CN109388845A - Based on backward learning and the complicated photovoltaic array parameter extracting method evolved of enhancing - Google Patents
Based on backward learning and the complicated photovoltaic array parameter extracting method evolved of enhancing Download PDFInfo
- Publication number
- CN109388845A CN109388845A CN201810946865.5A CN201810946865A CN109388845A CN 109388845 A CN109388845 A CN 109388845A CN 201810946865 A CN201810946865 A CN 201810946865A CN 109388845 A CN109388845 A CN 109388845A
- Authority
- CN
- China
- Prior art keywords
- point
- complex shape
- follows
- model
- complicated
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Photovoltaic Devices (AREA)
Abstract
The present invention relates to based on backward learning and the complicated photovoltaic array parameter extracting method evolved of enhancing, comprising: obtains the actual I-V characteristic curve of photovoltaic panel, and selects corresponding photovoltage model.Determine the objective function of this optimization problem.The position of initial point is optimized by backward learning algorithm (OBL) algorithm.Model parameter is extracted according to different circuit models using enhanced complicated evolution algorithm (ESCE).The model parameter of photovoltaic panel under the conditions of different actual measurements is extracted by the algorithm.A kind of photovoltaic array parameter extracting method based on backward learning strategy with enhanced complicated evolution algorithm proposed by the present invention, speed is fast, and convergence is strong, and stability is good.
Description
Technical field
It is especially a kind of multiple based on backward learning and enhancing the present invention relates to photovoltaic power generation array model parameter extraction technology
The photovoltaic array parameter extracting method of miscellaneous evolution.
Background technique
With global energy exhaustion, people are increasing for the demand of green novel energy source.Photovoltaic power generation turns solar energy
Change electric energy, green cleanliness without any pollution into, it is considered to be can most replace traditional one of scheme of fossil energy.As photovoltaic
The core of power generation, large-scale photovoltaic array mostly by photovoltaic module string and form.Therefore, to photovoltaic cell actual measurement condition I-
V characteristic carries out modeling and parameter extraction, for the Performance Evaluation of photovoltaic generating system entirety, Optimized System Design and real time fail
Detection has and its important meaning.
In order to assess photovoltaic generating system, it is necessary first to be modeled to photovoltaic array, current main photovoltaic
Model mainly has the five-parameter model of single diode and seven parameter models of double diode.Five-parameter model has centainly accurate
Rate and calculate simple and effective, structure is complicated for seven parameter models, and computational efficiency is low but more accuracy, estimated parameter can be preferably
Approach measured curve.
Parameter extracting method primarily now can be divided into analytic method, intelligent optimization algorithm and mixed method.Analytic method can
Directly quickly to obtain model parameter, but accuracy and robustness are poor and need complicated mathematical derivation.Therefore, it utilizes
The problem of intelligent optimization algorithm search such as photovoltage model parameter extraction, is widely favored.Currently, a large amount of intelligent optimization is calculated
Method has been applied to the parameter extraction (such as ABC, GOFPANM, Rcr-IJADE, STLBO, GOTLBO etc.) of photovoltaic array model, this
The features such as that there is convergence rates is slow for a little algorithms, computationally intensive, poor robustness, stability is low.Chaotic complicated (SCE) algorithm of evolving
It was most suggested early in 1993, mainly complex shape is optimized using deterministic competitive complicated evolution, is then passed through
New complex shape is extracted in mixing.The algorithm itself has the features such as robustness is good, and computational efficiency is high.But when in face of multiple target
The selection of nonlinear problem, initial point may will affect the convergence.In order to further enhance the convergence of SCE,
The algorithm that VCMarian proposed a kind of differential evolution (DE) in 2011 and SCE is mixed.The algorithm is gone down the hill adaptability
Simplex search algorithm (NMS) and differential evolution combine.This method can effectively improve the overall performance and calculating of SCE
Efficiency.
Summary of the invention
The purpose of the present invention is to provide a kind of based on backward learning and the complicated photovoltaic array parameter extraction evolved of enhancing
Method, to overcome defect existing in the prior art.
