TWI776199B - Parameter estimation method for solar cell double-diode model - Google Patents

Abstract

The invention provides parameter estimation method for solar cell double-diode model, which eliminates parameters with insignificant effect through sensitivity matrix, determines the effect of each parameter on the overall output measurement by principal component analysis, establishes an orthogonal eigenvalue between input variables to remove the dependency between parameters through applying Gram-Schmidt orthogonalization, completes the selection of parameters affecting the output, the Whale Optimization Algorithm is used to estimate the relevant parameters to obtain the best parameter solution. In the process of estimating the Whale Optimization Algorithm, a double-diode-based PV (solar cell) estimation model must be built in a Matlab/Simulink environment, and the parameter values must be corrected according to the input and output data obtained from the actual measurement, which enables the Whale Optimization Algorithm to generate the best parameter solution and stores it in the PV estimation model. Accordingly, by inputting the real-time illuminance and temperature into the PV estimation model, the corresponding and accurate power generation forecast value can be generated.

Description

Parameter Estimation Method of Solar Cell Double Diode Model

The present invention relates to a method for estimating parameters of a solar cell bidiode model, in particular to a method for estimating parameters for obtaining an optimal estimated parameter solution to accurately estimate the power generated by a solar cell.

Although solar energy has the advantages of being inexhaustible, inexhaustible, and does not require transportation and does not emit substances that adversely affect the environment, it is a clean energy source and can be used as a permanent energy source for human beings. The influence of climate, day and night is great and cannot be constant. The fact that it cannot be constant will seriously affect the power quality and the stability of the power system, thereby hindering the economic development of society and even the entire country.

Therefore, with the rapid development of the smart grid, the power estimation of the solar power generation system plays an increasingly important role. Through accurate power generation efficiency estimation, it can not only provide the smart grid power management system for efficient load scheduling, but also reduce wisdom The uncertainty of solar power generation on the grid, thereby improving system stability, maintaining power quality, and even improving the dispatchability of solar power generation systems.

The physical models of solar power generation system electricity estimation are roughly divided into single-diode model and double-diode model. Under the compromise of model accuracy and simplification, single-diode model is adopted by most people. However, since the single-diode model ignores the recombination loss of the diode in the depletion region, it is easy to produce a large estimation error under low illumination. Unlike the single-diode model, the saturation current in the dual-diode model is contributed by two diodes. This model produces accurate predictions at low light levels, but the One diode is added to the single-diode model, and the parameters of the solar cell power generation model may be offset due to the aging of the solar cell or the change of the climate. Therefore, the parameters must be estimated and corrected in a timely and real-time manner. Accurately estimating the power generation efficiency of the PV power generation model makes it more complicated to estimate the power generation efficiency of the PV power generation model using dual diodes.

The main purpose of the present invention is to provide a method for estimating parameters of a bi-diode model of a solar cell, which converts the 7 parameters to be estimated in the bi-diode model into 17 parameters under different operating points, and then transmits them through the solar cell. The establishment of the power generation model, the selection of parameters, and the optimization of parameter estimation and other processes provide more refined and accurate power generation efficiency estimates for the smart grid power management system to carry out effective power management.

