CN115600503A - Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method - Google Patents

Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method Download PDF

Info

Publication number
CN115600503A
CN115600503A CN202211382714.4A CN202211382714A CN115600503A CN 115600503 A CN115600503 A CN 115600503A CN 202211382714 A CN202211382714 A CN 202211382714A CN 115600503 A CN115600503 A CN 115600503A
Authority
CN
China
Prior art keywords
simplex
vertex
simulated annealing
compound battery
annealing method
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211382714.4A
Other languages
Chinese (zh)
Inventor
吕辉
郭灿
请求不公布姓名
姚育成
柴世一
陈浩
张益豪
李莎莎
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hubei University of Technology
Original Assignee
Hubei University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hubei University of Technology filed Critical Hubei University of Technology
Priority to CN202211382714.4A priority Critical patent/CN115600503A/en
Publication of CN115600503A publication Critical patent/CN115600503A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a multi-junction compound battery model parameter optimization method based on a simplex-simulated annealing method, which comprises the following steps: s100: obtaining a plurality of groups of measured data samples of the current and the voltage of the multijunction compound battery; s200: constructing a physical model of the multi-junction compound battery, and determining an objective function of an optimization problem, wherein the objective function is constructed based on the minimum difference between a predicted value and an actual measured value of the output current of the multi-junction compound battery; s300: searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then performing random search by using the simulated annealing method to obtain an optimal solution of the model parameter to be optimized. The method has the advantages of fast convergence, obviously improved search efficiency and search precision, and greatly improved precision and reliability of the multi-junction compound battery model, thereby obtaining the battery model with accurate prediction capability and having stronger applicability.