To achieve the above object, the technical scheme is that based on backward learning and the complicated photovoltaic battle array evolved of enhancing
Column parameter extracting method is realized in accordance with the following steps:
Step S1: the actual I-V characteristic curve of photovoltaic panel is obtained, and selects corresponding photovoltage model;
Step S2: the objective function of optimization problem is determined;
Step S3: by using backward learning algorithm, the position of initial point is optimized;
Step S4: by using enhanced complicated evolution algorithm, model parameter is extracted according to different circuit models;
Step S5: the model parameter of photovoltaic panel under the conditions of different actual measurements is extracted.
In an embodiment of the present invention, in the step S1, further include following steps:
Step S11: output Current Voltage number of photovoltaic panel under the conditions of different illumination, irradiation level is acquired by collection plate
According to, and it is saved as to the number table of N*1 respectively;Wherein, N is the number of collecting sample point, and the sample of the electric current and voltage acquired
This quantity is identical;
Step S12: select single diode model and double diode model as photovoltage model;
Single diode model are as follows:
Wherein, IpvIt is photogenerated current, IoIt is the saturation current of diode, a is the ideal factor of diode, RsBe it is equivalent simultaneously
Join resistance, RshIt is equivalent series resistance, K is Boltzmann constant (1.380653 × 10-23J/K), T is environment temperature, and q is electricity
The absolute value (1.60217646 × 10 of charge of the electron amount-19C);
Double diode model are as follows:
Wherein, IpvIt is photogenerated current, Io1,Io2The saturation current of two different diodes respectively, a1,a2Respectively two two
The ideal factor of pole pipe, RsIt is equivalent parallel resistance, RshEquivalent series resistance, K be Boltzmann constant (1.380653 ×
10-23J/K), T is environment temperature, and q is the absolute value (1.60217646 × 10 of electronic charge-19C)。
In an embodiment of the present invention, in the step S2, the objective function namely fitness function, according to such as
Lower method obtains:
Remember root-mean-square error are as follows:
Wherein, N is the total sample number on I-V characteristic curve, and f (V, I, X) is between measured current and model estimation electric current
Absolute error;
For the f (V, I, X) of single diode are as follows:
For the f (V, I, X) of double diode are as follows:
In an embodiment of the present invention, in the step S3, further include following steps:
Step S31: corresponding control parameter is initialized, comprising: solved in the quantity npg of complex shape, each complex shape
Count the lower coboundary BL and BU of ngs, the sample number npt of total candidate solution, candidate solution;
Step S32: random initializtion, process are carried out in the lower upper bound BL, BU to npt candidate samples point are as follows:
xi=BL+rand* (BU-BL), i=1,2 ..., npt
And calculate the fitness function value of its each position;
Step S33: opposition solution x is obtained in the way of backward learning to candidate solutiono, process are as follows:
xo=BU+BL-xi
The fitness function value of opposition solution is calculated, and by xoAnd xiMixing postscript is new candidate solution s, according to its fitness
The size of function carries out ascending order arrangement.
In an embodiment of the present invention, further include following steps in the step S4:
Step S41: npt optimal candidate solution is divided into npg complex shape;
Step S42: each complex shape is evolved by enhanced complicated evolution algorithm, obtains the position of new candidate solution
It sets;
Step S43: the optimization point in the complex shape after evolution is replaced into the adaptation corresponding with its of original complex shape candidate point
Spend functional value;
Step S44: when objective function calculation times reach maximum, end algorithm returns to optimal solution.