The object of the present invention is realized by the following technologies: A method for estimating parameters of a dual-diode model of a solar cell, the steps of which include: Step 1: Converting complex parameters that will change under different illumination and temperature to a sensitivity matrix at a steady-state operating point into the complex parameters Those with little effect on the output are eliminated, and the remaining parameters are determined by principal component analysis to determine the effect of each remaining parameter relative to the overall output measurement, and then the Gram-Schmidt orthogonal transformation is used to establish the remaining parameters. In order to eliminate the remaining dependencies between the parameters, and then select the parameters that affect the output; Step 2: The parameters that affect the output selected in Step 1 are established in the Matlab/Simulink environment as a double-diode Based on the PV estimation model, the whale optimization algorithm is applied to estimate the parameters that affect the output, so as to obtain the best parameter solution; wherein, the whale optimization algorithm includes: Step 2 -1: Take the parameter that affects the output selected in step 1 as the initial parameter, and randomly give each initial parameter an initial parameter solution position vector, and set the number of iterations at the same time; step 2-2: use the actual The measured input value and output value data are sent into the PV estimation model to calculate the estimated value, and then the estimated value is integrated with the actual measured input value and output value, and then each calculated value is obtained. The fitness value of the initial parameter solution position vector; Step 2-3: According to the encircling prey method in the bubble net attack method, in the process of reducing the mechanism of encircling the best feasible solution, first assume that the current best feasible solution is said For the best position of the initial parametric solution, use the spiral path of the helical equation to move towards the direction of the best position of the initial parametric solution, and update the position of the initial parametric solution according to the best position; Step 2-4: use The method of searching for prey explores and randomly selects parametric solutions, and the parametric solutions in these searches update their positions according to random positions, and then move towards the direction of the best solution in the whole domain; Steps 2-5: Decode the parameters after updating the positions Enter the PV estimation model to calculate the estimated power value, then integrate the estimated power value with the actual measured input value and output value, and then calculate the adaptive value of the parameter solution position vector after each updated position, and Update the best position solution; Step 2-6: Check whether the set number of iterations is met, if not, go back to step 2-3; if so, the updated best position solution is the best parameter solution , output the optimal parameter solution and end.

、

、β及E _g,ref 。 In the method for estimating parameters of a solar cell dual-diode model as described above, the parameters that may vary under different illumination and temperature in step 1 include: _IL,ref , I _sc,ref , V _oc,ref , I _mp,ref , V _mp,ref , G _ref , T _ref , n _{I 1,ref} , n _{I 2 ,ref} , R _s,ref , R _sh,ref , I _{o 1 ,ref} , I _{o 2 ,ref} , α

,

, β and E _g,ref .

，該擾動將導致第i個輸出產生變化，即

，則靈敏度係數可近似如下：

上式可建構出每一個參數相對於每一個輸出的靈敏度矩陣，對輸出影響微小的參數可藉此進行第一步的剔除。 In the above-mentioned method for estimating the parameters of the solar cell dual-diode model, in step 1, the judgment of the parameters that have little influence on the output is to consider that there is a disturbance on the parameter j , that is,

, the disturbance will cause the ith output to change, that is

, the sensitivity coefficient can be approximated as follows:

The above formula can construct the sensitivity matrix of each parameter relative to each output, and parameters that have little influence on the output can be eliminated in the first step.

的特徵向量(eigenvector)，其中第一個主成份為X矩陣中具有最大特徵值(eigenvalue)所對應的特徵向量，此即整體變化的最大方向，其餘主成份則依序排列其對於整體變化的貢獻度，因此，第j個參數的整體效應表示如下：

其中

為第j個參數相對於整體變數的效應，其值愈大愈佳，α _i 為第i個輸出的特徵值，C _ij 為第j個參數相對於第i個輸出的主成份元素(或貢獻度)，m為輸出的個數。 The above-mentioned method for estimating parameters of a solar cell dual-diode model, wherein, in the principal component analysis method described in step 1, the principal component is a covariance matrix

The eigenvector of , where the first principal component is the eigenvector corresponding to the largest eigenvalue (eigenvalue) in the X matrix, which is the maximum direction of the overall change, and the remaining principal components are arranged in order with respect to the overall change. Contribution, therefore, the overall effect of the jth parameter is expressed as:

in

is the effect of the jth parameter relative to the overall variable, the larger the value, the better, α _i is the eigenvalue of the ith output, C _ij is the principal component element (or contribution of the jth parameter relative to the ith output) degrees), m is the number of outputs.

特徵矩陣X可分解為：X=QR，其中R為一上三角矩陣(upper triangular matrix)，Q為一正交矩陣(orthogonal matrix)，即

其中q _i 為正交空間新的特徵向量，在使用Gram-Schmidt進行正交分解(orthogonal decomposition)過程中，下列程序用來建構正交矩陣：q ₁=x ₁，

其中，

將上述X=QR公式中的X空間映射至空間Q：Q=R ^-1 X，以建構出所述正交矩陣。 The above-mentioned method for estimating parameters of a solar cell dual-diode model, wherein, in step 1 to eliminate the dependence between parameters and parameters, it is assumed that X _i =[ X _i (1), X _i (2),..., X _i ( N )] ^T is the feature vector of the ith sample, and N is the number of samples, then the feature matrix is defined as follows:

The characteristic matrix X can be decomposed into: X = QR , where R is an upper triangular matrix, and Q is an orthogonal matrix, that is,

where q _i is the new eigenvector of the orthogonal space. In the orthogonal decomposition process using Gram-Schmidt, the following procedure is used to construct the orthogonal matrix: q ₁ = x ₁ ,

in,

The X space in the above formula of X = QR is mapped to the space Q : Q = R ^-1 X to construct the orthogonal matrix.

