Invention content
Goal of the invention is to establish a kind of photovoltaic module shade determination method based on peak counting and parameter identification, can
The shadow occlusion situation of photovoltaic module is effectively judged.
Technical solution is used by the present invention solves above-mentioned technical problem:
A kind of photovoltaic module shade determination method based on peak counting and parameter identification, wherein the foundation of determination method
Include the following steps:
Step 10:The overall situation is carried out using programmable DC electronic load to photovoltaic module I-V output characteristic curves quickly to sweep
It retouches, to record apparent multi-peak number;
Step 20:It is internal equivalent to sampled value progress in photovoltaic module I-V characteristic curve using artificial fish-swarm algorithm is improved
Parameter identification, in conjunction with slight shadow occlusion situation in root-mean-square error and then determination component.
To optimize above-mentioned technical proposal, the concrete measure taken further includes:
The counting process of the apparent multi-peak point of photovoltaic module is as follows in step 10:
It is scanned since the short circuiting work point of photovoltaic module using programmable DC electronic load, initializes PmPeak counting
Flag=0, if detecting, the output power of photovoltaic module meets:Pk>Pk-1And Pk>Pk+1When, then remember Flag=Flag+1, until
Complete the scanning process of whole photovoltaic module I-V curve, wherein Pk-1、Pk、Pk+1For the output power value of continuous photovoltaic module
Scanning element.
The decision process of the slight shadow occlusion situation of photovoltaic module is as follows in step 20:
It is root-mean-square error RMSE that photovoltaic module parameter identification, which chooses object function, and formula is:
In formula, θ=(Rs、Rsh、Iph、ISD, n) be parameter to be identified, fi(V, I, θ) is i-th group of measured value and simulation model
The difference of output, Rs、RshFor equivalent series and parallel compensated resistance, IphFor photogenerated current, ISDFor diode reverse saturation current, n is two poles
Quality factor is thought in management;
It is as follows using the specific implementation step for improving artificial fish-swarm algorithm for above-mentioned formula (1):
Step1, initialization operation, setup parameter are carried out to parameter:Population invariable number N, random initial position, greatest iteration time
Number Maxgen, sensing range [Visualstart,Visualend], step-length range [Stepstart,Stepend], crowding factor delta,
Maximum sounds out number Try_number and NM method space-number K;
Step2, the fitness value for seeking each Artificial Fish, and record global optimum's Artificial Fish state;
Step3, artificial fish-swarm algorithm parameter is adaptively adjusted;
Step4, the behavior of each Artificial Fish is evaluated, the most suitable behavior of Artificial Fish is selected to be acted;
After Step5, the corresponding behavior of execution, location information and global optimum's Artificial Fish state to Artificial Fish carry out more
Newly, optimal value is assigned to billboard;Meanwhile using reproductive behavior, eliminating the poor individual of fitness value;
Step6, migratory behaviour judge, probability P is migrated if meetinge, then migratory behaviour is executed, and update billboard state;
Otherwise, Step7 execution is passed directly to;
If Step7, meeting t modK=0, wherein t is current iteration number, executes NM methods;By K- mean clusters
Method determines cluster centre Artificial Fish;NM search is executed to each class center individual, its fitness value is calculated and updates billboard;
Step8, the search of NM methods is executed to global extremum Artificial Fish individual, optimal value is assigned to billboard;
Step9, judge end condition, if meeting end condition, export optimal value, algorithm terminates;Otherwise, continue iteration
Step2~Step8 is executed, until algorithm end condition is satisfied.
Principle:The present invention is based on the photovoltaic module shade determination method of peak counting and parameter identification, purpose has been intended to
Shadow occlusion situation in effect judgement photovoltaic module.
Advantageous effect:Establish a kind of photovoltaic module shade determination method based on peak counting and parameter identification, this hair
Bright the case where being difficult to judge for shade in photovoltaic module, using above-mentioned two-step method, shadow occlusion in photovoltaic module can be carried out
Effectively judgement.
Specific implementation mode
Below in conjunction with the accompanying drawings, the present invention is described in detail.