In an embodiment of the present invention, further include following steps in the step S52:
Step S521: initialization control parameter;Note simplex number of vertex is q, and control coefrficient is respectively α, beta, gamma, and primary and secondary is multiple
Miscellaneous shape evolution number is (m, n);
Step S522: the vertex that q point makees ngs as simplex is selected from npt point of complex shape, and calculates the q
The central point ce of point, calculation method are as follows:Wherein, ujIt is q-1 simplex vertex
Position;
Step S523: corresponding reflection point u is obtained according to central pointr=ce- α (ce-uq), after boundary Control,
Calculate the fitness function value of reflection point;If the fitness function value of reflection point is less than the optimum point in the complex shape, according to
As under type calculates the numerical value of its extension point: ue=ce+ γ (ce-uq), and optimal position in reflection point and extension point is taken
Instead of position worst in this vertex;
Step S524: when reflection point position is poorer than the Best Point in complex shape, then according in the complex shape optimum point,
Central point, most not good enough and reflection point, generate compression point as follows:
Wherein, ugAnd uqIt is globe optimum and most not good enough position, u respectivelyrIt is the position of reflection point, β is contraction factor, Fq-1It is time
Not good enough fitness function value, FrIt is the fitness function value of reflection point, FqIt is most not good enough fitness function value;
Step S525: if compression point cannot still update most almost, according to globe optimum and local best points, according to such as
Under type optimizes worst point: uz=uq+β(ug-uq)+γ×r·(ub-uq), r ∈ (- 1,1), wherein uzIt is to update point
Position, ugIt is globe optimum, ubIt is local best points, r is the random number between -1 and 1, and γ is spreading factor;
Step S526: it repeats step S522 to step S525n times, step S521 to step S525m times.
Compared to the prior art, the invention has the following advantages: it is proposed by the present invention based on backward learning and enhancing
The photovoltaic array parameter extracting method of complexity evolution (ESCE-OBL), using the strategy of backward learning, to the unknown parameter of model
Secondary initialization is carried out, then chooses complex shape, then optimize to the complex shape chosen by improved complicated evolution algorithm,
Mix simultaneously by the complex shape that the complex shape after evolution is instead preceding, and by multiple complex shapes, after being ranked up, then separate new
Complex shape optimizes.During optimization, by solving the minimum value of objective function, final approximation by target function value most
The model parameter value of hour, it is believed that optimization objective function value when being convergence.Compared to traditional SCE algorithm, the present invention is in light
That lies prostrate parameter extraction has faster convergence rate and robustness using upper.Identical precision can be reached with SCE, but it is received
It holds back speed and is apparently higher than SCE, and it with better stability and robustness, stability is with RMSE in different data sample
Under variance make reference, up to more than an order of magnitude.
Detailed description of the invention
Fig. 1 is the photovoltaic array parameter extraction side in the present invention based on backward learning strategy and enhanced complicated evolution algorithm
The flow diagram of method
Fig. 2 is the idiographic flow block diagram of ESCE-OBL algorithm in one embodiment of the invention
Fig. 3 is the flow diagram of the CCE evolution strategy enhanced in one embodiment of the invention
Fig. 4 is ESCE-OBL and existing algorithm the parameter extraction result under single diode model in one embodiment of the invention
Comparison schematic diagram.
Fig. 5 is ESCE-OBL and existing algorithm the parameter extraction result under double diode model in one embodiment of the invention
Comparison schematic diagram.
Fig. 6 be in one embodiment of the invention ESCE-OBL and SCE algorithm under single diode model to photovoltaic cell and light
The further comparison schematic diagram of Fu Mo group parameter extraction result.
Fig. 7 be in one embodiment of the invention ESCE-OBL and SCE algorithm under double diode model to photovoltaic cell and light
The further comparison schematic diagram of Fu Mo group parameter extraction result.
Fig. 8 is convergence curve signal of the ESCE-OBL and SCE algorithm when extracting for two kinds in one embodiment of the invention
Figure.
Specific embodiment
With reference to the accompanying drawing, technical solution of the present invention is specifically described.
The present invention provides a kind of photovoltaic array parameter extraction based on backward learning strategy with enhanced complicated evolution algorithm
Method is realized in accordance with the following steps as shown in Figure 1 and Figure 2:
Step S1: the actual I-V characteristic curve of photovoltaic panel is obtained, and selects corresponding photovoltage model;
Step S2: the objective function of optimization problem is determined;
Step S3: by using backward learning algorithm, the position of initial point is optimized;
Step S4: by using enhanced complicated evolution algorithm, model parameter is extracted according to different circuit models;
Step S5: the model parameter of photovoltaic panel under the conditions of different actual measurements is extracted.