(t)=[經所述步驟1後所有被選取的參數](t為疊代次數)。 In the method for estimating parameters of the solar cell dual diode model as described above, the position vector of each initial parameter solution is randomly generated in step 2-1, as shown below: x _i,j (0)= x _i,min + rand ×( x _i,max - x _i,min ), i =1,2,..., S , j =1,2,..., P ; where X _i,j (0) is the jth The i -th parameter (variable) is solved by a parameter. X _i,min and X _i,max are the upper and lower limits of the parameters. The upper and lower limits of the parameters are set by user experience. rand is a uniform random number between 0 and 1. S is the number of parameters, P is the number of parameter solutions, and the position vector of the jth parameter solution can be expressed as follows:

( t )=[all selected parameters after step 1] (t is the number of iterations).

In the above-mentioned method for estimating parameters of a dual-diode model of a solar cell, the input data in step 2-2 are illuminance and temperature.

In the above-mentioned method for estimating parameters of a dual-diode model of a solar cell, the output data in step 2-2 are voltage, current, and power.

其中P _j,est 為PV發電模型估測值，P _j,mea 為實際量測值。 In the above-mentioned method for estimating the parameters of the dual-diode model of a solar cell, the fitness value of each initial parameter solution position vector is calculated as follows:

Among them, P _j,est is the estimated value of the PV power generation model, and P _j,mea is the actual measured value.

Figure 1: Physical model of the dual-diode model of the present invention

The second figure: the flow chart of the method for estimating the parameters of the solar cell bidiode model of the present invention

Figure 3: Flow chart of the whale optimization algorithm of the present invention

Figure 4: Actual measured values of sunny weather patterns and estimated results before and after parameter optimization

Figure 5: The actual measured value of the rainy weather pattern and the estimated results before and after parameter optimization

Figure 6: The actual measured values of cloudy weather patterns and the estimated results before and after parameter optimization

Figure 7: The actual measured values of cloudy weather patterns in sunny days and the estimated results before and after parameter optimization

Figure 8: The actual measured value of the weather pattern on a sunny day and the parameter estimation comparison between the double-diode model and the single-diode model

Figure 9: The actual measured values in cloudy weather patterns and the parameter estimation comparison between the double-diode model and the single-diode model

Figure 10: The actual measured values in cloudy weather patterns and the parameter estimation comparison between the dual-diode model and the single-diode model

Figure 11: Actual measured values of cloudy weather patterns in clear weather and comparison of parameter estimates between the dual-diode model and the single-diode model

In order to have a more complete and clear disclosure of the technical content, the purpose of the invention and the effect achieved by the present invention, it is explained in detail below, and please refer to the disclosed drawings and drawing numbers: please refer to the first ~Three pictures.

The method for estimating parameters of a solar cell dual-diode model of the present invention uses the bi-diode model circuit shown in the first figure as the physical model of the solar cell of the present invention. The steps of the estimation method include:

最大功率點下溫度係數)、

短路電流下溫度係數)、β(開路電壓下溫度係數)及E _g,ref (間隙能量)。 Step 1: First, the parameters that will change under different illuminance and temperature are eliminated by the sensitivity matrix at a steady-state operating point, and the parameters that have little influence on the output are then determined by the principal component analysis method. The effect of the parameters relative to the overall output measurement, and then applying the Gram-Schmidt orthogonal transformation to establish the orthogonal features between the remaining parameters, so as to eliminate the dependencies between the remaining parameters, and then select the parameters that affect the output; wherein , the parameters that will change under different illumination and temperature include: I _L,ref (photocurrent), I _sc,ref (short-circuit current), V _oc,ref (open circuit voltage), I _mp,ref (maximum power point current), V _mp,ref (maximum power point voltage), G _ref (illuminance), T _ref (module temperature), n _{I 1,ref} (first diode ideality factor), n _{I 2 ,ref} ( second diode ideality factor), R _s,ref (series resistance), R _sh,ref (parallel resistance), I _{o 1 ,ref} (first diode saturation current), I _{o 2 ,ref} ( second diode saturation current),

