sleet influence and generate such as short-circuit, open circuit, the failures such as shade.The production of failure
Life can reduce photovoltaic efficiency, cause unexpected energy waste and economic loss, and What is more can leave security risk, cause
Fire endangers personal safety.So the working condition to photovoltaic system is monitored, the failure of appearance is detected in time and is divided
Class simultaneously gives a warning, it will be able to maintenance be effectively performed, reduce because of photovoltaic array failure bring energy loss, prevent peace
Full accident has good economic and social benefit.
Domestic and foreign scholars successively propose a series of method for diagnosing faults, realize to different fault types
Detection and positioning.Typical fault detection method has Capacitive current measuring method, Time Domain Reflectometry analytic approach, infrared thermal imaging etc..This
Outside, with the fast development of artificial intelligence, scholars are proposed with support vector machines, neural network, and decision tree etc. is based on machine
The fault diagnosis scheme of device learning algorithm.Capacitive current measuring method is judged according to the measurement of the direct-to-ground capacitance to photovoltaic group string
Whether it occurs open circuit and positioning failure.Time Domain Reflectometry analytic approach is to inject a pulse to photovoltaic group string, to return signal
Shape and delay time, which are analyzed, to be judged in photovoltaic group string with this with the presence or absence of failure.Direct-to-ground capacitance mensuration and Time Domain Reflectometry
Analytic approach requires to carry out offline inspection, lacks real-time, can expend a large amount of manpower and financial resources in this way.And work in normal shape
Solar cell under state and malfunction has that there are the apparent temperature difference, thus can also be carried out using infrared thermography analysis method therefore
Barrier diagnosis.Although infrared thermography analysis method can efficiently carry out fault diagnosis, a large amount of infrared video camera must be equipped with,
Deficiency in economic performance, it is difficult to be promoted.What application was most at present is fault diagnosis and classification method based on machine learning algorithm.
This method has stronger self-learning capability, and strong robustness, accuracy rate is high, has become the hot spot of research.It is most of at present
Machine learning algorithm is to be based on maximum power point of photovoltaic array electric current IMPP, maximum power point voltage VMPP, short circuit current ISC,
Open-circuit voltage VOC and warm illumination G, the variables such as environment temperature T are trained study as feature, and exploring new feature vector is to grind
Study carefully the important research direction of diagnosing failure of photovoltaic array.Accurate fault diagnosis model generally requires the data of multidimensional, model instruction
The white silk time is long, and the continuous transformation of environment also brings challenge to the accuracy of model.
Summary of the invention
In view of this, the purpose of the present invention is to propose to a kind of photovoltaic power generation array based on wavelet multiresolution analysis and SVM
The method of fault diagnosis and classification, ambient adaptability is strong, there is stronger novelty, and trained model is able to achieve the event of degree of precision
Barrier diagnosis and classification.
The present invention is realized using following scheme: a kind of photovoltaic power generation array failure based on wavelet multiresolution analysis and SVM
The method of diagnosis and classification, comprising the following steps:
Step S1: the not electric current of equality of temperature illumination and voltage sample signal under acquisition photovoltaic array different working condition；
Step S2: the electric current and voltage signal of acquisition are normalized, and obtain current changing rate, voltage change
Four rate, power variation rate and conductance change rate sample signals；
Step S3: the sliding window that setting window size is L carries out sliding to four sample signals that step S2 is obtained and takes window
Message number carries out the multiresolution analysis of wavelet transformation to four kinds of window signals of acquirement；
Step S4: 2 norms of the high-frequency signal that four window signal n-th layers are decomposed are calculated；
Step S5: 2 models of the high-frequency signal for taking the n-th layer of window signal of four sample signals when failure occurs to decompose
Number is used as characteristic value, the feature vector that dimension is 4；
Step S6: obtaining multiple feature value vectors by the multiple groups sample signal of different working condition, wherein by m feature to
Amount is used as training data, using n feature vector as test data；
Step S7: being arranged the parameter of support vector machines, the training fault diagnosis model from training data, with test number
According to the accuracy for the classification for detecting the model.
Further, in step S1, the different working condition of photovoltaic array include: normal, 1 component short circuit of single group string,
One component open circuit of single group string and the 2 component short circuits of single group string；Wherein, one the 1 component short circuit of single group string, single group string component
Open circuit and the 2 component short circuits of single group string be malfunction, the sample signal of malfunction contain photovoltaic array by normally to
Failure arrives the stable process of failure again, captures the variation characteristic of array current and voltage when failure occurs.Wherein, electric current and electricity
Press acquisition time the window t=10s, sample frequency 200Hz of sample signal.