Description

Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method
Technical Field
The application belongs to the technical field of photovoltaic power generation, and particularly relates to a multi-junction compound battery model parameter optimization method based on a simplex-simulated annealing method.
Background
The multijunction compound cell is a photovoltaic cell formed by sequentially laminating and extending different materials according to the forbidden band width from high to low, and can absorb light energy of corresponding wave bands in a light source and convert the light energy into electric energy. In this way, the multijunction compound cell has a wider External Quantum Efficiency (EQE), thereby absorbing wider spectral light energy and finally having higher conversion efficiency. At present, the conversion efficiency of the multijunction compound cell reaches 46.5% under the condition of high-concentration, and the multijunction compound cell is the cell with the highest photoelectric conversion efficiency in various photovoltaic cells.
The multijunction compound battery generally adopts a three-junction mode, namely a three-junction compound battery, and the structure and the equivalent circuit diagram thereof are shown in fig. 1: the first layer is GaInP (forbidden band width Eg =1.86 eV), the middle layer is InGaAs (forbidden band width Eg =1.4 eV), and the bottom layer is Ge (forbidden band width Eg =0.65 eV). The sensitive wavelength range reaches 280nm-1950nm, and the energy components of incident light in various wavelengths can be fully utilized. Germanium Ge has good mechanical properties as the bottom material, is not easy to break, has relatively low price and mature production technology. Compared with the traditional photovoltaic cell, the multijunction compound cell has excellent high-temperature characteristic, can continue to work when the temperature reaches 200 ℃, can improve the output power by high-power light condensation, and is widely applied to aerospace and ground light condensation systems at present.
In order to more accurately analyze and predict the electrical characteristics of the multijunction compound battery, accurate modeling simulation needs to be performed on the multijunction compound battery, and therefore how to obtain the optimal parameter value of the multijunction compound battery model is also particularly important. At present, battery model parameters are optimized, an analysis method is usually adopted to obtain initial values of the parameters, and then an iteration method is adopted to search optimal solutions of the parameters in a region near the initial values. For example, the chinese patent CN 113343410A discloses an optimal parameter solution method for a photovoltaic cell model, this method has problems: the searching of the optimal solution of the parameters depends on the initial values of the parameters, and once the initial values are improperly calculated, the accuracy of the optimal solution is influenced; and also tends to fall into a locally optimal solution.
Disclosure of Invention
In view of the above problems in the prior art, the present application provides a method for optimizing parameters of a multijunction compound battery model based on a simplex-simulated annealing method.
According to the method, a simplex searching method based on a convex polyhedron structure in a high-dimensional Euclidean space and a simulated annealing method with probability snap-through characteristics are organically combined, the simplex searching method has the characteristic of rapid convergence, the simulated annealing method has global searching capability due to the probability snap-through characteristics, and the searching capability and the searching efficiency can be enhanced by combining the simplex searching method and the simulated annealing method, so that the accuracy is improved.
The application provides a multi-junction compound battery model parameter optimization method based on a simplex-simulated annealing method, which comprises the following steps:
s100: obtaining multiple groups of measured data samples of current and voltage of the multijunction compound battery, wherein the multiple groups of measured data samples of the current and the voltage are obtained by performing I-V characteristic test on the multijunction compound battery;
s200: constructing a physical model of the multi-junction compound battery, and determining an objective function f (X) of an optimization problem, wherein X represents a parameter vector of the model to be optimized; the objective function is constructed based on the minimum difference value between the predicted value and the measured value of the output current of the multi-junction compound battery, and the predicted value is calculated by using the physical model;
s300: searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then performing random search by using the simulated annealing method to obtain an optimal solution of the model parameter to be optimized.
In some embodiments of the present application, the physical model of the multi-junction compound cell is a physical model in which the sub-cell model is selected from a single exponential model, such as a single diode model.
In some embodiments of the present application, the objective function f (X) is constructed as follows:
Figure BDA0003929172970000031