temperature coefficient at the maximum power point),

temperature coefficient at short-circuit current), β (temperature coefficient at open circuit voltage) and E _g,ref (gap energy).

，該擾動將導致第i個輸出產生變化，即

，則靈敏度係數可近似如下：

上式可建構出每一個參數相對於每一個輸出的靈敏度矩陣，對輸出影響微小的參數可藉此進行第一步的剔除。 In step 1, in the judgment of the parameters that have little influence on the output, it is better to consider that there is a disturbance on the parameter j , that is,

, the disturbance will cause the ith output to change, that is

, the sensitivity coefficient can be approximated as follows:

The above formula can construct the sensitivity matrix of each parameter relative to each output, and parameters that have little influence on the output can be eliminated in the first step.

的特徵向量(eigenvector)，其中第一個主成份為X矩陣中具有最大特徵值(eigenvalue)所對應的特徵向量，此即整體變化的最大方向，其餘主成份則依序排列其對於整體變化的貢獻度，因此，第j個參數的整體效應可表示如下：

其中

為第j個參數相對於整體變數的效應，其值愈大愈佳，α _i 為第i個輸出的特徵值，C _ij 為第j個參數相對於第i個輸出的主成份元素(或貢獻度)，m為輸出的個數。 Wherein, in the principal component analysis method in step 1, it is preferable to make the principal component a covariance matrix

The eigenvector of , where the first principal component is the eigenvector corresponding to the largest eigenvalue (eigenvalue) in the X matrix, which is the maximum direction of the overall change, and the remaining principal components are arranged in order with respect to the overall change. contribution, so the overall effect of the jth parameter can be expressed as:

in

is the effect of the jth parameter relative to the overall variable, the larger the value, the better, α _i is the eigenvalue of the ith output, C _ij is the principal component element (or contribution of the jth parameter relative to the ith output) degrees), m is the number of outputs.

特徵矩陣X可分解為：X=QR，其中R為一上三角矩陣(upper triangular matrix)，Q為一正交矩陣(orthogonal matrix)，即

其中q _i 為正交空間新的特徵向量，在使用Gram-Schmidt進行正交分解(orthogonal decomposition)過程中，下列程序用來建構正交矩陣： q ₁=x ₁，

其中，

將上述X=QR公式中的X空間映射至空間Q：Q=R ^-1 X，以建構出所述正交矩陣。 Among them, in the dependency relationship between the remaining parameters and the parameters in step 1, it is assumed that X _i =[ X _i (1), X _i (2),..., X _i ( N )] ^T is the i -th sample The eigenvector, N is the number of samples, the eigenmatrix is defined as follows:

The characteristic matrix X can be decomposed into: X = QR , where R is an upper triangular matrix, and Q is an orthogonal matrix, that is,

where q _i is the new eigenvector of the orthogonal space. In the orthogonal decomposition process using Gram-Schmidt, the following procedure is used to construct the orthogonal matrix: q ₁ = x ₁ ,

in,

The X space in the above formula of X = QR is mapped to the space Q : Q = R ^-1 X to construct the orthogonal matrix.

Step 2: The parameters that affect the output selected in Step 1 are established in the Matlab/Simulink environment to establish a PV (solar cell) estimation model based on the bi-diode, and the whale optimization algorithm is applied to perform these parameters. The estimation of the parameters that affect the output to generate the best estimated parameter solution; wherein, the whale optimization algorithm comprises:

Step 2-1: First, take the parameters (control variables) that affect the output selected in step 1 as the initial parameters, and give the initial parameter solution position vector, and set the number of iterations at the same time, each initial parameter solution position vector is randomly generated. , as follows: x _i,j (0)= x _i,min + rand ×( x _i,max - x _i,min ); i =1,2,..., S ; j =1,2 ,..., P ; where X _i,j (0) is the j -th initial parameter solution of the i -th initial parameter (variable), X _i,min and X _i,max are the upper and lower limits of the parameters, and rand is Uniform random number between 0 and 1, S is the number of parameters, P is the number of parameter solutions; the position vector of the jth initial parameter solution can be expressed as follows:

Among them, the number of parameters S in the above formula depends on the selected parameters.