Further, step S2 the following steps are included:
Step S21: the electric current and voltage signal of acquisition are normalized: by array current divided by short circuit current,
By array voltage divided by open-circuit voltage；The influence of different illuminance and different temperatures, normalizing can be overcome by normalized
The formula of change is as follows:
iPV(t)=IPV(t)/ISC(t)；
vPV(t)=VPV(t)/VOC(t)
In formula, IPV(t) the array current sample signal of acquisition, I are indicatedSC(t) array short circuit current signal, i are indicatedPV(t)
Array current sample signal after indicating normalization, VPV(t) the array voltage sample signal of acquisition, V are indicatedOC(t) array is indicated
Open circuit signaling, vPV(t) the array voltage sample signal after normalization is indicated；Electric current and voltage letter after normalized
The variation tendency for number reflecting array signal under different working condition, highlights variation characteristic；
Step S22: current changing rate, voltage change ratio, power variation rate and conductance change rate four are obtained using following formula
Sample signal:
Di (t)=[iPV(n+1)-iPV(n)]/t；
Dv (t)=[vPV(n+1)-vPV(n)]/t；
Dp (t)=[iPV(n+1)v(n+1)-iPV(n)v(n)]/t；
Dg (t)=[iPV(n+1)/vPV(n+1)-iPV(n)/iPV(n)]/t；
In formula, di (t), dv (t), dp (t), dg (t) respectively describe the curent change of photovoltaic array different working condition
Rate, voltage change ratio, the signal of power variation rate and conductance change rate.
Further, in step S3, the window size L is 20, using DB small echo to four kinds of window signals of acquirement into
The wavelet transformation of 6 layers of Multiresolution Decomposition of row；The multiresolution analysis be with high pass and low-pass filter by signal decomposition at
The high frequency detail and approximation of multiple ranks.
Preferably, wavelet transformation is a kind of widely used digital signal processing method, it is suitable for analysis nonperiodic signal,
Time and frequency analysis can be carried out to signal simultaneously.Signal x (t) is discrete to turn to x (n), and the DWT of the signal can be asked by following formula
Solution:
In formula, ψ (t) is morther wavelet, and n is the points of signal x (n), a=a0 jIt is scale factor, b=kb0a0 jShift factor.
a0> 0 and be constant, b0≠ 0, j, k ∈ Z.
Multiresolution analysis is with the high frequency detail and approximation of high pass and low-pass filter by signal decomposition at multiple ranks
Value.Time-domain signal f (t) can be indicated by scaling function Φ (t) and wavelet function Ψ (t), be shown below:
In formula, dj(k) be decomposition at different levels detail coefficients, express high-frequency information, and aN(k) afterbody N grades of approximate system
Number expresses low-frequency information.
Further, in step S4,2 norms of the high-frequency signal of four window signal n-th layers decomposition are calculated using following formula:
In formula, djnIt indicates, N is indicated, djIndicate the high frequency coefficient of jth layer, fiIndicate the feature of i-th of signal extraction.
Further, in step S5, the feature vector that the dimension is 4 is f=[f1,f2,f3,f4]。
Further, in step S7, the parameter of the setting support vector machines specifically: by the penalty factor of SVM
1000 are set as, sets 5 for the sum of the distance γ of two different classes of supporting vectors to hyperplane.Specifically:
The basic thought of support vector machines is that an optimal separating hyper plane is found by the training sample set of linear separability
To realize the division of different classes of sample data.For nonlinear classification problem, sample can be mapped from luv space
To the high-dimensional feature space of a linear separability, to find a suitable optimal separating hyper plane.
Given training sample data collection, D={ xi,yi, i=1,2,3 ..., m, yi∈ { -1,1 }, wherein xiFor sample number
According to m is training sample sum, and d is the dimension of sample space, and yi is the corresponding label of sample.They can be optimal super by one
Plane is separated, which can be expressed as wTX+b=0, wherein w ∈ Rd, it is normal vector, determines the direction of hyperplane；b
∈ R, for be displaced item, determine hyperplane between origin at a distance from, the threshold value of classification.