Fig. 1 show the photovoltaic module characteristic curved scanning circuit block diagram based on programmable DC electronic load.As schemed
Show, using programmable DC electronic load as the load of photovoltaic module, the equivalent resistance of programmable electronic load is by control circuit
Infinity is changed stepwise by zero in the control of output signal, the equivalent change in resistance range, and photovoltaic module operating point is also by short circuit
Open circuit point is changed stepwise in point.In the process, output voltage, electric current on the continuous work of photovoltaic module point are sampled,
Just photovoltaic module I-V characteristic curve under current working is obtained.
Programmable DC electronic load selects MOSFET in the present invention, permanent in photovoltaic module I-V output characteristic curves to make
It is more accurate to flow source region scanning, often takes multiple MOSFET in parallel, with reduce its it is fully on when equivalent resistance.By right
The control to DC Electronic Loads can be realized in the control of MOSFET driving voltages, based on programmable DC electronic load to I-V
Characteristic curve is scanned achievable scanning process and controllably changes.More conventional capacitance dynamic charging I-V characteristic curved scanning method, should
Method has many advantages, such as that small, at low cost, precision is high, scanning process is controllable.
The present invention is that hardware circuit realizes simple, for whole photovoltaic module I-V characteristic curve by short circuit current point,
Setting electronic load is in constant pressure operating mode, controls photovoltaic module output voltage with fixed step size, and synchronized sampling component is defeated
Go out electric current and voltage, until photovoltaic module is in open-circuit condition, completes whole I-V characteristic curved scanning.
Normal conditions, multi-peak is presented in photovoltaic module output characteristics under the conditions of local shades.Using programmable DC electronics
Load is scanned since the short circuiting work point of photovoltaic module, initializes PmPeak counting Flag=0, if detecting photovoltaic module
Output power meets:Pk>Pk-1And Pk>Pk+1When, then remember Flag=Flag+1, under shadowed condition the counting of photovoltaic module multi-peak show
It is intended to as shown in Figure 2.
As shown in Figure 2, the accuracy of photovoltaic module multi-peak point counting method is easily influenced by two aspect reasons under shadowed condition:
The sweep spacing of one side programmable DC electronic load, sweep spacing is shorter, sampled point is more, precision is higher, but can cause
Scanning circuit takes longer;On the other hand it is shadow occlusion degree, shade degree is lower, curvilinear motion more unobvious.
Based on above-mentioned two aspects reason, suitable I-V characteristic curved scanning interval is chosen, when the apparent multi-peak feelings of appearance
When condition, peak-valley counting method can be screened effectively, but still inevitably will appear misjudgment phenomenon.
In general, the equivalent-circuit model of photovoltaic cell monomer is as shown in figure 3, model parameter includes mainly:Photogenerated current
(Iph), diode reverse saturation current (ISD), diode desired qualities factor (n) and equivalent series and parallel compensated resistance (Rs、Rsh)。
In practical applications, one piece of photovoltaic module is by NsA photovoltaic cell is connected, then the kirchhoff of photovoltaic module
Current law (Kirchhoff ' s current law, KCL) Equivalent circuit equations are
In formula:Q is electronic charge (1.602 × 10-19C);K is Boltzmann constant (1.381 × 10-23J/K);T is
Thermodynamic temperature (room temperature is approximately 300K).
By formula (1) it is found that containing 5 unknown parameters in the equation, respectively:Iph、ISD、n、RsAnd Rsh, and manufacturer
Databook above-mentioned parameter value will not be generally provided.
Photovoltaic module mathematical model is an implicit and nonlinear transcendental equation, and direct function solves more difficult.This hair
The bright Lambert W functions that first pass through are carried out explicitization processing, for reducing the difficulty of parametric solution in former mathematical model;Into
And it proposes by improving artificial fish-swarm algorithm (IAFSA) come identification model parameter.
It can thus be concluded that the electric current explicit expression of photovoltaic module:
In formula,
Before carrying out IAFSA and recognizing the model parameter, accurate object function need to be established, following deformation is made to formula (2):
In formula, V, I are voltage, current sampling data in I-V characteristic curve.