其中P _j,est 為PV發電模型估測值，P _j,mea 為實際量測值，計算後，所得到的適應值fit(j)愈小愈好。 Step 2-2: The actual measured input value (such as illuminance, temperature) and output value (such as voltage, current, power) data are sent into the PV estimation model to calculate the estimated value, and then the estimated value Integrate with the actual measurement value, and then calculate the fitness value of each initial parameter solution position vector; wherein, the formula for calculating the fitness value of each initial parameter solution position vector is as follows:

Among them, P _j,est is the estimated value of the PV power generation model, and P _j,mea is the actual measured value. After calculation, the smaller the fitness value fit ( j ), the better.

Step 2-3: Use the bubble net attack method to update the initial parameter solution position. The bubble net attack method includes the mechanism of shrinking and encircling the prey and updating the position in a spiral manner; it is the mechanism process of shrinking and encircling the best feasible solution according to the method of encircling the prey. , first assume that the current best feasible solution is the target prey (that is, the best position of the initial parameter solution), use the spiral path of the spiral equation to move towards the direction of the best position of the initial parameter solution, and update the initial parameters according to the best position solution location.

Step 2-4: Use the prey search method to explore and randomly select some parametric solutions. The parametric solutions in these searches update their positions according to the random positions, and then move towards the best solution in the whole domain.

Step 2-5: Send the updated optimal parameter solution into the PV estimation model to calculate the estimated power value, and then integrate the estimated power value with the actual measured input value and output value, and then calculate to obtain The fitness value of the parameter solution position vector after each updated position, and the optimal position solution is updated.

Step 2-6: Determine whether the set number of iterations is met, if not, go back to step 2-3; if so, it means that the best parameter solution is found, the end condition is met, and the best parameter solution is output.

After the optimal parameter solution is output through the above steps, as long as the real-time illuminance and temperature data are input into the PV estimation model, a corresponding and accurate predicted value of the generated power can be generated.

<Example>

其中P _fore 為估測值，P _true 為實際值，P _cap 為PV發電容量，N為訓練資料數。 The method proposed by the present invention is applied to a 500kWp solar power generation system. The data is sampled once every minute, and the data obtained within one hour are averaged as the PV power generation for that hour. Due to the relationship of sunshine, the summer moon data is obtained from 06:00 in the morning to 19:00 in the evening at 14:00, and in the non-summer month from 06:00 in the morning to 17:00 in the evening at 12:00. The data collection period was from January 2018 to December 2018. In order to evaluate the accuracy of prediction, the present invention adopts the mean relative error (Mean Relative Error, MRE ), as follows:

Among them, P _fore is the estimated value, P _true is the actual value, P _cap is the PV power generation capacity, and N is the number of training data.

Among them, the first figure is used as the solar cell double-diode model of the present invention, and its parameters that will change under different illumination and temperature include: I _L,ref (photocurrent), I _sc,ref ( short circuit current), V _oc,ref (open circuit voltage), I _mp,ref (maximum power point current), V _mp,ref (maximum power point voltage), G _ref (illuminance), T _ref (module temperature), n _{I 1,ref} (first diode ideality factor), n _{I 2 ,ref} (second diode ideality factor), R _s,ref (series resistance), R _sh,ref (parallel resistance), I _{o 1 ,ref} (first diode saturation current), I _{o 2 ,ref} (second diode saturation current), α _Imp (temperature coefficient at maximum power point), α _Isc (temperature coefficient at short-circuit current) ), β (temperature coefficient at open circuit voltage) and E _g,ref (gap energy). Table 1 shows the above 17 related parameters and their possible variation ranges.