wherein: n represents the total group number of the actually measured data samples, and j represents the group number of the actually measured data samples; x denotes the parameter vector of the model to be optimized, I L And V L Respectively, the output current and the output voltage of the multijunction compound battery, F (V) L ,I L X) represents the difference between the predicted value and the measured value of the output current of the multijunction compound battery, F (V) L ,I L ,X) j F (V) corresponding to the j-th group of measured data samples L ,I L ,X)。
In some specific implementation manners of the embodiments of the present application, step S300 specifically includes:
(1) Searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then executing the step (2);
(2) Randomly updating the individuals by using a simulated annealing method; after the updating individual at each temperature is finished, judging whether convergence occurs or not, if so, ending, wherein the current updating individual is the optimal solution of the model parameter to be optimized; otherwise, the temperature is updated, the vertex of the simplex is reconstructed based on the current updated individuals, and then the step (1) is executed.
In some specific implementation manners of the embodiments of the present application, the step (1) further includes the sub-steps of:
s310: initializing, namely initializing m +1 vertexes of a simplex in a given search range, and presetting the calculation times D of function values at each temperature;
s320: calculating function value f (X) of each vertex to obtain optimal value f G The second optimum value f H The worst value f L And divide intoRespectively corresponding vertex X G 、X H 、X L (ii) a Calculating the removed vertex X G Centroid of posterior remaining vertex
Figure BDA0003929172970000032
S330: obtain vertex X G About the centroid
Figure BDA0003929172970000034
Point of symmetry X of m+2 Calculating the symmetric point X m+2 Function value f (X) m+2 ) (ii) a Judgment of f (X) m+2 ) And f L The size of (2):
when f (X) m+2 )＜f L When it is used, order
Figure BDA0003929172970000033
Gamma is a preset expansion coefficient; computing vertex X m+3 Function value f (X) m+3 ) Let D = D-1, and further compare f (X) m+3 ) And f (X) m+2 ) The size of (c): when f (X) m+2 )＜f(X m+3 ) Using the vertex X m+3 Replacement vertex X G Then go to step S350; otherwise, use vertex X m+2 Replacement vertex X G Then go to step S350;
when f is L ≤f(X m+2 )≤f H Using the vertex X m+2 Replacement vertex X G Then go to step S350;
when f (X) m+2 )＞f H If yes, go directly to step S340;
s340: take f (X') = min { f (X) m+2 ),f(X H ) } order of
Figure BDA0003929172970000041
X 'represents a vertex corresponding to the function value f (X'), and beta is a preset expansion coefficient; computing vertex X m+4 Function value f (X) m+4 ) Let D = D-1, and further compare f (X) m+4 ) And f (X') are: when f (X) m+4 ) When f (X') is less than or equal to f, the vertex X is used m+4 Replacement of vertex X G Then go to step S350; when f (X) m+4 )＞If f (X'), performing edge narrowing, and making D = D-m, and then proceeding to step S350;
s350: judging whether D is less than or equal to 0, if so, ending simplex search, and finally obtaining X G I.e. local minima; otherwise, go to step S320 and repeat the process.
In some specific implementation manners of the embodiments of the present application, the step (2) further includes the sub-steps of:
s360: taking the local minimum point as an initial individual of a simulated annealing method, randomly updating the individual by using the simulated annealing method, and calculating a function value of the updated individual;
s370: judging whether to receive the updated individual based on the Metropolis acceptance criterion, and executing the step S380 when judging to receive the updated individual; otherwise, go to step S360 to renew the individual randomly;
s380: updating the current temperature, and then executing step S390;
s390: judging whether convergence occurs or not, and if yes, ending; otherwise, reconstructing the vertex of the simplex based on the currently received updated individuals, and then executing the step (1).
Further, the individual is randomly updated by using a simulated annealing method, and the individual X is randomly updated according to the step length and the normally distributed random number 1 '＝X 1 +ρS 0 Wherein X is 1 And X 1 ' represents a pre-update individual and an update individual, respectively; rho represents a normally distributed random number, rho is in the range of (-1,1); s 0 Representing a preset step vector.
Further, whether to receive the updated individual is judged based on the Metropolis acceptance criterion, which specifically includes:
when the function value of the updated individual is not larger than the updated individual, judging to receive the updated individual; otherwise, further calculating the probability
Figure BDA0003929172970000051
And generates a random number delta E [0,1]If delta is less than or equal to P, judging to receive the updated individual; otherwise, judging not to receive the updated individual; wherein T represents the current temperature, f (X) 1 ) And f (X) 1 ') indicates the individual before update and the updated individual, respectivelyThe function value of the volume.
Compared with the prior art, the application has the following advantages and beneficial effects:
the simplex searching method and the simulated annealing method are organically combined, and the simplex searching method has high convergence rate; the simulated annealing method has global searching capability due to the probability jump characteristic, and can effectively avoid trapping in a local optimal solution; the combination of the two is beneficial to enriching the search behavior in the optimization process, increasing the search capability and the search efficiency and improving the search precision, thereby obtaining a battery model with more accurate prediction capability.
The method can avoid the dependence on the initial value of the parameter and the partial optimal value in the optimization process, has fast convergence compared with the optimization method in the background technology, can obviously improve the search efficiency and the search precision, and can greatly improve the precision and the reliability of the multi-junction compound battery model, thereby obtaining the battery model with accurate prediction capability and having stronger applicability.
Drawings
Fig. 1 is a structural view of a triple junction compound battery and an equivalent circuit diagram thereof;
fig. 2 is a flowchart of an embodiment of the present application.
Detailed Description
The technical solutions of the embodiments of the present application will be described clearly and completely with reference to the accompanying drawings, and it is to be understood that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
The target cell in the present embodiment is a lattice-matching type triple-junction compound cell, but the method of the present invention is not limited to the triple-junction compound cell, and the following description will be given only by taking a triple-junction compound cell as an example. Referring to fig. 1, there is shown a structure of a triple junction compound battery used in the embodiment of the present application and an equivalent circuit diagram thereof. For a multijunction compound cell, there are classified into a lattice matching type, a band gap matching type and an inverse type according to an epitaxial structure and a manufacturing processAnd growing a gradually changed buffer layer type. The three-junction compound battery shown in FIG. 1 is a lattice-matched GaInP/GaInAs/Ge three-junction compound battery, specifically Ga 0.51 In 0.49 P/Ga 0.99 In 0.01 The As/Ge type three-junction compound battery has a structure in which sub-batteries are connected in series, and a single-index model can be selected for each sub-battery model.
Referring to fig. 2, a detailed flowchart of an embodiment of the present application is shown, and a detailed implementation process of the embodiment of the present application will be provided with reference to fig. 2, which includes the following steps:
s100: and acquiring multiple groups of measured data samples of the current and the voltage of the multijunction compound battery, wherein the multiple groups of measured data samples of the current and the voltage are obtained by carrying out I-V characteristic test on the multijunction compound battery.
The acquisition mode of the measured data sample in the embodiment of the application is as follows: under the condition of the temperature of 25 ℃, a sunlight simulator is used as a light source, light beams are irradiated on the triple-junction compound battery, the triple-junction compound battery converts light energy into electric energy, and current and voltage data of the triple-junction compound battery can be tested through an electronic load.
S200: and (3) constructing a physical model of the multijunction compound battery, and determining an objective function f (X) of the optimization problem.
First, a physical model of a multijunction compound cell was constructed.
In the embodiment of the application, each sub-battery model of the multi-junction compound battery selects a single-exponential model, specifically selects a single-diode model, and the formula is as follows:
Figure BDA0003929172970000061
in formula (1):
I L represents the output current of a multijunction compound battery, I ph,i Denotes the photo-generated current of the sub-cell I 0,i Represents the reverse saturation current of a single diode of the sub-battery i; v i Indicating the output voltage, V, of the sub-cell i L Representing the total output voltage of the multijunction compound battery; a. The i Representing the actual effective area of the sub-cell i,R s,i denotes the series resistance, R, of the subcell i sh,i Represents the parallel resistance of the sub-cell i; n is i A diode quality factor representing the subcell i; v T K denotes a boltzmann constant, T denotes a kelvin temperature of the multi-junction compound battery, and q denotes an electron charge.
The above i represents the number of the sub-battery, and in the embodiment of the present application, when the target battery is a triple-junction compound battery, i is 1,2,3 in sequence. The single diode model for a three-junction compound battery then contains ten unknown model parameters, i.e. model parameters to be optimized, expressed as follows:
X＝{I ph,1 ,I 0,1 ,n 1 ,I ph,2 ,I 0,2 ,n 2 ,I ph,3 ,I 0,3 ,n 3 ,R s } (2)
in the formula (2), R s Represents the series resistance of the multijunction compound cell, which is equivalent to the sum of the series resistances of the sub-cells, i.e., R s ＝R s,1 +R s,2 +R s,3 。
Next, an objective function of the optimization problem is determined based on a physical model of the multijunction compound cell.
According to the embodiment of the application, the target function is constructed by minimizing the difference value between the predicted value and the measured value of the output current of the multijunction compound battery; the predicted value is the output current calculated by using the physical model, and the measured value is the output current from the measured data sample.
The constructed objective function f (x) is shown in formulas (3) to (4):
Figure BDA0003929172970000071
Figure BDA0003929172970000072
in the formulae (3) to (4), RMSE represents a root mean square error; n represents the total group number of the actually measured data samples, and j represents the group number of the actually measured data samples; s represents the number of sub-cells in the multi-junction compound cellIn the embodiment of the application, s =3; f (V) L ,I L X) represents the difference between the predicted value and the measured value of the output current of the multijunction compound battery calculated in the iteration, and F (V) L ,I L ,X) j F (V) corresponding to the j-th group of measured data samples L ,I L X); when the predicted value is infinitely close to the measured value, namely the objective function f (X) is minimum, the corresponding model parameter X is the optimal parameter value of the multi-junction compound battery model.
In the formula (4), the reaction mixture is,
Figure BDA0003929172970000073
representing the predicted value calculated in iteration, I L ' denotes an actual measurement value from an actual measurement data sample.
And each model parameter to be optimized has a corresponding search range, the search range of each model parameter to be optimized is preset before parameter optimization, and the optimization is carried out in the preset search range. For diode quality factor n i The value of the multijunction compound battery is generally between 1 and 2; for photo-generated current I ph,i The value is generally 11mA/cm 2 To 20mA/cm 2 To (c) to (d); reverse saturation current I for a single diode 0,i Generally, the value is between 0 muA and 1 muA; for the series resistance R s And generally ranges from 0 Ω to 0.1 Ω. The boundary settings for the ten model parameters to be optimized in the examples of the present application are listed in table 1.
TABLE 1 boundaries of model parameters to be optimized
Parameter(s) n 1 n 2 n 3 I ph,1 I ph,2 I ph,3 I 0,1 I 0,2 I 0,3 R s
Upper limit of 2 2 2 20 20 20 1 1 1 0.1
Lower limit of 1 1 1 11 11 11 0 0 0 0
S300: searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then performing random search by using the simulated annealing method to obtain an optimal solution of the model parameter to be optimized.
In the embodiment of the application, a simplex search method is firstly utilized to search a local minimum point of a model parameter to be optimized in a given search range, in order to avoid trapping in a local optimal solution, the local minimum point is taken as an initial individual of a simulated annealing method, then the simulated annealing method is adopted to carry out random search, the local minimum point is jumped away, and a global optimal solution is obtained through circulation along with annealing operation. When a simulated annealing method is used for random search, the individual is randomly updated at each temperature, when the updating of the individual is completed, whether convergence occurs or not is judged, and if the convergence occurs, the current updating individual is output, namely the optimal solution of the model parameter to be optimized; otherwise, updating the temperature, reconstructing the vertex of the simplex at the new temperature based on the current updated individuals, and then searching a new local minimum point by adopting the simplex searching method again until a global optimal solution is obtained.
The specific implementation of this step will be provided below as follows:
s310: initialization of simplex search method, including at least setting initial temperature T 0 Initialization step S 0 Margin of error sigma, annealing end temperature epsilon T Initializing m +1 vertexes of the simplex, wherein each vertex corresponds to one X, and setting the calculation times D of the objective function at each temperature.
In the embodiment of the present application, the initial temperature T is set 0 =100 ℃, initial step vector S 0 ＝[1,1,1,1,1,1,1,1,1,1]Error margin σ =10 -8 Annealing end temperature ε T ＝0.001℃，D＝200。
S320: respectively solving function values f (X) corresponding to m +1 vertexes by using formulas (3) - (4) to obtain an optimal value f G The second optimum value f H The worst value f L Optimum value f G Sub-optimum value f H The worst value f L The corresponding vertexes are respectively marked as vertex X G 、X H 、X L (ii) a Removing vertex X G Calculating the centroids of the remaining m vertices
Figure BDA0003929172970000093
. The optimum value refers to the smallest function value, the second best value refers to the second smallest function value, and the worst value refers to the largest function value.
S330: obtain vertex X G About centroid
Figure BDA0003929172970000094
Is marked as X m+2 Calculating the symmetric point X m+2 Function value f (X) m +2 ) (ii) a Comparison f (X) m+2 ) And f L The size of (2):
when f (X) m+2 )＜f L When it is used, order
Figure BDA0003929172970000091
The expansion coefficient gamma is generally a value larger than 1, in the embodiment of the application, the expansion coefficient gamma =1+2/dim, dim represents the dimension of a parameter vector of a model to be optimized, namely dim is 10; then, X is calculated m+3 Function value f (X) m+3 ) Let D = D-1, and further compare f (X) m+3 ) And f (X) m+2 ) The size of (2): when f (X) m+2 )＜f(X m+3 ) Then use the vertex X m+3 Replacement vertex X G And go to step S350; otherwise, use vertex X m+2 Replacement of vertex X G Then go to step S350;
when f is L ≤f(X m+2 )≤f H Then use the vertex X m+2 Replacement vertex X G Then, go to step S350;
when f (X) m+2 )＞f H If so, it indicates that the step size is too large and needs to be retracted, and the process goes to step S340.
S340: take f (X') = min { f (X) m+2 ),f(X H ) Instruction of
Figure BDA0003929172970000092
X 'represents a vertex corresponding to the function value f (X'), the value range of β is (0,1), in the embodiment of the present application, β is 0.5; computing vertex X m+4 Function value f (X) m+4 ) Let D = D-1, and further compare f (X) m+4 ) And f (X') is: when f (X) m+4 ) When f (X') is less than or equal to f, the vertex X is used m+4 Replacement of vertex X G Then go to step S350; when f (X) m+4 ) If f (X'), then performing edge reduction on all the current vertexes, namely making X k ＝(X k +X L ) And/2, where k =1,2,.., m, and let D = D-m, then go to step S350.
S350: judging whether D is less than or equal to 0, if so, ending simplex search, and finally obtaining X G Namely local minimum point, then executing step S360 to enter into simulated degradation search; otherwise, go to step S320 to repeat the iteration.
S360: using local minimum point as initial individual of simulated annealing method to call simulated annealing method, and updating individual X randomly according to step length and normally distributed random number 1 '＝X 1 +ρS 0 Wherein X is 1 And X 1 ' denotes a pre-update individual and an update individual, respectively, X 1 Initializing to local minimum points obtained by a simplex searching method; ρ represents a random number of a normal distribution, and ρ ∈ (-1,1).
S370: judging whether to receive the updated individual based on the Metropolis acceptance criterion:
comparing the function value f (X) of the pre-update individual and the update individual 1 ) And f (X) 1 ') size, when f (X) 1 ')≤f(X 1 ) Then receive the updated individual X 1 ', then step S380 is performed; otherwise, calculating the probability
Figure BDA0003929172970000101
And generates a random number delta epsilon [0,1]When delta is less than or equal to P, receiving the updated individual X 1 ', then step S380 is performed; otherwise, not receiving the updated individual X 1 ' go to step S360 to update the individuals randomly again. Wherein T represents the current temperature, with the stackThe times are increased and the temperature is gradually reduced.
S380: update the temperature T l+1 ＝ωT l Then, step S390 is executed; wherein, T l Indicating the current temperature, T l+1 Indicates the updated temperature, ω indicates the cooling index, and the value is between 0 and 1, in the embodiment of the present application, ω =0.95.
S390: judging whether convergence occurs or not, and if yes, ending; otherwise, reset D and go to step S320, and re-execute simplex search at the next temperature. Turning to step S320, the updated individuals received by the simulated annealing method reconstruct m +1 vertices of the simplex.
The convergence condition is preset, and the convergence condition in the embodiment of the application is as follows: and when the difference value of the individual function values before and after updating is smaller than a preset error tolerance or the temperature after updating is smaller than a preset annealing end temperature, judging that the temperature is convergent.
And constructing a multi-junction compound battery model by using the optimal solution of the model parameters to be optimized, wherein the constructed multi-junction compound battery model can be used for predicting the performance of the multi-junction compound battery. The optimal solution of the determined model parameters in the embodiment of the application is shown in table 2 below, and the root mean square error RMSE is calculated according to the optimized model parameters, wherein the predicted value and the measured value of the output current of the multijunction compound battery are calculated, and the root mean square error is 9.96 multiplied by 10 -7 It shows that the method of the present application has excellent accuracy.
TABLE 2 optimal solution of model parameters in the examples of the present application
Model parameters Optimum value
I ph,1 1.84
I 0,1 1.59
n 1 1.37
I ph,2 1.20×10 1
I 0,2 1.12×10 1
n 2 1.67×10 1
I ph,3 2.75×10 -9
I 0,3 7.71×10 -13
n 3 5.25×10 -1
R s 1.16×10 -2
RMSE 9.96×10 -7
The method is realized based on the combination of a simplex searching method and a simulated annealing method, firstly, the simplex searching method is utilized to quickly obtain the local minimum point of the model parameter to be optimized, the local minimum point is taken as an initial individual, then, the simulated annealing method is utilized to optimize the model parameter to be optimized, the jump characteristic of the simulated annealing method enables the local minimum value to jump out, and the global optimal solution is obtained through circulation along with the annealing operation.
It should be noted that the foregoing is only illustrative of the preferred embodiments of the present application and the technical principles employed. It will be understood by those skilled in the art that the present application is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the application. Therefore, although the present application has been described in more detail with reference to the above embodiments, the present application is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present application.

Claims (7)

1. The method for optimizing the parameters of the multi-junction compound battery model based on the simplex-simulated annealing method is characterized by comprising the following steps of:
s100: obtaining a plurality of groups of measured data samples of the current and the voltage of the multijunction compound battery, wherein the plurality of groups of measured data samples of the current and the voltage are obtained by carrying out I-V characteristic test on the multijunction compound battery;
s200: constructing a physical model of the multi-junction compound battery, and determining an objective function f (X) of an optimization problem, wherein X represents a parameter vector of the model to be optimized; the objective function is constructed based on the minimum difference value between the predicted value and the measured value of the output current of the multi-junction compound battery, and the predicted value is calculated by using the physical model;
s300: searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then performing random search by using the simulated annealing method to obtain an optimal solution of the model parameter to be optimized.
2. The method for optimizing parameters of a multijunction compound battery model based on simplex-simulated annealing method according to claim 1, wherein:
in the physical model of the multi-junction compound battery, a single exponential model is selected as a sub-battery model.
3. The method for optimizing parameters of a multijunction compound battery model based on simplex-simulated annealing method according to claim 1, wherein:
the objective function
Figure FDA0003929172960000011
Wherein:
n represents the total group number of the actually measured data samples, and j represents the group number of the actually measured data samples; x represents the parameter vector of the model to be optimized, I L And V L Respectively, the output current and the output voltage of the multijunction compound battery, F (V) L ,I L X) represents the difference between the predicted value and the measured value of the output current of the multijunction compound battery, F (V) L ,I L ,X) j F (V) corresponding to the j-th group of measured data samples L ,I L ,X)。
4. The method for optimizing parameters of a multijunction compound battery model based on simplex-simulated annealing method according to claim 1, wherein:
step S300 specifically includes:
(1) Searching a local minimum point of a model parameter to be optimized in a given search range by using a simplex search method, taking the local minimum point as an initial individual of a simulated annealing method, and then executing the step (2);
(2) Randomly updating the individuals by using a simulated annealing method; after the updating individual at each temperature is finished, judging whether convergence occurs or not, if so, ending, wherein the current updating individual is the optimal solution of the model parameter to be optimized; otherwise, the temperature is updated, the vertex of the simplex is reconstructed based on the current updated individuals, and then the step (1) is executed.
5. The simplex-simulated annealing method-based multijunction compound battery model parameter optimization method of claim 4, wherein:
the step (1) further comprises:
s310: initializing, namely initializing m +1 vertexes of a simplex in a given search range, and presetting the calculation times D of function values at each temperature;
s320: calculating function values f (X) of all vertexes to obtain optimal values f G Sub-optimum value f H The worst value f L And respectively corresponding vertexes X G 、X H 、X L (ii) a Calculating the removed vertex X G Centroid of posterior remaining vertex
Figure FDA0003929172960000021
S330: obtain vertex X G About the centroid
Figure FDA0003929172960000022
Point of symmetry X m+2 Calculating the symmetric point X m+2 Function value f (X) m+2 ) (ii) a Judgment of f (X) m +2 ) And f L The size of (c):
when f (X) m+2 )＜f L When it is used, order
Figure FDA0003929172960000023
Gamma is a preset expansion coefficient; computing vertex X m+3 Function value f (X) m+3 ) And let D = D-1, further compare f (X) m+3 ) And f (X) m+2 ) The size of (2): when f (X) m+2 )＜f(X m+3 ) Using the vertex X m+3 Replacement of vertex X G Then go to step S350; otherwise, use vertex X m+2 Replacement of vertex X G Then go to step S350;
when f is L ≤f(X m+2 )≤f H Using the vertex X m+2 Replacement of vertex X G Then go to step S350;
when f (X) m+2 )＞f H If yes, go directly to step S340;
s340: take f (X') = min { f (X) m+2 ),f(X H ) Instruction of
Figure FDA0003929172960000024
X 'represents a vertex corresponding to the function value f (X'), and beta is a preset expansion coefficient; computing vertex X m+4 Function value f (X) m+4 ) Let D = D-1, and further compare f (X) m+4 ) And f (X') is: when f (X) m+4 ) When f (X') is less than or equal to f, the vertex X is used m+4 Replacement of vertex X G Then go to step S350; when f (X) m+4 ) If f (X'), performing edge shrinking, and enabling D = D-m, and then turning to the step S350;
s350: judging whether D is less than or equal to 0, if so, ending simplex search, and finally obtaining X G I.e. local minima; otherwise, go to step S320 and repeat the process.
6. The simplex-simulated annealing method-based multijunction compound battery model parameter optimization method of claim 4, wherein:
the step (2) further comprises the following steps:
s360: taking the local minimum point as an initial individual of a simulated annealing method, and randomly updating the individual by using the simulated annealing method;
s370: judging whether to receive the updated individual based on the Metropolis acceptance criterion, and executing the step S380 when judging to receive the updated individual; otherwise, go to step S360 to renew the individual randomly;
s380: updating the current temperature, and then executing step S390;
s390: judging whether convergence occurs or not, and if yes, ending; otherwise, reconstructing the vertex of the simplex based on the currently received updated individuals, and then executing the step (1).
7. The simplex-simulated annealing method-based multijunction compound battery model parameter optimization method of claim 6, wherein:
the individuals are randomly updated by using a simulated annealing method, specifically, the individuals X are randomly updated according to step length and normally distributed random numbers 1 '＝X 1 +ρS 0 Wherein X is 1 And X 1 ' represents a pre-update individual and an update individual, respectively; ρ represents a normal scoreA random number of the cloth, rho epsilon (-1,1); s. the 0 Representing a preset step vector.
CN202211382714.4A 2022-11-07 2022-11-07 Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method Pending CN115600503A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211382714.4A CN115600503A (en) 2022-11-07 2022-11-07 Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211382714.4A CN115600503A (en) 2022-11-07 2022-11-07 Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method

Publications (1)

Publication Number Publication Date
CN115600503A true CN115600503A (en) 2023-01-13

Family

ID=84853293

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211382714.4A Pending CN115600503A (en) 2022-11-07 2022-11-07 Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method

Country Status (1)

Country Link
CN (1) CN115600503A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115828818A (en) * 2023-02-02 2023-03-21 湖北工业大学 Photovoltaic cell parameter identification method and storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115828818A (en) * 2023-02-02 2023-03-21 湖北工业大学 Photovoltaic cell parameter identification method and storage medium

Similar Documents

Publication Publication Date Title
Zhong et al. Prediction of photovoltaic power generation based on general regression and back propagation neural network
Kurtz et al. A comparison of theoretical efficiencies of multi‐junction concentrator solar cells
Zhang et al. Modeling and prediction of PV module performance under different operating conditions based on power-law I–V model
CN111245365B (en) Photovoltaic module fault diagnosis method based on dynamic weighted depth forest
CN108694276B (en) Method for calculating output characteristics of series-parallel photovoltaic modules
CN111815021B (en) Photovoltaic power prediction method based on solar radiation climate characteristic identification
Khan et al. A review of estimating solar photovoltaic cell parameters
CN113437938B (en) Photovoltaic array output power calculation method considering field characteristics due to regional differences
Boukortt et al. Development of high-efficiency PERC solar cells using Atlas Silvaco
CN115600503A (en) Multi-junction compound battery model parameter optimization method based on simplex-simulated annealing method
CN110909310A (en) Photovoltaic short-term power generation capacity prediction method and system based on model parameter optimization
CN112465271A (en) Energy storage battery model selection method for energy storage stabilizing wind power fluctuation scene
Lang et al. Parameter identification and performance estimation for PV modules based on reduced forms model
Leem et al. Theoretical modeling and optimization of III–V GaInP/GaAs/Ge monolithic triple-junction solar cells
CN115907339A (en) Power distribution network photovoltaic and electric vehicle charging station collaborative planning method based on GMM
Figueiredo et al. Comparison between P&O-based and PSO-based MPPT algorithms for photovoltaic system under partially shaded conditions
Su et al. Parameter extraction of photovoltaic single-diode model using integrated current–voltage error criterion
Hou et al. Research on the MPPT of Photovoltaic Power Generation Based on the CSA-INC Algorithm
Shannan et al. Two diode model for parameters extraction of PV module
CN109388845B (en) Photovoltaic array parameter extraction method based on reverse learning and enhanced complex evolution
Chin et al. Modifications to Accelerate the iterative algorithm for the two-diode model of PV module
Kebir et al. An overview of solar cells parameters extraction methods
CN109558632A (en) A kind of photovoltaic module parameter identification method
CN109494747B (en) Power grid probability load flow calculation method based on alternating gradient algorithm
CN115034084A (en) Parameter optimization method of multi-junction compound battery model for improving photoelectric conversion efficiency

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