Further, in the present embodiment, in step sl, further include following steps:
Step S11: output Current Voltage number of photovoltaic panel under the conditions of different illumination, irradiation level is acquired by collection plate
According to, and it is saved as to the number table of N*1 respectively;Wherein, N is the number of collecting sample point, and the sample of the electric current and voltage acquired
This quantity is identical;
Step S12: select single diode model and double diode model as photovoltage model;
Single diode model are as follows:
Wherein, IpvIt is photogenerated current, IoIt is the saturation current of diode, a is the ideal factor of diode, RsBe it is equivalent simultaneously
Join resistance, RshIt is equivalent series resistance, K is Boltzmann constant (1.380653 × 10-23J/K), T is environment temperature, and q is electricity
The absolute value (1.60217646 × 10 of charge of the electron amount-19C);
Double diode model are as follows:
Wherein, IpvIt is photogenerated current, Io1,Io2The saturation current of two different diodes respectively, a1,a2Respectively two two
The ideal factor of pole pipe, RsIt is equivalent parallel resistance, RshEquivalent series resistance, K be Boltzmann constant (1.380653 ×
10-23J/K), T is environment temperature, and q is the absolute value (1.60217646 × 10 of electronic charge-19C)。
Further, in the present embodiment, it selects single diode model to carry out parameter extraction to photovoltaic array, passes through double two
Pole pipe model carries out parameter extraction to photovoltaic cell and further verifies its advantage.
Further, in the present embodiment, in step s 2, objective function namely fitness function, as follows
It obtains:
Remember root-mean-square error are as follows:
Wherein, N is the total sample number on I-V characteristic curve, and f (V, I, X) is between measured current and model estimation electric current
Absolute error;
For the f (V, I, X) of single diode are as follows:
For the f (V, I, X) of double diode are as follows:
Further, in the present embodiment, in step s3, further include following steps:
Step S31: corresponding control parameter is initialized, comprising: solved in the quantity npg of complex shape, each complex shape
Count the lower coboundary BL and BU of ngs, the sample number npt of total candidate solution, candidate solution;
Step S32: random initializtion, process are carried out in the lower upper bound BL, BU to npt candidate samples point are as follows:
xi=BL+rand* (BU-BL), i=1,2 ..., npt
And calculate the fitness function value of its each position;
Step S33: opposition solution x is obtained in the way of backward learning to candidate solutiono, process are as follows:
xo=BU+BL-xi
The fitness function value of opposition solution is calculated, and by xoAnd xiMixing postscript is new candidate solution s, according to its fitness
The size of function carries out ascending order arrangement.
Further, in the present embodiment, in step s 4, further include following steps:
Step S41: npt optimal candidate solution is divided into npg complex shape;
Step S42: each complex shape is evolved by enhanced complicated evolution algorithm, obtains the position of new candidate solution
It sets;
Step S43: the optimization point in the complex shape after evolution is replaced into the adaptation corresponding with its of original complex shape candidate point
Spend functional value;
Step S44: when objective function calculation times reach maximum, end algorithm returns to optimal solution.