Assuming that hyperplane can correctly classify training sample, then for { xi,yi∈ D, if yi=+1, then there is wTX+b >
0；If yi, then there is w in=- 1TX+b < 0.It enables:
Then several training sample points nearest apart from hyperplane set up the equal sign of above formula, they be referred to as support to
It measures (Support vectors, SVs), then the sum of the distance of two different classes of supporting vectors to hyperplane is
The distance is referred to as class interval.If it is desired that γ is maximum, then require | | w | |2Minimum, while requiring classifying face to all samples
This correct classification, needs to meet
yi(wTxi+ b) >=1, i=1,2,3 ..., l
Therefore, optimal separating hyper plane problem is sought, a quadratic programming problem can be converted into, optimization aim can be written as
s.t.yi(wTxi+ b) >=1, i=1,2,3 ..., l
For the training sample in sample space linear separability, can be divided by optimal separating hyper plane.Then,
In realistic task, the case where being usually present linearly inseparable, there are part sample datas to be unsatisfactory in training sample at this time
, there is the error centainly classified in formula (1).Therefore, by introducing slack variable ξi(ξi>=0) this is solved the problems, such as.Therefore, public
Formula (1) can be written as
yi(wTxi+b)≥1-ξi, i=1,2,3 ..., l
While maximizing interval, it is desirable that the sample for being unsatisfactory for constraint will lack as far as possible.Therefore, optimization object function can
To be rewritten as
Wherein,Referred to as penalty term, C are penalty factor.Therefore, optimal classification when available linearly inseparable
Face, referred to as broad category hyperplane, can be expressed as following optimization problem
s.t.yi(wTxi+b)≥1-ξi
ξi>=0, i=1,2,3 ..., l
Solving the optimization problem by method of Lagrange multipliers, can obtain its dual problem such as following formula:
0≤αi≤ C, i, j=1,2,3 ..., l
According to Karush-Kuhn-Tucker (KKT) condition, available αi(yi(wTxi+ b) -1)=0.If αi> 0,
Then corresponding sample point is located on largest interval boundary, is exactly supporting vector.Can then it pass throughIt solves
W, and according to yi(wTxi+ b) -1=0 solves b, wherein xiFor supporting vector, n is the quantity of supporting vector.Determine w and b it
Afterwards, it is as follows to obtain categorised decision function
Secondly, being directed to Nonlinear Classification problem, sample is mapped to the spy of more higher-dimension using kernel function by SVM from luv space
Space is levied, so that sample linear separability in this feature space.φ (x) is enabled to indicate the feature vector after mapping x, then in height
Optimization aim corresponding to hyperplane is represented by dimension space
s.t.yi(wTφ(xi)+b)≥1-ξi
ξi>=0, i=1,2,3 ..., l
Corresponding dual problem is
0≤αi≤ C, i, j=1,2,3 ..., l
To avoid calculating sample xiAnd xjInner product operation in higher dimensional space constructs kernel function K (), xiAnd xj?
The inner product of feature space is converted to the result calculated in original sample space by the function.K () is expressed as follows
K(xi,xj)=φ (xi)Tφ(xj)
Then previous formula is become
0≤αi≤ C, i, j=1,2,3 ..., l
Categorised decision function becomes at this time
Common kernel function mainly have Polynomial kernel function, radial base (Radial based kernel, RBF) kernel function,
Tanh (Sigmoid) kernel function etc..SVM is applied using RBF kernel function in the present invention.
RBF kernel function is expressed as
K(xi, x) and=exp (- γ | | xi-x||2)
The classifying quality of support vector machines will be largely dependent on the selection of C and γ, set C in the present invention
1000, γ are set as 10.With parameter SVM is set, fault grader is trained from the training data of construction, realizes photovoltaic
The fault diagnosis and classification of power generation array.
Compared with prior art, the invention has the following beneficial effects:
1, the present invention only acquires normal and failure electric current and voltage signal under not equality of temperature illumination, thus obtains electric current
Change rate, voltage change ratio, four sample signals of power variation rate and conductance change rate highlight characteristic when failure occurs.
Other data characteristicses are not needed, and fault detection and classification can be carried out in the case where not influencing photovoltaic generating system work.
2, the present invention is based on the methods of wavelet multiresolution analysis to be extracted the feature vector of 4 dimensions, with SVM training failure modes
Model realizes the fault diagnosis and classification of photovoltaic power generation array.This detection scheme ambient adaptability is strong, there is stronger innovation
Property, trained model is able to achieve the fault diagnosis and classification of degree of precision.The present invention can using wavelet multi_resolution analysis technology
To extract the detail coefficients of signal, the feature that prominent failure occurs.The fault diagnosis of feature vector training with SVM based on extraction
Model, ambient adaptability is strong, realizes the accurately fault detection and classification of photovoltaic power generation array.Classification accuracy of the invention
Up to 95% or more.
Detailed description of the invention
Fig. 1 is the method flow schematic diagram of the embodiment of the present invention.
Fig. 2 is the photovoltaic generating system topological diagram of the embodiment of the present invention.
Fig. 3 is the photovoltaic generating system experiment porch figure of the embodiment of the present invention.