The above 17 parameters that will change under different illumination and temperature are placed at a steady-state operating point, and the parameters that have little influence on the output are eliminated by the sensitivity matrix. The parameters eliminated at this stage are α _Isc and β; The component analysis method determines the effect of the remaining parameters on the overall output measurement after eliminating the small parameters that affect the output. Finally, the Gram-Schmidt orthogonal transformation technique is used to establish the orthogonal features between the remaining parameters to eliminate the remaining parameters. 14 parameters that can affect the output are selected. Among them, the parameters that have not been selected in this step are α _Imp , α _Isc , β, and the other 14 parameters that are selected to affect the output are shown in Table 1. Finally, The selected 14 parameters are: I _L,ref , I _sc,ref , V _oc,ref , I _mp,ref , V _mp,ref , G _ref , T _ref , n _{I 1,ref} , n _{I 2 ,ref} , R _s,ref , R _sh,ref , I _{o 1 ,ref} , I _{o 2 ,ref} , E _g,ref ). Based on these 14 selected parameters, a solar cell power generation estimation model is established in the Matlab/Simulink environment. The 14 parameters to be selected are used as the initial parameters, and the initial parameters are given randomly to solve the position vector, and the number of iterations is set at the same time; then the actual measured illuminance, temperature input value and voltage, current, power output value are sent to PV estimation The estimated value is calculated in the model, and then the estimated value is integrated with the actual measurement value, and then the fitness value of each initial parameter solution position vector is calculated; then, the initial parameter solution position is updated according to the encircling prey method, and the updated After that, the optimal parameter solution is sent into the PV estimation model to calculate the estimated power value, and then the estimated power value is integrated with the actual measured input value and output value, and then the position of the parameter solution after each update position is calculated. The fitness value of the vector, and update the optimal position solution. During the operation, if it meets the set number of iterations, it means that the best parameter solution is found. Steps continue until the set number of iterations is met. When the optimal parameter solution is found, as long as the real-time illuminance and temperature data are input into the PV estimation model, the corresponding and accurate predicted power generation value can be generated.

In order to explore the estimation results under different weather conditions, the present invention conducts research on four different weather patterns, such as sunny days, rainy days, cloudy days, and cloudy days when clear. The estimation results of four different weather patterns such as cloudy before and after parameter optimization. From these curve trends, it can be observed that after the parameters are optimized, the estimation of power generation is significantly improved, and the estimation error before and after optimization The size can be known from Table 2. Figures 8 to 11 show the parameter estimation comparison between the double-diode model and the single-second body model for four different weather patterns, including sunny days, rainy days, cloudy days, and cloudy days, respectively (where the single-second body model is used). Through the parameter selection process, 9 parameters were selected from the original 13 parameters for parameter optimization). From these curve trends, it can be observed that the dual-diode model has better estimation results. Table 3 compares the estimation results of the two models in each time period. The two-diode model has better estimation results in the low illumination period in the morning and afternoon, and the two models lead each other in the high-illumination period at noon. The overall average error shows that the dual-diode model has better estimation results. good estimates.

<Table 1>

The above-mentioned examples are only some embodiments of the present invention, and are not intended to limit the present invention. According to the creative spirit and characteristics of the present invention, those made with slight changes and modifications should also be included in the scope of this patent.

To sum up, the embodiment of the present invention can indeed achieve the expected use effect, and the specific technical means disclosed by it have not only not been seen in similar products, but also have not been disclosed before the application, which fully complies with the patent law. According to the regulations and requirements, it is really grateful to file an application for an invention patent in accordance with the law, and ask for the review and approval of the patent.