Further, in the present embodiment, as shown in figure 3, further including following steps in step S52:
Step S521: initialization control parameter;Note simplex number of vertex is q, and control coefrficient is respectively α, beta, gamma, and primary and secondary is multiple
Miscellaneous shape evolution number is (m, n);Preferably, q=6 (single diode model), q=8 (double diode model), α=1, β=0.5,
γ=2, m=1, n=1;
Step S522: the vertex that q point makees ngs as simplex is selected from npt point of complex shape, and calculates the q
The central point ce of point, calculation method are as follows:Wherein, ujIt is q-1 simplex vertex
Position;
Step S523: corresponding reflection point u is obtained according to central pointr=ce- α (ce-uq), after boundary Control,
Calculate the fitness function value of reflection point;If the fitness function value of reflection point is less than the optimum point in the complex shape, according to
As under type calculates the numerical value of its extension point: ue=ce+ γ (ce-uq), and optimal position in reflection point and extension point is taken
Instead of position worst in this vertex;
Step S524: when reflection point position is poorer than the Best Point in complex shape, then according in the complex shape optimum point,
Central point, most not good enough and reflection point, generate compression point as follows:
Wherein, ugAnd uqIt is globe optimum and most not good enough position, u respectivelyrIt is the position of reflection point, β is contraction factor, Fq-1It is time
Not good enough fitness function value, FrIt is the fitness function value of reflection point, FqIt is most not good enough fitness function value;;
Step S525: if compression point cannot still update most almost, according to globe optimum and local best points, according to such as
Under type optimizes worst point: uz=uq+β(ug-uq)+γ×r·(ub-uq), r ∈ (- 1,1), wherein uzIt is to update point
Position, ugIt is globe optimum, ubIt is local best points, r is the random number between -1 and 1, and γ is spreading factor;
Step S526: it repeats step S522 to step S525n times, step S521 to step S525m times.
In order to allow those skilled in the art to further appreciate that technical solution proposed by the present invention, below with reference to specific example into
Row explanation.
Further, as shown in figure 4, for algorithm proposed by the invention and the ABC, STLBO, the GOTLBO that have proposed,
The comparison of GOFPANM, Rcr-IJADE under single diode model, wherein what RMSE was represented is the root-mean-square error of objective function,
Its numerical value is smaller, illustrates that the error between the sample data between measured data and fitting data is smaller, i.e. parameter extraction process
Accuracy is higher, and MNFES refers to the maximum times of calculating target function, and numerical value is smaller, illustrates that its convergence is better.From figure
It is obvious that precision as the ESCE-OBL proposed and other algorithms have in 4, but the ESCE-OBL proposed
In MNFES other algorithms are substantially better than, illustrate that it, will be below specifically with the comparison of SCE with better convergence
As a result it is provided in.
Further, as shown in figure 5, for algorithm proposed by the invention and the ABC, STLBO, the GOTLBO that have proposed,
The comparison of GOFPANM, Rcr-IJADE under double diode model, it is obvious that the ESCE-OBL proposed from Fig. 5
Precision as having with other algorithms, but the ESCE-OBL proposed will be substantially better than other algorithms in MNFES, illustrate it
With better convergence, specifically with the comparison of SCE, will be provided in the result below.
Further, as shown in fig. 6, for SCE and ESCE-OBL to photovoltaic cell parameter extraction under single diode model
Statistical method comparison, can with ESCE-OBL in the result of variance and maximum value will be less than SCE, illustrate that ESCE-OBL exists
Robustness and stability are had more in the case of 5000 iteration of single diode model.
Further, as shown in fig. 7, for SCE and ESCE-OBL to photovoltaic cell parameter extraction under single bipolar tube model
Statistical method comparison, can with ESCE-OBL in the result of variance and maximum value will be less than SCE, illustrate that ESCE-OBL exists
Robustness and stability are had more in the case of 5000 iteration of single diode model.
Further, as shown in figure 8, for SCE and ESCE-OBL to photovoltaic cell parameter extraction under single bipolar tube model
Convergence curve, can be to be significantly faster than that SCE on ESCE-OBL first speed, the speed of decline will be faster than SCE, explanation
ESCE-OBL convergence traditional SCE algorithm is effectively promoted.
The above are preferred embodiments of the present invention, all any changes made according to the technical solution of the present invention, and generated function is made
When with range without departing from technical solution of the present invention, all belong to the scope of protection of the present invention.