Photovoltaic module parameter identification can be summarized as an optimization problem, basic thought be by minimize object function with
Seek parameter optimal value.The object function that the present invention chooses is that root-mean-square error (RMSE) is:
In formula, θ=(Rs、Rsh、Iph、ISD, n) be parameter to be identified, fi(V, I, θ) is i-th group of measured value and simulation model
The difference of output.
The target of photovoltaic module parameter identification is to solve min RMSE, and Fig. 4 is photovoltaic module parameter identification strategy, wherein
S is intensity of illumination.
Artificial fish-swarm algorithm simulates the cluster foraging behavior of fish in nature, uses optimizing pattern from bottom to top, leads to
Cross the purpose that the cooperation in the shoal of fish between individual makes group be optimal selection.Every Artificial Fish explores the ring itself being presently in
Border, selection execute one such behavior operator, and by constantly adjusting the position of individual, it is larger finally to concentrate at food density
Around region, global optimum is obtained.Foraging behavior establishes the basis of algorithmic statement, and behavior of bunching enhances the stabilization of algorithmic statement
Property and it is of overall importance, behavior of knocking into the back enhances the rapidity of algorithmic statement and of overall importance.Artificial Fish is by the perception to environment come autonomous
Coordinate search mechanisms, which finally can be near optimizing to global optimum, to make optimization problem solving.
Nelder-Mead methods (NM methods) are also referred to as simplex method, are different from the simplex method of linear programming, it is applicable in
In seeking n-ary function f (x1,x2,…,xn) without constraint minimum value.Its algorithm idea be in n-dimensional space, can by n+1 vertex
To form the figure of " most simple ", simplex is.NM methods are exactly first to build a simplex that is initial, covering set point, so
In each step searched for afterwards, generates and compare from current simplex using possible 4 kinds of modes (reflection expands, compresses and shrinks)
Close point can compare with the value on each vertex of simplex in new point superior function value, can generally be substituted, be produced there are one vertex
A raw new simplex, repeats step as above, until the functional value of simplex is less than predetermined threshold value.
It is similar with other intelligent optimization algorithms, when being in random movement state there are Artificial Fish in AFSA or in local pole
Value point occur artificial fish-swarm aggregation it is serious when, cause algorithm the convergence speed to slow down, and then influence final convergence precision.
Therefore, the problems such as late convergence slows down, precision reduces, the dynamic in algorithm operational process are run for AFSA
Relevant parameter is adjusted, while introducing reproductive behavior, migratory behaviour and NM methods to improve the whole optimizing performance of algorithm, preferable Horizon
The global and local search capability of weighing apparatus innovatory algorithm, further speeds up arithmetic speed.
Algorithm iteration runs early period, and larger Visual and Step can enhance the ability of searching optimum and convergence speed of algorithm
Degree;Iteration runs the later stage, and algorithm is gradually evolved into fining search process, fine search is carried out within the scope of optimal solution neighborhood.
Based on this, Artificial Fish sensing range Visual and moving step length Step can be adjusted into Mobile state by formula (5):
In formula, Visualstart、VisualendThe initial value and final value of Visual are indicated respectively;Stepstart、StependRespectively
Indicate the initial value and final value of Step;T is current iteration number, and Maxgen is maximum iteration.
In the iterative process of IAFSA, K- means clustering methods are introduced in fixed iteration interval step number to artificial fish-swarm
Classify, and NM method precise search is executed to cluster centre individual.In addition, to accelerate the convergence rate reconciliation of algorithm entirety
Quality, the search of NM method is performed both by global extremum point in each iterative process billboard.Based on this, IAFSA is preferably
Using the optimum results obtained by AFSA, while appropriateness reduces NM method calculation amounts, and the particular flow sheet of IAFSA is as shown in Figure 5.
In conclusion the specific implementation step of the carried IAFSA of the present invention is as follows:
Step1, to parameter carry out initialization operation, population invariable number N, random initial position, maximum iteration Maxgen,
Sensing range [Visualstart,Visualend], step-length range [Stepstart,Stepend], crowding factor delta, maximum sound out time
The parameters such as number Try_number and NM method space-numbers K.