Claims (6)

A method for estimating parameters of a dual-diode model of a solar cell, the steps of which include: Step 1: Among the complex parameters that will change under different illumination and temperature, use a sensitivity matrix to calculate the complex parameters at a steady-state operating point. Among them, those with little effect on the output are eliminated, and the remaining parameters are determined by principal component analysis to determine the effect of each remaining parameter relative to the overall output measurement, and then the Gram-Schmidt orthogonal transformation is applied to establish the remaining parameters. In order to eliminate the remaining dependencies between the parameters, and then select the parameters that affect the output; Step 2: The parameters that affect the output selected in step 1 are established in the Matlab/Simulink environment as double two A polar body-based PV estimation model, and the whale optimization algorithm is applied to estimate the parameters that affect the output, so as to obtain the best parameter solution; wherein, the whale optimization algorithm includes: Step 2-1: Take the parameter that affects the output selected in Step 1 as the initial parameter, and randomly give each initial parameter an initial parameter solution position vector, and set the number of iterations at the same time; Step 2-2: System The actual measured input value and output value are sent into the PV estimation model to calculate the estimated value, and then the estimated value is integrated with the actual measured input value and output value, and then each calculated value is obtained. 1. The adaptation value of the position vector of the initial parameter solution; Step 2-3: According to the method of encircling the prey in the bubble net attack method, in the process of reducing the mechanism of encircling the best feasible solution, first assume that the current best feasible solution is the the initial parameter solution The best position of the initial parameter solution is used to move towards the direction of the best position of the initial parameter solution using the spiral path of the spiral equation, and the position of the initial parameter solution is updated according to the best position; Step 2-4: Use the prey search method Exploring and randomly selecting parametric solutions, the parametric solutions in these searches update their positions according to random positions, and then move towards the best solution in the whole domain; Step 2-5: The parametric solutions after updating the positions are sent to PV estimation Calculate the estimated power value in the measurement model, then integrate the estimated power value with the actual measured input value and output value, and then calculate the fitness value of the parameter solution position vector after each updated position, and update the most Step 2-6: Check whether it meets the set number of iterations, if not, go back to step 2-3; if it matches, the updated best position solution is the best parameter solution, and output The optimal parameter solution and end. The method for estimating parameters of a solar cell bidiode model according to claim 1, wherein, in the principal component analysis method described in step 1, the principal component is a covariance matrix

The eigenvector of , where the first principal component is the eigenvector corresponding to the largest eigenvalue (eigenvalue) in the X matrix, which is the maximum direction of the overall change, and the remaining principal components are arranged in order with respect to the overall change. Contribution, therefore, the overall effect of the jth parameter is expressed as:

in

is the effect of the jth parameter relative to the overall variable, α _i is the eigenvalue of the ith output, C _ij is the principal component element (or contribution) of the jth parameter relative to the ith output, m is the output number. The method for estimating parameters of a solar cell bidiode model according to claim 2, wherein, in step 1 to eliminate the dependencies between parameters, X _i =[ X _i (1), X _i (2), …, X _i ( N )] ^T is the eigenvector of the ith sample, and N is the number of samples, then the eigenmatrix is defined as follows:

The characteristic matrix X can be decomposed into: X = QR ; where R is an upper triangular matrix, and Q is an orthogonal matrix, that is,

where q _i is the new eigenvector of the orthogonal space. In the orthogonal decomposition process using Gram-Schmidt, the following procedure is used to construct the orthogonal matrix: q ₁ = x ₁ ;

in,

; Map the X space in the above formula of X = QR to the space Q : Q = R ^-1 X , to construct the orthogonal matrix. The method for estimating parameters of a solar cell bidiode model according to claim 5, wherein in step 2-1, the position vector of each initial parameter solution is randomly generated, as shown below: x _i,j (0)= x _i,min + rand ×( x _i,max - x _i,min ), i =1,2,..., S,j =1,2,..., P ; where X _i,j (0 ) is the jth parameter to solve the ith parameter (variable), X _i,min and X _i,max are the upper and lower limits of the parameters, the upper and lower limits of the parameters are set by the user according to experience, rand is between 0 and 1 Uniform random number, S is the number of parameters, P is the number of parameter solutions, and the position vector of the jth parameter solution can be expressed as follows:

( t )=[all selected parameters after step 1] (t is the number of iterations). The method for estimating parameters of a solar cell bidiode model according to claim 6, wherein in step 2-2, the input values are illuminance and temperature, and the output values are voltage, current, and power. The method for estimating parameters of a solar cell bidiode model according to claim 5, wherein the calculation of the fitness value of each initial parameter solution position vector is as follows:

Among them, P _j,est is the estimated value of the PV power generation model, and P _j,mea is the actual measured value.

Images