Claims (6)
1. based on backward learning and the complicated photovoltaic array parameter extracting method evolved of enhancing, which is characterized in that according to following step
It is rapid to realize:
Step S1: the actual I-V characteristic curve of photovoltaic panel is obtained, and selects corresponding photovoltage model;
Step S2: the objective function of optimization problem is determined;
Step S3: by using backward learning algorithm, the position of initial point is optimized;
Step S4: by using enhanced complicated evolution algorithm, model parameter is extracted according to different circuit models;
Step S5: the model parameter of photovoltaic panel under the conditions of different actual measurements is extracted.
2. the photovoltaic array parameter extracting method according to claim 1 based on backward learning and the complicated evolution of enhancing,
It is characterized in that, further includes following steps in the step S1:
Step S11: acquiring output current and voltage data of photovoltaic panel under the conditions of different illumination, irradiation level by collection plate,
And it is saved as to the number table of N*1 respectively;Wherein, N is the number of collecting sample point, and the sample point of the electric current and voltage acquired
Quantity is identical;
Step S12: select single diode model and double diode model as photovoltage model;
Single diode model are as follows:
Wherein, IpvIt is photogenerated current, IoIt is the saturation current of diode, a is the ideal factor of diode, RsIt is equivalent parallel electricity
Resistance, RshIt is equivalent series resistance, K is Boltzmann constant, and T is environment temperature, and q is the absolute value of electronic charge;
Double diode model are as follows:
Wherein, IpvIt is photogenerated current, Io1,Io2The saturation current of two different diodes respectively, a1,a2Respectively two diodes
Ideal factor, RsIt is equivalent parallel resistance, RshIt is equivalent series resistance, K is Boltzmann constant, and T is environment temperature, and q is
The absolute value of electronic charge.
3. the photovoltaic array parameter extracting method according to claim 2 based on backward learning and the complicated evolution of enhancing,
It is characterized in that, in the step S2, the objective function namely fitness function obtain as follows:
Remember root-mean-square error are as follows:
Wherein, N is the total sample number on I-V characteristic curve, and f (V, I, X) is absolute between measured current and model estimation electric current
Error;
For the f (V, I, X) of single diode are as follows:
For the f (V, I, X) of double diode are as follows:
4. the photovoltaic array parameter extracting method according to claim 3 based on backward learning and the complicated evolution of enhancing,
It is characterized in that, further includes following steps in the step S3:
Step S31: corresponding control parameter is initialized, comprising: the number solved in the quantity npg of complex shape, each complex shape
Ngs, the sample number npt of total candidate solution, candidate solution lower coboundary BL and BU;
Step S32: random initializtion, process are carried out in the lower upper bound BL, BU to npt candidate samples point are as follows:
xi=BL+rand* (BU-BL), i=1,2 ..., npt
And calculate the fitness function value of its each position;
Step S33: opposition solution x is obtained in the way of backward learning to candidate solutiono, process are as follows:
xo=BU+BL-xi
The fitness function value of opposition solution is calculated, and by xoAnd xiMixing postscript is new candidate solution s, according to its fitness function
Size carry out ascending order arrangement.
5. the photovoltaic array parameter extracting method according to claim 4 based on backward learning and the complicated evolution of enhancing,
It is characterized in that, further includes following steps in the step S4:
Step S41: npt optimal candidate solution is divided into npg complex shape;
Step S42: each complex shape is evolved by enhanced complicated evolution algorithm, obtains the position of new candidate solution;
Step S43: the optimization point in the complex shape after evolution is replaced into original complex shape candidate point fitness letter corresponding with its
Numerical value;
Step S44: when objective function calculation times reach maximum, end algorithm returns to optimal solution.