Step2, the fitness value for seeking each Artificial Fish, and record global optimum's Artificial Fish state.
Step3, AFSA algorithm parameters are adaptively adjusted.
Step4, the behavior of each Artificial Fish is evaluated, the most suitable behavior of Artificial Fish is selected to be acted.
After Step5, the corresponding behavior of execution, location information and global optimum's Artificial Fish state to Artificial Fish carry out more
Newly, optimal value is assigned to billboard.Meanwhile using reproductive behavior, eliminating the poor individual of fitness value.
Step6, migratory behaviour judge, probability P is migrated if meetinge, then migratory behaviour is executed, and update billboard state;
Otherwise, Step7 execution is passed directly to.
If Step7, meeting tmodK=0, NM methods are executed.By K- means Methods, cluster centre Artificial Fish is determined;
NM search is executed to each class center individual, its fitness value is calculated and updates billboard.
Step8, the search of NM methods is executed to global extremum Artificial Fish individual, optimal value is assigned to billboard.
Step9, judge end condition, if meeting end condition, export optimal value, algorithm terminates;Otherwise, continue iteration
Step2~Step8 is executed, until algorithm end condition is satisfied.
Using TSM-250PC05A types photovoltaic module as research object, using experiment porch to the component I- under different operating modes
V output characteristic curves are scanned.To simplify data volume, 32 point samplings only are carried out to every I-V characteristic curve of photovoltaic module,
And it is updated in IAFSA and carries out parameter identification.
To verify the accuracy and rapidity of IAFSA identified parameters, to photovoltaic module in intensity of illumination 950W/m2, environment
Newton methods are respectively adopted in 25 DEG C of next group of measurement data of temperature, GA, ABSO and IAFSA carry out parameter identification and compare reality
It tests, equally using iterations 100 times as end condition, IAFSA parameter settings are as shown in table 1.Table 2 provides each algorithm and transports respectively
Optimal identified parameters value after row 20 times, as seen from table, the object function RMSE value obtained using IAFSA are minimum, show IAFSA
With stronger ability of searching optimum and local mining ability, can obtain closest to actual model parameter value.
1 IAFSA parameter settings of table
Wherein, reflectance factor α (α>0), compressed coefficient β (0<β<1), lengthening coefficient γ (γ>1), constriction coefficient λ (0<λ<
1);TknmIt is the iterations that cluster centre point carries out NM search;TgnmIt is the iterations that global extremum point carries out NM search;K
For number of classifying, TkFor the iterations of K- means Methods.
Photovoltaic module parameter identification object function RMSE value under the different identification algorithms of table 2
In addition, for further illustrate IAFSA validity, to every I-V characteristic curve under photovoltaic module difference operating mode into
32 point sampling of row, and be updated in IAFSA and carry out identification of Model Parameters, choose object function RMSE under wherein 4 kinds of operating modes
Value, as shown in table 3.
Photovoltaic module parameter identification object function RMSE value under the different operating modes of table 3
As shown in Table 3, it is based on IAFSA and optimizes the obtained equal very little of object function RMSE value, illustrate that above-mentioned parameter recognizes
As a result very close with the actual parameter value of component.
In general, traditional photovoltaic module inside equivalent parameters formula is based on the description under single peak output characteristics, it cannot
Accurate description is carried out to output characteristics under component shadow occlusion, therefore can judge that photovoltaic module is hidden by shade in conjunction with RMSE value
Gear.The present invention keeps maximum power value variation within 1W, RMSE value is 0.01 by repeatedly slight shadow occlusion experiment
Compared with fluctuating in small neighbourhood, therefore propose to use RMSE>0.01 as the foundation for judging slight shadow occlusion.
In conclusion by parameter identification and based on programmable DC electronic load to I-V output characteristic curve peak countings
Flag values are combined, and can must judge particular flow sheet that whether photovoltaic module is blocked by shadow as indicated with 6.