Fig. 4 is the failure failure modes result of the embodiment of the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
It is noted that described further below be all exemplary, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
As shown in Figure 1, present embodiments providing a kind of photovoltaic power generation array failure based on wavelet multiresolution analysis and SVM
The method of diagnosis and classification, Fig. 2 are the photovoltaic generating system topological diagram of the present embodiment, and system is by S × P solar components group
At, realization is attached with power grid by inverter and is generated electricity by way of merging two or more grid systems, the fault state occurred by simulation photovoltaic power generation array, packet
Include open circuit 1, three kinds of failures of short circuit 1 and short-circuit 2 failure.Under not equality of temperature photograph, real time fail is carried out for every kind of fault condition and is examined
It is disconnected, specifically includes the following steps:
Step S1: the not electric current of equality of temperature illumination and voltage sample signal under acquisition photovoltaic array different working condition；
Step S2: the electric current and voltage signal of acquisition are normalized, and obtain current changing rate, voltage change
Four rate, power variation rate and conductance change rate sample signals；
Step S3: the sliding window that setting window size is L carries out sliding to four sample signals that step S2 is obtained and takes window
Message number carries out the multiresolution analysis of wavelet transformation to four kinds of window signals of acquirement；
Step S4: 2 norms of the high-frequency signal that four window signal n-th layers are decomposed are calculated；
Step S5: 2 models of the high-frequency signal for taking the n-th layer of window signal of four sample signals when failure occurs to decompose
Number is used as characteristic value, the feature vector that dimension is 4；
Step S6: obtaining multiple feature value vectors by the multiple groups sample signal of different working condition, wherein by m feature to
Amount is used as training data, using n feature vector as test data；
Step S7: being arranged the parameter of support vector machines, the training fault diagnosis model from training data, with test number
According to the accuracy for the classification for detecting the model.
Preferably, the system parameter of the present embodiment is as shown in the table.
1 system detail parameters of table
The present invention extracts the photovoltaic array current changing rate of different working condition, voltage change with wavelet multiresolution analysis
4 Wei Tes of the SVM parameter from construction are arranged as feature in rate, the detail coefficients of four signals of power variation rate and conductance change rate
Training fault grader in vector is levied, realizes the fault diagnosis and classification of photovoltaic array.
In the present embodiment, in step S1, the different working condition of photovoltaic array includes: that normal, 1 component of single group string is short
Road (hereinafter referred to as short circuit 1), one component open circuit (hereinafter referred to as open circuit 1) of single group string and the 2 component short circuits of single group string are (hereinafter referred to as short
Road 2)；Wherein, the 1 component short circuit of single group string, one component open circuit of single group string and the 2 component short circuits of single group string are failure shape
State, the sample signal of malfunction contain photovoltaic array by normally to the failure process stable to failure again, capturing failure
The variation characteristic of array current and voltage when generation.Wherein, the acquisition time window t=10s of electric current and voltage sample signal, is adopted
Sample frequency is 200Hz.
The method of the present embodiment can be to the short circuit 1 under different illuminance, and open circuit 1 and short-circuit 2 failures are detected.It is different
The electric signal of photovoltaic array has similar variation characteristic in failure of the same race under environment, and the method proposed is sent out in string type photovoltaic
The stronger ambient adaptability of electric system.Particularly, the present embodiment simulation photovoltaic power generation system is normal, short circuit 1, open circuit 1 and short circuit 2 four
Kind working condition carries out data acquisition.In part of in August, 2018 point multiple periods, sample signal is carried out under different warm illumination
Random acquisition, every kind of working condition acquire 150 sample signals, obtain 150 feature vectors with wavelet multi_resolution analysis,
100 feature vectors are randomly selected as training data, remaining 50 feature vectors are as test data.Fig. 3 is this implementation
Photovoltaic generating system experiment porch figure in example.The current sample signal specifying information of acquisition is as shown in table 2.
2 sample signal of table acquires information
In the present embodiment, step S2 the following steps are included:
Step S21: the electric current and voltage signal of acquisition are normalized: by array current divided by short circuit current,
By array voltage divided by open-circuit voltage；The influence of different illuminance and different temperatures, normalizing can be overcome by normalized
The formula of change is as follows:
iPV(t)=IPV(t)/ISC(t)；
vPV(t)=VPV(t)/VOC(t)
In formula, IPV(t) the array current sample signal of acquisition, I are indicatedSC(t) array short circuit current signal, i are indicatedPV(t)
Array current sample signal after indicating normalization, VPV(t) the array voltage sample signal of acquisition, V are indicatedOC(t) array is indicated
Open circuit signaling, vPV(t) the array voltage sample signal after normalization is indicated；Electric current and voltage letter after normalized
The variation tendency for number reflecting array signal under different working condition, highlights variation characteristic；
Step S22: current changing rate, voltage change ratio, power variation rate and conductance change rate four are obtained using following formula
Sample signal:
Di (t)=[iPV(n+1)-iPV(n)]/t；
Dv (t)=[vPV(n+1)-vPV(n)]/t；
Dp (t)=[iPV(n+1)v(n+1)-iPV(n)v(n)]/t；
Dg (t)=[iPV(n+1)/vPV(n+1)-iPV(n)/iPV(n)]/t；
In formula, di (t), dv (t), dp (t), dg (t) respectively describe the curent change of photovoltaic array different working condition
Rate, voltage change ratio, the signal of power variation rate and conductance change rate.
In the present embodiment, in step S3, the window size L is 20, is believed using four kind windows of the DB small echo to acquirement
Number carry out 6 floor Multiresolution Decomposition wavelet transformation；The multiresolution analysis is to be divided signal with high pass and low-pass filter
Solution at multiple ranks high frequency detail and approximation.
Preferably, in the present embodiment, wavelet transformation is a kind of widely used digital signal processing method, it is suitable for dividing
Nonperiodic signal is analysed, time and frequency analysis can be carried out to signal simultaneously.Signal x (t) is discrete to turn to x (n), the signal
DWT can be solved by following formula:
In formula, ψ (t) is morther wavelet, and n is the points of signal x (n), a=a0 jIt is scale factor, b=kb0a0 jShift factor.
a0> 0 and be constant, b0≠ 0, j, k ∈ Z.
Multiresolution analysis is with the high frequency detail and approximation of high pass and low-pass filter by signal decomposition at multiple ranks
Value.Time-domain signal f (t) can be indicated by scaling function Φ (t) and wavelet function Ψ (t), be shown below:
In formula, dj(k) be decomposition at different levels detail coefficients, express high-frequency information, and aN(k) afterbody N grades of approximate system
Number expresses low-frequency information.
In the present embodiment, in step S4,2 norms for calculating the high-frequency signal of four window signal n-th layers decomposition are used
Following formula:
In formula, djnIt indicates, N is indicated, djIndicate the high frequency coefficient of jth layer, fiIndicate the feature of i-th of signal extraction.This
Invention extracts the high frequency coefficient of the 6th layer of 4 sample signals decomposition as characteristic value, therefore i=1, and 2,3,4, j=6.
In the present embodiment, in step S5, the feature vector that the dimension is 4 is f=[f1,f2,f3,f4]。
In the present embodiment, in step S7, it is described setting support vector machines parameter specifically: by the punishment of SVM because
Sub- C is set as 1000, sets 5 for the sum of the distance γ of two different classes of supporting vectors to hyperplane.Specifically:
The basic thought of support vector machines is that an optimal separating hyper plane is found by the training sample set of linear separability
To realize the division of different classes of sample data.For nonlinear classification problem, sample can be mapped from luv space
To the high-dimensional feature space of a linear separability, to find a suitable optimal separating hyper plane.
Given training sample data collection, D={ xi,yi, i=1,2,3 ..., m, yi∈ { -1,1 }, wherein xiFor sample number
According to m is training sample sum, and d is the dimension of sample space, yiFor the corresponding label of sample.They can be optimal super by one
Plane is separated, which can be expressed as wTX+b=0, wherein w ∈ Rd, it is normal vector, determines the direction of hyperplane；b
∈ R, for be displaced item, determine hyperplane between origin at a distance from, the threshold value of classification.
Assuming that hyperplane can correctly classify training sample, then for { xi, yi } ∈ D, if yi=+1, then there is wTX+b >
0；If yi, then there is w in=- 1TX+b < 0.It enables:
Then several training sample points nearest apart from hyperplane set up the equal sign of above formula, they be referred to as support to
It measures (Support vectors, SVs), then the sum of the distance of two different classes of supporting vectors to hyperplane is
The distance is referred to as class interval.If it is desired that γ is maximum, then require | | w | |2Minimum, while requiring classifying face to all samples
This correct classification, needs to meet
yi(wTxi+ b) >=1, i=1,2,3 ..., l
Therefore, optimal separating hyper plane problem is sought, a quadratic programming problem can be converted into, optimization aim can be written as
s.t.yi(wTxi+ b) >=1, i=1,2,3 ..., l
For the training sample in sample space linear separability, can be divided by optimal separating hyper plane.Then,
In realistic task, the case where being usually present linearly inseparable, there are part sample datas to be unsatisfactory in training sample at this time
, there is the error centainly classified in formula (1).Therefore, by introducing slack variable ξi(ξi>=0) this is solved the problems, such as.Therefore, public
Formula (1) can be written as
yi(wTxi+b)≥1-ξi, i=1,2,3 ..., l
While maximizing interval, it is desirable that the sample for being unsatisfactory for constraint will lack as far as possible.Therefore, optimization object function can
To be rewritten as
Wherein,Referred to as penalty term, C are penalty factor.Therefore, optimal classification when available linearly inseparable
Face, referred to as broad category hyperplane, can be expressed as following optimization problem
s.t.yi(wTxi+b)≥1-ξi
ξi>=0, i=1,2,3 ..., l
Solving the optimization problem by method of Lagrange multipliers, can obtain its dual problem such as following formula:
0≤αi≤ C, i, j=1,2,3 ..., l
According to Karush-Kuhn-Tucker (KKT) condition, available αi(yi(wTxi+ b) -1)=0.If αi> 0,
Then corresponding sample point is located on largest interval boundary, is exactly supporting vector.Can then it pass throughIt solves
W, and according to yi(wTxi+ b) -1=0 solves b, wherein xiFor supporting vector, n is the quantity of supporting vector.Determine w and b it
Afterwards, it is as follows to obtain categorised decision function
Secondly, being directed to Nonlinear Classification problem, sample is mapped to the spy of more higher-dimension using kernel function by SVM from luv space
Space is levied, so that sample linear separability in this feature space.φ (x) is enabled to indicate the feature vector after mapping x, then in height
Optimization aim corresponding to hyperplane is represented by dimension space
s.t.yi(wTφ(xi)+b)≥1-ξi
ξi>=0, i=1,2,3 ..., l
Corresponding dual problem is
0≤αi≤ C, i, j=1,2,3 ..., l
To avoid calculating sample xiAnd xjInner product operation in higher dimensional space constructs kernel function K (), xiAnd xj?
The inner product of feature space is converted to the result calculated in original sample space by the function.K () is expressed as follows
K(xi,xj)=φ (xi)Tφ(xj)
Then previous formula is become
0≤αi≤ C, i, j=1,2,3 ..., l
Categorised decision function becomes at this time
Common kernel function mainly have Polynomial kernel function, radial base (Radial based kernel, RBF) kernel function,
Tanh (Sigmoid) kernel function etc..SVM is applied using RBF kernel function in the present invention.
RBF kernel function is expressed as
K(xi, x) and=exp (- γ | | xi-x||2)
The classifying quality of support vector machines will be largely dependent on the selection of C and γ, set C in the present invention
1000, γ are set as 10.With parameter SVM is set, fault grader is trained from the training data of construction, realizes photovoltaic
The fault diagnosis and classification of power generation array.The precision of the classification is detected with 50 test datas, Fig. 4 divides for the failure suggested plans
Class result.
It is corresponding, short circuit 1 label be 1, open a way 1 label be 2, short circuit 2 label be 3, short circuit 2 label be 4.?
In classification results figure, if prediction label and physical tags are overlapped, the prediction result of the data is accurate.As shown in figure 4,
There are the prediction label of 1 data and physical tags inconsistent in 50 1 test datas of short circuit, diagnostic accuracy 98%, open circuit 1 is surveyed
Examination data in have 4 data prediction malfunction, precision of prediction 92%, open a way 2 test datas in have predicting for 2 data
Mistake, precision of prediction 96%, and the predictablity rate of proper testing data is 100%.The program realizes whole 96.5% essence
The fault diagnosis and classification of degree.To sum up, the fault diagnosis in the present embodiment and classifying quality are as shown in table 3.
3 fault detection of table and classification results
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (7)
1. a kind of method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM, feature exist
In: the following steps are included:
Step S1: the not electric current of equality of temperature illumination and voltage sample signal under acquisition photovoltaic array different working condition；
Step S2: being normalized the electric current and voltage signal of acquisition, and obtain current changing rate, voltage change ratio,
Four sample signals of power variation rate and conductance change rate；
Step S3: the sliding window that setting window size is L carries out sliding to four sample signals that step S2 is obtained and window is taken to believe
Number, four kinds of window signals of acquirement are carried out with the multiresolution analysis of wavelet transformation；
Step S4: 2 norms of the high-frequency signal that four window signal n-th layers are decomposed are calculated；
Step S5: 2 norms of the high-frequency signal for taking the n-th layer of window signal of four sample signals when failure occurs to decompose are made
It is characterized value, the feature vector that dimension is 4；
Step S6: obtaining multiple feature value vectors by the multiple groups sample signal of different working condition, wherein m feature vector is made
For training data, using n feature vector as test data；
Step S7: being arranged the parameter of support vector machines, and training fault diagnosis model, is examined with test data from training data
Survey the accuracy of the classification of the model.
2. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 1 and
The method of classification, it is characterised in that: in step S1, the different working condition of photovoltaic array includes: normal, 1 component of single group string
Short circuit, one component open circuit of single group string and the 2 component short circuits of single group string；Wherein, the 1 component short circuit of single group string, single group string one
A component open circuit and the 2 component short circuits of single group string are malfunction, the sample signal of malfunction contain photovoltaic array by
It normally arrives the stable process of failure again to failure, captures the variation characteristic of array current and voltage when failure occurs.
3. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 1 and
The method of classification, it is characterised in that: step S2 the following steps are included:
Step S21: the electric current and voltage signal of acquisition are normalized: by array current divided by short circuit current, by battle array
Column voltage is divided by open-circuit voltage；The influence of different illuminance and different temperatures can be overcome by normalized, it is normalized
Formula is as follows:
iPV(t)=IPV(t)/ISC(t)；
vPV(t)=VPV(t)/VOC(t)
In formula, IPV(t) the array current sample signal of acquisition, I are indicatedSC(t) array short circuit current signal, i are indicatedPV(t) it indicates
Array current sample signal after normalization, VPV(t) the array voltage sample signal of acquisition, V are indicatedOC(t) array open circuit is indicated
Signal, vPV(t) the array voltage sample signal after normalization is indicated；
Step S22: four current changing rate, voltage change ratio, power variation rate and conductance change rate samples are obtained using following formula
Signal:
Di (t)=[iPV(n+1)-iPV(n)]/t；
Dv (t)=[vPV(n+1)-vPV(n)]/t；
Dp (t)=[iPV(n+1)v(n+1)-iPV(n)v(n)]/t；
Dg (t)=[iPV(n+1)/vPV(n+1)-iPV(n)/iPV(n)]/t；
In formula, di (t), dv (t), dp (t), dg (t) respectively describe the current changing rate of photovoltaic array different working condition, electricity
Buckling rate, the signal of power variation rate and conductance change rate.
4. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 1 and
The method of classification, it is characterised in that: in step S3, the window size L is 20, is believed using four kind windows of the DB small echo to acquirement
Number carry out 6 floor Multiresolution Decomposition wavelet transformation；The multiresolution analysis is to be divided signal with high pass and low-pass filter
Solution at multiple ranks high frequency detail and approximation.
5. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 1 and
The method of classification, it is characterised in that: in step S4,2 norms for calculating the high-frequency signal of four window signal n-th layers decomposition are used
Following formula:
In formula, djnIt indicates, N is indicated, djIndicate the high frequency coefficient of jth layer, fiIndicate the feature of i-th of signal extraction.
6. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 5 and
The method of classification, it is characterised in that: in step S5, the feature vector that the dimension is 4 is f=[f1,f2,f3,f4]。
7. a kind of photovoltaic power generation array fault diagnosis based on wavelet multiresolution analysis and SVM according to claim 1 and
The method of classification, it is characterised in that: in step S7, the parameter of the setting support vector machines specifically: by the punishment of SVM
Factor C is set as 1000, sets 5 for the sum of the distance γ of two different classes of supporting vectors to hyperplane.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811561729.0A CN109617526A (en) | 2018-12-20 | 2018-12-20 | A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811561729.0A CN109617526A (en) | 2018-12-20 | 2018-12-20 | A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109617526A true CN109617526A (en) | 2019-04-12 |
Family
ID=66009694
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811561729.0A Pending CN109617526A (en) | 2018-12-20 | 2018-12-20 | A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109617526A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111046790A (en) * | 2019-12-11 | 2020-04-21 | 江苏大学镇江流体工程装备技术研究院 | Pump bearing fault diagnosis method |
| CN113746132A (en) * | 2021-08-11 | 2021-12-03 | 国网江苏省电力有限公司 | Photovoltaic power station based on cloud edge cooperation and control method thereof |
| CN113916536A (en) * | 2021-09-03 | 2022-01-11 | 山东科技大学 | Sinusoidal roller screen coupling fault diagnosis method and system |
| CN114993438A (en) * | 2022-05-30 | 2022-09-02 | 济南建和机电设备有限公司 | Method for monitoring granary inventory by using distributed pressure sensor |
| CN115758284A (en) * | 2022-11-14 | 2023-03-07 | 南京工业大学 | Photovoltaic hot spot fault detection method and system based on fusion of nuclear entropy and information gain |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104638668A (en) * | 2014-06-12 | 2015-05-20 | 华中科技大学 | Photovoltaic power generation grid connection control method and photovoltaic power generation grid connection control system |
| CN104753461A (en) * | 2015-04-10 | 2015-07-01 | 福州大学 | Method for diagnosing and classifying faults of photovoltaic power generation arrays on basis of particle swarm optimization support vector machines |
| CN105375461A (en) * | 2015-07-10 | 2016-03-02 | 华北电力大学（保定） | Active power distribution network power supply capacity real-time assessment method based on prediction technology |
| CN108875796A (en) * | 2018-05-28 | 2018-11-23 | 福州大学 | Diagnosing failure of photovoltaic array method based on linear discriminant analysis and support vector machines |
-
2018
- 2018-12-20 CN CN201811561729.0A patent/CN109617526A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104638668A (en) * | 2014-06-12 | 2015-05-20 | 华中科技大学 | Photovoltaic power generation grid connection control method and photovoltaic power generation grid connection control system |
| CN104753461A (en) * | 2015-04-10 | 2015-07-01 | 福州大学 | Method for diagnosing and classifying faults of photovoltaic power generation arrays on basis of particle swarm optimization support vector machines |
| CN105375461A (en) * | 2015-07-10 | 2016-03-02 | 华北电力大学（保定） | Active power distribution network power supply capacity real-time assessment method based on prediction technology |
| CN108875796A (en) * | 2018-05-28 | 2018-11-23 | 福州大学 | Diagnosing failure of photovoltaic array method based on linear discriminant analysis and support vector machines |
Non-Patent Citations (2)
| Title |
|---|
| KUN XIA等: ""Wavelet Packet and Support Vector Machine Analysis of Series DC Arc Fault Detection in Photovoltaic System"", 《IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING》 * |
| ZHEHAN YI等: ""Line-to-Line Fault Detection for Photovoltaic Arrays Based on Multiresolution Signal Decomposition and Two-Stage Support Vector Machine"", 《IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111046790A (en) * | 2019-12-11 | 2020-04-21 | 江苏大学镇江流体工程装备技术研究院 | Pump bearing fault diagnosis method |
| CN113746132A (en) * | 2021-08-11 | 2021-12-03 | 国网江苏省电力有限公司 | Photovoltaic power station based on cloud edge cooperation and control method thereof |
| CN113746132B (en) * | 2021-08-11 | 2024-01-30 | 国网江苏省电力有限公司 | Photovoltaic power station based on cloud edge cooperation and control method thereof |
| CN113916536A (en) * | 2021-09-03 | 2022-01-11 | 山东科技大学 | Sinusoidal roller screen coupling fault diagnosis method and system |
| CN114993438A (en) * | 2022-05-30 | 2022-09-02 | 济南建和机电设备有限公司 | Method for monitoring granary inventory by using distributed pressure sensor |
| CN115758284A (en) * | 2022-11-14 | 2023-03-07 | 南京工业大学 | Photovoltaic hot spot fault detection method and system based on fusion of nuclear entropy and information gain |
| CN115758284B (en) * | 2022-11-14 | 2023-05-16 | 南京工业大学 | Photovoltaic hot spot fault detection method and system based on fusion kernel entropy and information gain |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108647716B (en) | Photovoltaic array fault diagnosis method based on composite information | |
| CN106777984B (en) | A method of photovoltaic array Working state analysis and fault diagnosis are realized based on density clustering algorithm | |
| CN109617526A (en) | A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM | |
| CN104753461B (en) | Method for diagnosing and classifying faults of photovoltaic power generation arrays on basis of particle swarm optimization support vector machines | |
| CN109738776A (en) | Fan converter open-circuit fault recognition methods based on LSTM | |
| CN109460618A (en) | A kind of rolling bearing remaining life on-line prediction method and system | |
| CN109660206B (en) | Wasserstein GAN-based photovoltaic array fault diagnosis method | |
| CN109672406B (en) | Photovoltaic power generation array fault diagnosis and classification method based on sparse representation and SVM | |
| CN110336534A (en) | A kind of method for diagnosing faults based on photovoltaic array electric parameter time series feature extraction | |
| Zaki et al. | Deep‐learning–based method for faults classification of PV system | |
| CN109635928A (en) | A kind of voltage sag reason recognition methods based on deep learning Model Fusion | |
| CN109871976A (en) | A kind of prediction technique of power quality containing distributed power distribution network based on cluster and neural network | |
| CN108983749A (en) | Diagnosing failure of photovoltaic array method based on K-SVD training sparse dictionary | |
| CN109766952A (en) | Photovoltaic array fault detection method based on Partial Least Squares and extreme learning machine | |
| CN107807860B (en) | Power failure analysis method and system based on matrix decomposition | |
| Iturrino-García et al. | An innovative single shot power quality disturbance detector algorithm | |
| Zu et al. | A simple gated recurrent network for detection of power quality disturbances | |
| He et al. | Compound fault diagnosis for photovoltaic arrays based on multi-label learning considering multiple faults coupling | |
| Serrano-Luján et al. | Case of study: Photovoltaic faults recognition method based on data mining techniques | |
| Hussain et al. | Deployment of AI-based RBF network for photovoltaics fault detection procedure | |
| CN113222067B (en) | Island intelligent detection method based on SVM-Adaboost algorithm | |
| Zgraggen et al. | Physics informed deep learning for tracker fault detection in photovoltaic power plants | |
| Lu et al. | Efficient fault diagnosis approach for solar photovoltaic array using a convolutional neural network in combination of generative adversarial network under small dataset | |
| Gnetchejo et al. | Faults diagnosis in a photovoltaic system based on multivariate statistical analysis | |
| CN113642255A (en) | Photovoltaic power generation power prediction method based on multi-scale convolution cyclic neural network |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190412 |
|
| RJ01 | Rejection of invention patent application after publication |