6. the photovoltaic array parameter extracting method according to claim 5 based on backward learning and the complicated evolution of enhancing,
It is characterized in that, further includes following steps in the step S52:
Step S521: initialization control parameter;Note simplex number of vertex is q, and control coefrficient is respectively α, beta, gamma, primary and secondary complex shape
Evolution number is (m, n);
Step S522: the vertex that q point makees ngs as simplex is selected from npt point of complex shape, and calculates this q point
Central point ce, calculation method are as follows:Wherein, ujIt is the position on q-1 simplex vertex
It sets;
Step S523: corresponding reflection point u is obtained according to central pointr=ce- α (ce-uq), after boundary Control, calculate
The fitness function value of reflection point;If the fitness function value of reflection point is less than the optimum point in the complex shape, according to as follows
Mode calculates the numerical value of its extension point: ue=ce+ γ (ce-uq), and optimal position in reflection point and extension point is replaced and is replaced
Worst position in this vertex;
Step S524: when reflection point position is poorer than the Best Point in complex shape, then according to the optimum point in the complex shape, center
Point, most not good enough and reflection point, generate compression point as follows:Its
In, ugAnd uqIt is globe optimum and most not good enough position, u respectivelyrIt is the position of reflection point, β is contraction factor, Fq-1It is secondary poor
The fitness function value of point, FrIt is the fitness function value of reflection point, FqIt is most not good enough fitness function value;
Step S525: if compression point cannot still update most almost, according to globe optimum and local best points, according to such as lower section
Formula optimizes worst point: uz=uq+β(ug-uq)+γ×r·(ub-uq), r ∈ (- 1,1), wherein uzIt is the position for updating point
It sets, ugIt is globe optimum, ubIt is local best points, r is the random number between -1 and 1, and γ is spreading factor;
Step S526: it repeats step S522 to step S525n times, step S521 to step S525m times.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810946865.5A CN109388845B (en) | 2018-08-19 | 2018-08-19 | Photovoltaic array parameter extraction method based on reverse learning and enhanced complex evolution |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810946865.5A CN109388845B (en) | 2018-08-19 | 2018-08-19 | Photovoltaic array parameter extraction method based on reverse learning and enhanced complex evolution |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109388845A true CN109388845A (en) | 2019-02-26 |
CN109388845B CN109388845B (en) | 2022-12-13 |
Family
ID=65417702
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810946865.5A Active CN109388845B (en) | 2018-08-19 | 2018-08-19 | Photovoltaic array parameter extraction method based on reverse learning and enhanced complex evolution |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109388845B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112182883A (en) * | 2020-09-28 | 2021-01-05 | 上海交通大学 | Photovoltaic module parameter estimation method based on self-adaptive LM |
CN113037213A (en) * | 2021-03-04 | 2021-06-25 | 国网新疆电力有限公司信息通信公司 | Photovoltaic cell model parameter identification method and device based on goblet sea squirt group algorithm |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104484833A (en) * | 2014-12-02 | 2015-04-01 | 常州大学 | Photovoltaic power generation output power tracking algorithm based on genetics algorithm improved RBF-BP neural network |
CN104820877A (en) * | 2015-05-21 | 2015-08-05 | 河海大学 | Photovoltaic system generation power prediction method based on cloud adaptive PSO-SNN |
CN105138065A (en) * | 2015-07-29 | 2015-12-09 | 三峡大学 | Photovoltaic power generation system MPPT algorithm based on voltage pre-estimating method |
KR20160032536A (en) * | 2014-09-16 | 2016-03-24 | 한국전자통신연구원 | Signal process algorithm integrated deep neural network based speech recognition apparatus and optimization learning method thereof |
CN106485075A (en) * | 2016-10-12 | 2017-03-08 | 福州大学 | A kind of based on hawk strategy and the photovoltage model parameter identification method of adaptive N M simplex |
CN108111125A (en) * | 2018-01-29 | 2018-06-01 | 福州大学 | The scanning of IV characteristic curves and the parameter identification system and method for a kind of photovoltaic array |
-
2018