It will be appreciated from fig. 6 that being scanned to I-V output characteristic curves by programmable DC electronic load and recording peak value
Number, if Flag>1, then show V-P curves there are multi-peak, photovoltaic module is blocked by shadow;Otherwise, it is further calculated by IAFSA
Whether method carries out parameter identification to I-V characteristic curve sampled value, hidden by shade with the RMSE value judgement photovoltaic module of object function
Gear.Accurate judgement can be made by the above method to whether photovoltaic module is blocked by shadow.
When photovoltaic module is under shadow occlusion, multi-peak is presented in I-V characteristic curve.Similarly, it is in intensity of illumination
950W/m2, 25 DEG C of environment temperature when, it is assumed that 20 photovoltaic electrics in the same administrative branch of bypass diode in TSM-250PC05A types
Pond is all blocked by shadow, and is acquired to photovoltaic module operation data, using programmable DC electronic load to above-mentioned I-V spies
Linearity curve is scanned, and can obtain peak counting Flag=2.At this point, still with the object function of formula (4) algorithm as an optimization, respectively
Parameter identification is carried out using Newton methods, GA, ABSO and IAFSA and compares experiment, wherein each parameter setting in above-mentioned algorithm,
Iterations with it is consistent above.Table 4 provides each algorithm and is in RMSE value under shadow occlusion to photovoltaic module.
When 4 shadow occlusion of table, photovoltaic module parameter identification object function RMSE value under different identification algorithms
As shown in Table 4, I-V characteristic curve carries out parameter when using above-mentioned 4 kinds of distinct methods to photovoltaic module shadow occlusion
Identification, the RMSE value of gained object function is larger, parameter optimisation procedure when compared to photovoltaic module shadow-free, even if algorithm
Iterations and test number (TN) further increase, but still cannot photovoltaic module at this time optimal internal equivalent parameters value, show
Formula (4) cannot characterize output characteristics under photovoltaic module shade.Parameter identification result based on above-mentioned 4 kinds of methods establishes photovoltaic group
Part model, to obtain different simulation result outputs, as shown in Figure 7.
As shown in Figure 7, it is unimodal curve based on model output obtained by above-mentioned 4 kinds of parameter identification results, with photovoltaic module
Multimodal output characteristic curve is significantly different under shade.Meanwhile compared to other several methods, based on the imitative of model obtained by IAFSA
True curve of output highest opposite with measured curve fitting degree.
Further to study accuracys of the IAFSA to parameter identification result under photovoltaic module shade, keep intensity of illumination and
Environment temperature is constant, chooses measured value under 3 kinds of photovoltaic module shades and carries out parameter identification, respectively:(1) in photovoltaic module only
1 photovoltaic cell is blocked by shadow;(2) there are 2 photovoltaic cells by different degrees of shadow occlusion in photovoltaic module, and be distributed in
In 2 bypass diode branches;(3) there are 3 photovoltaic cells by the shadow occlusion of same degree in photovoltaic module, and be distributed in 3
In a bypass diode branch.Using IAFSA, to photovoltaic module, the RMSE value under different shades is as shown in table 5.
When 5 difference shadow occlusion of table, photovoltaic module parameter identification object function RMSE value
Simulation model output is established based on above-mentioned parameter identification result to be compared with measured value.
By table 5 and Fig. 8 it is found that when the shielding rate for being blocked by shadow part in photovoltaic module is smaller or every branch top shadow
When coverage extent is identical, multimodality unobvious of V-P output characteristic curves or to be unimodal, as shown in Fig. 8 (a), (c) and (e),
If carrying out peak counting to I-V output characteristic curves only in accordance with programmable DC electronic load at this time to be difficult to determine that it is hidden by shade
Situation is kept off, causes to judge by accident.Parameter is carried out using IAFSA to Fig. 8 (b), (d) and photovoltaic module I-V output characteristic curves in (f) to distinguish
Know, shadow condition can be judged;But gained partial parameters identification result has lost practical significance, is only under curve matching
Optimal value of the parameter.
The above is only the preferred embodiment of the present invention, protection scope of the present invention is not limited merely to above-described embodiment,
All technical solutions belonged under thinking of the present invention all belong to the scope of protection of the present invention.It should be pointed out that for the art
For those of ordinary skill, several improvements and modifications without departing from the principles of the present invention should be regarded as the protection of the present invention
Range.