- 2018-08-19 CN CN201810946865.5A patent/CN109388845B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20160032536A (en) * | 2014-09-16 | 2016-03-24 | 한국전자통신연구원 | Signal process algorithm integrated deep neural network based speech recognition apparatus and optimization learning method thereof |
CN104484833A (en) * | 2014-12-02 | 2015-04-01 | 常州大学 | Photovoltaic power generation output power tracking algorithm based on genetics algorithm improved RBF-BP neural network |
CN104820877A (en) * | 2015-05-21 | 2015-08-05 | 河海大学 | Photovoltaic system generation power prediction method based on cloud adaptive PSO-SNN |
CN105138065A (en) * | 2015-07-29 | 2015-12-09 | 三峡大学 | Photovoltaic power generation system MPPT algorithm based on voltage pre-estimating method |
CN106485075A (en) * | 2016-10-12 | 2017-03-08 | 福州大学 | A kind of based on hawk strategy and the photovoltage model parameter identification method of adaptive N M simplex |
CN108111125A (en) * | 2018-01-29 | 2018-06-01 | 福州大学 | The scanning of IV characteristic curves and the parameter identification system and method for a kind of photovoltaic array |
Non-Patent Citations (1)
Title |
---|
刘萌萌: "太阳电池性能参数的提取方法和影响因素的研究", 《万方数据学位论文库》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112182883A (en) * | 2020-09-28 | 2021-01-05 | 上海交通大学 | Photovoltaic module parameter estimation method based on self-adaptive LM |
CN112182883B (en) * | 2020-09-28 | 2022-06-24 | 上海交通大学 | Photovoltaic module parameter estimation method based on self-adaptive LM |
CN113037213A (en) * | 2021-03-04 | 2021-06-25 | 国网新疆电力有限公司信息通信公司 | Photovoltaic cell model parameter identification method and device based on goblet sea squirt group algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN109388845B (en) | 2022-12-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ismaeel et al. | Gradient-based optimizer for parameter extraction in photovoltaic models | |
CN109873610B (en) | Photovoltaic array fault diagnosis method based on IV characteristic and depth residual error network | |
CN109932627A (en) | A kind of localization method and system of GIS pressure test shelf depreciation | |
CN109165774A (en) | A kind of short-term photovoltaic power prediction technique | |
CN107103154A (en) | A kind of photovoltaic module model parameter identification method | |
CN106295899B (en) | Wind power probability density Forecasting Methodology based on genetic algorithm Yu supporting vector quantile estimate | |
WO2019056753A1 (en) | Dynamic equivalent modeling method for distributed photovoltaic power station cluster | |
CN103106544B (en) | A kind of photovoltaic generation prognoses system based on T-S Fuzzy neutral net | |
CN110503153B (en) | Photovoltaic system fault diagnosis method based on differential evolution algorithm and support vector machine | |
CN106126863A (en) | Based on artificial fish-swarm and the photovoltaic cell parameter identification method of the algorithm that leapfrogs | |
CN103955521B (en) | Cluster classification method for wind power plant | |
CN105574615A (en) | Spatial correlation and genetic algorithm (GA) based wind power forecast method for wavelet-BP neural network | |
CN108062720A (en) | A kind of load forecasting method based on similar day selection and random forests algorithm | |
CN108537683A (en) | A kind of load forecasting method based on similar day selection and random forests algorithm | |
CN106485075A (en) | A kind of based on hawk strategy and the photovoltage model parameter identification method of adaptive N M simplex | |
CN109388845A (en) | Based on backward learning and the complicated photovoltaic array parameter extracting method evolved of enhancing | |
CN103996079A (en) | Wind power weighting predication method based on conditional probability | |
CN112418346A (en) | Numerical weather forecast total radiation system error classification calculation method | |
CN115759467A (en) | Time-division integrated learning photovoltaic prediction method for error correction | |
CN115759445A (en) | Machine learning and cloud model-based classified flood random forecasting method | |
CN109543269B (en) | Photovoltaic model parameter identification method based on mixed trust interval reflection | |
WO2022105060A1 (en) | Method for identifying parameters of photovoltaic cell having single diode structure | |
CN109767353A (en) | A kind of photovoltaic power generation power prediction method based on probability-distribution function | |
CN113595132A (en) | Photovoltaic online parameter identification method based on global maximum power point tracking and hybrid optimization algorithm | |
Chen et al. | Graph neural network-based wind farm cluster speed 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |