Optimization method for copper metallization of silicon heterojunction solar cell

Abstract

The invention discloses a method for optimizing copper metallization of a silicon heterojunction solar cell, which belongs to the technical field of photovoltaic copper electroplating and comprises the following steps: acquiring copper metallization production data variables of the silicon heterojunction solar cell within a preset time period; and inputting the copper metallization production data variable into the trained optimization model to obtain a copper metallization optimization condition prediction value of the target silicon heterojunction solar cell product. According to the method, the Pearson analysis method and the XGBoost algorithm are utilized to carry out consistency sorting on the copper metallization production data variables, the optimal variable feature set is obtained through variable feature screening, and the SVM algorithm is adopted to carry out optimization model training on the optimal variable feature set; and a copper metallization optimization condition prediction value of a target silicon heterojunction solar cell product is obtained through the trained silicon heterojunction solar cell copper metallization optimization model, so that a current field is uniformly distributed during electroplating, the surface of an electroplated layer is compact and bright, the morphology is uniform and smooth, and the photoelectric conversion efficiency of the cell is improved.

Description

Technical Field

[0001] The invention relates to the technical field of photovoltaic copper electroplating, and in particular to an optimization method for copper metallization of a silicon heterojunction solar cell. Background Art

[0002] Silicon heterojunction solar cells have the characteristics of high conversion efficiency, low process temperature, simple process and high stability, and have become the focus of research in the current photovoltaic industry. At present, silicon heterojunction solar cells need to be screen-printed at low temperatures due to the sensitive amorphous silicon characteristics of the passivation layer. However, the resistivity of low-temperature silver paste is much higher than that of high-temperature silver paste, which means that a large amount of low-temperature silver paste is needed to obtain a sufficiently low line resistance, which greatly increases the production cost of silicon heterojunction solar cells. With the continuous expansion of the production scale of the photovoltaic industry, the demand for low-temperature silver paste will also increase sharply. Therefore, in order to reduce production costs, the copper metallization of silicon heterojunction solar cells has received more and more attention from the industry.

[0003] Copper metallization of silicon heterojunction solar cells has become a research focus in this field because of its lower line resistance, higher aspect ratio and lower production cost, which can significantly reduce shadow loss, increase electrode conductivity and filling efficiency. In the existing copper metallization technology of silicon heterojunction solar cells, physical vapor deposition (PVD) technology is usually used to deposit a seed layer first, and then copper metallization is achieved on this seed layer. However, this vacuum process often adjusts a single parameter and debugs the process while keeping other parameters unchanged. It is not possible to use precise linkage electroplating parameters to ensure the uniformity of current distribution during electroplating. In particular, when the size of the wafer increases and the speed of the production line increases, the adhesion between the electroplated copper grid line and TCO is poor, and the surface morphology and phase structure of the copper electroplating layer are significantly affected, which in turn affects the photoelectric conversion efficiency of the battery. Summary of the invention

[0004] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to achieve uniform current field distribution during electroplating, dense and bright surface of the electroplating layer, and uniform and smooth morphology by jointly adjusting parameters such as current density, ratio of anode and cathode areas, and distance between anode and cathode, so as to improve the photoelectric conversion efficiency of the battery.

[0005] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:

[0006] The present invention provides a method for optimizing copper metallization of a silicon heterojunction solar cell, comprising:

[0007] Obtaining copper metallization production data variables of silicon heterojunction solar cells within a preset time period, wherein the production data includes current density, anode-cathode area ratio, anode-cathode spacing, electroplating waveform, electrode aspect ratio, electroplating requirement ratio, average electroplating area, and actual electroplating area;

[0008] Calculate the TP value, hole copper electroplating efficiency, and electroplating uniformity according to the production data;

[0009] Pearson analysis method was used to calculate the correlation coefficient between TP value, hole copper plating efficiency, plating uniformity and current density, anode-cathode area ratio, anode-cathode distance, plating waveform, electrode aspect ratio, plating requirement ratio, average plating area and actual plating area;

[0010] According to the correlation analysis results of TP value, hole copper electroplating efficiency and electroplating uniformity, the data of four dimensional factors with larger correlation coefficients are selected and input into the trained silicon heterojunction solar cell copper metallization optimization model to obtain the predicted value of copper metallization optimization conditions for the target silicon heterojunction solar cell product;

[0011] Furthermore, before calculating the TP value, hole copper electroplating efficiency, and electroplating uniformity according to the production data, the method further includes:

[0012] Preprocessing the acquired production data;

[0013] The TP value, hole copper electroplating efficiency and electroplating uniformity are calculated based on the production data after pretreatment.

[0014] Furthermore, according to the correlation analysis results of TP value, four dimensional factors with larger correlation coefficients, namely current density, anode-cathode spacing, electrode aspect ratio and electroplating requirement ratio, are selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to TP value.

[0015] Furthermore, according to the correlation analysis results of the hole copper plating efficiency, four dimensional factors with larger correlation coefficients, namely current density, anode and cathode area ratio, electrode aspect ratio and average plating area, were selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to the hole copper plating efficiency.

[0016] Furthermore, according to the correlation analysis results of electroplating uniformity, four dimensional factors with larger correlation coefficients, namely current density, anode-cathode area ratio, anode-cathode spacing and electroplating waveform, were selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to hole copper plating efficiency.

[0017] Furthermore, the highly correlated production data is input into a trained silicon heterojunction solar cell copper metallization optimization model, further comprising:

[0018] The historical data sets of copper metallization of silicon heterojunction solar cells with high correlation with TP value, hole copper plating efficiency and plating uniformity are constructed respectively;

[0019] The XGBoost algorithm is used to perform variable feature screening on the historical data set of copper metallization of silicon heterojunction solar cells to obtain the optimal variable feature set.

[0020] The SVM algorithm is used to train the model of the optimal variable feature set, and a trained copper metallization optimization model for silicon heterojunction solar cells is obtained.

[0021] Furthermore, the XGBoost algorithm is used to perform variable feature screening on the historical data set of copper metallization of silicon heterojunction solar cells to obtain the optimal variable feature set, including:

[0022] The historical data set of copper metallization of silicon heterojunction solar cells is divided into a training set and a test set, wherein the historical data set of copper metallization of silicon heterojunction solar cells includes production data with a large correlation with TP value, electroplating uniformity and hole copper electroplating efficiency, and an ideal condition value of copper metallization;

[0023] Obtaining the importance of each silicon heterojunction solar cell copper metallization historical data variable in the training set;

[0024] According to the importance of the data variables and the dimensional factors with greater correlation with TP value, electroplating uniformity and hole copper electroplating efficiency, consistency sorting is performed;

[0025] The RMSE mean and R-square mean of the importance of the historical data set variables of copper metallization of silicon heterojunction solar cells in the three sequences are calculated; the two historical data set variables of copper metallization of silicon heterojunction solar cells with the best and second-best RMSE mean and R-square mean are retained to obtain the optimal variable feature set.

[0026] Furthermore, the obtaining of the importance of the historical data variables of copper metallization of each silicon heterojunction solar cell in the training set includes:

[0027]

[0028] in:

[0029] Im represents the importance value of the historical data variable of copper metallization of silicon heterojunction solar cells, and the important value represents the importance of the historical data variable of copper metallization of silicon heterojunction solar cells;

[0030] G j Represents the cumulative sum of the first-order partial derivatives of the samples contained in each XGBoost tree leaf node j;

[0031] H jRepresents the cumulative sum of the second-order partial derivatives of the samples contained in each XGBoost tree leaf node j;

[0032] λ represents the value of randomly changing the training set at a variable in a silicon heterojunction solar cell copper metallization historical data set, which is used to control the size of the leaf node weight.

[0033] Furthermore, the SVM algorithm is used to perform model training on the optimal variable feature set to obtain a trained silicon heterojunction solar cell copper metallization optimization model, including:

[0034] Using the Laplace kernel function, the grid search method was used to obtain the optimal model parameters;

[0035] According to the optimal model parameters and the optimal variable feature set, a trained silicon heterojunction solar cell copper metallization optimization model is obtained.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The invention provides an optimization method for copper metallization of a silicon heterojunction solar cell. The method comprises the following steps: calculating electroplating effect evaluation parameters, including TP value, hole copper electroplating efficiency and electroplating uniformity, using a Pearson analysis method to obtain electroplating parameters with large correlation coefficients with the TP value, hole copper electroplating efficiency and electroplating uniformity, using an XGBoost algorithm to perform consistency sorting with the electroplating parameters with large correlation coefficients, performing feature screening processing on historical data of copper metallization of silicon heterojunction solar cells, obtaining a feature set of optimal variables, performing model training on the feature set of optimal variables through a SVM algorithm, obtaining a trained copper metallization optimization model for silicon heterojunction solar cells, and using the trained copper metallization optimization model for silicon heterojunction solar cells to predict copper metallization optimization conditions output by a target silicon heterojunction solar cell. The method can solve the problems of uneven current field distribution during electroplating and significant influence on the surface morphology and phase structure of a copper electroplating layer caused by adjusting only a single parameter and inaccurate parameter adjustment during the electroplating process, achieving uniform current field distribution, dense and bright surface of the electroplating layer and uniform and smooth morphology, and improving the photoelectric conversion efficiency of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. It is obvious that the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 It is a schematic flow chart of a method for optimizing copper metallization of a silicon heterojunction solar cell provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. It should be noted that the following embodiments are only used to illustrate the implementation method and typical parameters of the present invention, and are not used to limit the parameter range described in the present invention. Reasonable changes derived therefrom are still within the scope of protection of the claims of the present invention.

[0041] In order to illustrate the technical solution of the present invention, a specific embodiment is provided below for illustration.

[0042] Figure 1 This is a schematic flow chart of an optimization method for copper metallization of a silicon heterojunction solar cell provided by an embodiment of the present invention, which is described in detail as follows:

[0043] S1. Obtain copper metallization production data variables of silicon heterojunction solar cells within a preset time period, wherein the obtained production data refers to electroplating process production data derived from a database in the past 30 days.

[0044] Among them, the copper metallization production data variables of the silicon heterojunction solar cell include current density, anode-cathode area ratio, anode-cathode spacing, electroplating waveform, electrode aspect ratio, electroplating requirement ratio, average electroplating area, actual electroplating area and other field data.

[0045] S2. Preprocess the acquired copper metallization production data variables of silicon heterojunction solar cells, including but not limited to: removing unimportant data, deleting duplicate data, deleting data outliers, and deleting data missing records.

[0046] S3, calculating the TP value, hole copper electroplating efficiency and electroplating uniformity according to the copper metallization production data variables of the pre-treated silicon heterojunction solar cell, including the following steps:

[0047] Calculate the TP value based on the deep plating capacity calculation formula and production data, including:

[0048]

[0049] in:

[0050] The B series points represent the coating thickness on the hole wall;

[0051] The C series points represent the coating thickness around the orifice;

[0052] The A series of points represent the coating thickness on the board surface.

[0053] As an example, the average values ​​of the coating thicknesses at points B1 and B2, B3 and B4, C1 and C2, A1 and A2, and A3 and A4 in the database in the past 30 days are 25.22μm, 23.21μm, 24.69μm, 27.85μm, and 27.68μm, respectively; according to the TP value calculation formula and the above data, the TP value can be obtained to be 87.78%;

[0054] Calculate the hole copper plating efficiency based on the actual plating area and current density, including:

[0055] For example, hole copper plating efficiency = actual plating area / current density;

[0056] The electroplating uniformity evaluation index is obtained based on the calculated uniformity coefficient, including:

[0057]

[0058] in:

[0059] CoV is the ratio of the standard deviation of the coating thickness to the mean value, usually expressed as a percentage;

[0060] σ is the standard deviation;

[0061] μ is the average value of the coating thickness.

[0062] S4. Use Pearson analysis method to calculate the correlation coefficient between TP value, hole copper plating efficiency, plating uniformity and current density, anode-cathode area ratio, anode-cathode distance, plating waveform, electrode aspect ratio, plating requirement ratio, average plating area and actual plating area.

[0063] For example, the correlation coefficients of TP value, hole copper plating efficiency, plating uniformity and current density, cathode-cathode distance, cathode-cathode area ratio, electrode height-width ratio, plating requirement ratio, average plating area and plating waveform calculated by Pearson analysis are shown in Table 1 below:

[0064] Table 1

[0065] TP value Hole copper plating efficiency Plating uniformity Current density 0.75 0.65 0.73 Anode-Cathode Area Ratio 0.43 0.74 0.68 Anode-Cathode Distance 0.65 0.41 0.62 Electrode aspect ratio 0.63 0.58 0.48 Electroplating requirement ratio 0.59 0.46 0.36 Average plating area 0.23 0.55 0.45 Electroplating waveform 0.36 0.21 0.59

[0066] As an example, as shown in Table 1, the correlation coefficient between the TP value and the current density is 0.75.

[0067] S5. According to the correlation analysis results of TP value, four dimensional factors with larger correlation coefficients, namely, current density, anode-cathode spacing, electrode aspect ratio and electroplating requirement ratio, are selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to TP value;

[0068] According to the correlation analysis results of copper hole electroplating efficiency, four dimensional factors with larger correlation coefficients, namely current density, anode and cathode area ratio, electrode aspect ratio and average electroplating area, were selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to copper hole electroplating efficiency.

[0069] According to the correlation analysis results of electroplating uniformity, four dimensional factors with larger correlation coefficients, namely current density, anode-cathode area ratio, anode-cathode spacing and electroplating waveform, were selected as input parameters of the copper metallization optimization model for silicon heterojunction solar cells related to hole copper plating efficiency.

[0070] S6. Inputting the highly correlated production data into the trained silicon heterojunction solar cell copper metallization optimization model to obtain the predicted value of the copper metallization optimization condition of the target silicon heterojunction solar cell product, including:

[0071] A historical data set of copper metallization of silicon heterojunction solar cells with a high correlation with TP value, hole copper plating efficiency and plating uniformity was constructed respectively.

[0072] S7. Use the XGBoost algorithm to perform variable feature screening on the historical data set of copper metallization of silicon heterojunction solar cells to obtain the optimal variable feature set. Specifically:

[0073] The historical data set of copper metallization of silicon heterojunction solar cells, which is highly correlated with TP value, hole copper electroplating efficiency and electroplating uniformity, is divided into training set and test set, where the capacity of the historical data set of copper metallization is 150 groups, the capacity of the training set is 105 groups, and the capacity of the test set is 35 groups;

[0074] The silicon heterojunction solar cell copper metallization historical data set includes production data with a high correlation with TP value, electroplating uniformity and hole copper electroplating efficiency, as well as ideal copper metallization condition values, specifically:

[0075] The production data with a greater correlation with electroplating uniformity and hole copper electroplating efficiency include: current density, cathode-cathode spacing, electrode aspect ratio and electroplating requirement ratio parameters with a greater correlation with TP value; current density, cathode-cathode area ratio, electrode aspect ratio and average electroplating area parameters with a greater correlation with hole copper electroplating efficiency; current density, cathode-cathode area ratio, cathode-cathode spacing and electroplating waveform parameters with a greater correlation with electroplating uniformity; the ideal condition value of copper metallization is specifically the parameter value under the ideal electroplating condition of each parameter, and an indicator system set is constructed based on the historical parameter variables of copper metallization:

[0076] The current density, the ratio of the anode and cathode areas, the distance between the anode and cathode, the aspect ratio of the electrodes, the required ratio of electroplating, the average electroplating area, and the electroplating waveform are set as independent variables, and the TP value, electroplating uniformity, and hole copper electroplating efficiency are regarded as dependent variables;

[0077] Obtaining the importance of the copper metallization historical data variables of each silicon heterojunction solar cell in the training set, wherein the calculation of the importance of the copper metallization historical variable data includes:

[0078]

[0079] in:

[0080] Im represents the importance value of the historical data variable of copper metallization of silicon heterojunction solar cells, and the important value represents the importance of the historical data variable of copper metallization of silicon heterojunction solar cells;

[0081] G j Represents the cumulative sum of the first-order partial derivatives of the samples contained in each XGBoost tree leaf node j;

[0082] H j Represents the cumulative sum of the second-order partial derivatives of the samples contained in each XGBoost tree leaf node j;

[0083] λ represents the value of randomly changing the training set at a variable of a silicon heterojunction solar cell copper metallization historical data set, which is used to control the weight of the leaf node. Among them, the importance of the copper metallization historical variable data is calculated, specifically:

[0084] In this embodiment, when tuning the parameters of the XGBoost algorithm, a relatively high learning rate is usually selected, usually between 0.05 and 0.3, to determine the number of trees;

[0085] Adjust the tree-specific parameter max_delta_step to limit the maximum step size of each tree weight change and set the regularization parameter lambda to 1.0;

[0086] Gradually reduce the learning rate to determine the optimal parameters.

[0087] S8. The calculation results of the importance of the copper metallization historical variable data are compared with the results of the Pearson analysis method for calculating the TP value, hole copper electroplating efficiency, electroplating uniformity and current density, anode and cathode area ratio, anode and cathode spacing, electroplating waveform, electrode aspect ratio, electroplating requirement ratio, average electroplating area and actual electroplating area correlation coefficient, and the results of the two calculation methods are ranked according to consistency.

[0088] S9, calculate the RMSE mean and R-square mean of the importance of the historical dataset of copper metallization of silicon heterojunction solar cells in the three sequences;

[0089] The two silicon heterojunction solar cell copper metallization historical data set variables with the best and second best RMSE mean and R-square mean are retained to obtain the optimal variable feature set. In this embodiment, the RMSE and R-square optimal variables related to the TP value are current density, and the RMSE and R-square second best variables are the ratio of the anode and cathode areas; the RMSE and R-square optimal variables related to the hole copper plating efficiency are the anode and cathode spacing, and the RMSE and R-square second best variables are the electrode height-to-width ratio; the RMSE and R-square optimal variables related to the plating uniformity are the ratio of the anode and cathode areas, and the RMSE and R-square second best variables are the plating waveform.

[0090] S10, using the SVM algorithm to perform model training on the optimal variable feature set, and obtain a trained silicon heterojunction solar cell copper metallization optimization model, specifically:

[0091] The SVM algorithm involves many parameters, such as regularization parameters, kernel functions, maximum number of iterations, gamma and sigma parameters. This embodiment adopts the Laplace kernel function with fast operation speed, strong adaptability and excellent generalization performance, and adopts the grid search method to obtain the optimal model parameters. The specific values ​​of the optimal model parameters of the Laplace kernel function are gamma=0.1 and sigma=3, where the sigma parameter is used to control the smoothness of the Laplace kernel function, and the gamma value determines the distribution form of the Laplace curve.

[0092] According to the optimal model parameters and the optimal variable feature set, a trained silicon heterojunction solar cell copper metallization optimization model is obtained, and the copper metallization optimization condition prediction value of the target silicon heterojunction solar cell product is obtained. The TP value, electroplating uniformity and hole copper electroplating efficiency are calculated according to the predicted value to evaluate the electroplating effect.

[0093] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for optimizing copper metallization of silicon heterojunction solar cells, characterized in that: The following steps are involved: Obtaining copper metallization production data variables of silicon heterojunction solar cells within a preset time period, wherein the production data includes current density, anode-cathode area ratio, anode-cathode spacing, electroplating waveform, electrode aspect ratio, electroplating requirement ratio, average electroplating area, and actual electroplating area; Calculate the TP value, hole copper electroplating efficiency, and electroplating uniformity according to the production data; Pearson analysis method was used to calculate the correlation coefficient between TP value, hole copper plating efficiency, plating uniformity and current density, anode-cathode area ratio, anode-cathode distance, plating waveform, electrode aspect ratio, plating requirement ratio, average plating area and actual plating area; According to the correlation analysis results of TP value, hole copper plating efficiency and plating uniformity, the data of four dimensional factors with larger correlation coefficients are selected and input into the trained silicon heterojunction solar cell copper metallization optimization model to obtain the predicted value of copper metallization optimization conditions for the target silicon heterojunction solar cell product.

2. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 1, characterized in that: Before calculating the TP value, hole copper electroplating efficiency, and electroplating uniformity according to the production data, the method further includes: Preprocessing the acquired production data; The TP value, hole copper electroplating efficiency and electroplating uniformity are calculated according to the production data after pretreatment.

3. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 1, characterized in that: According to the correlation analysis results of TP value, four dimensional factors with larger correlation coefficients, namely current density, anode-cathode spacing, electrode aspect ratio and electroplating requirement ratio, are selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to TP value.

4. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 1, characterized in that: According to the correlation analysis results of the hole copper plating efficiency, four dimensional factors with larger correlation coefficients, namely current density, anode and cathode area ratio, electrode aspect ratio and average plating area, were selected as input parameters of the copper metallization optimization model of silicon heterojunction solar cells related to the hole copper plating efficiency.

5. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 1, characterized in that: According to the correlation analysis results of electroplating uniformity, four dimensional factors with larger correlation coefficients, namely current density, anode-cathode area ratio, anode-cathode spacing and electroplating waveform, were selected as input parameters of the copper metallization optimization model for silicon heterojunction solar cells related to hole copper plating efficiency.

6. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 1, characterized in that: Inputting the highly relevant production data into the trained silicon heterojunction solar cell copper metallization optimization model, further comprising: The historical data sets of copper metallization of silicon heterojunction solar cells with high correlation with TP value, hole copper plating efficiency and plating uniformity are constructed respectively; The XGBoost algorithm is used to perform variable feature screening on the historical data set of copper metallization of silicon heterojunction solar cells to obtain the optimal variable feature set. The SVM algorithm is used to train the model of the optimal variable feature set, and a trained copper metallization optimization model for silicon heterojunction solar cells is obtained.

7. The method for optimizing copper metallization of silicon heterojunction solar cells according to claim 6, characterized in that: The XGBoost algorithm is used to perform variable feature screening on the historical data set of copper metallization of silicon heterojunction solar cells to obtain the optimal variable feature set, including: The historical data set of copper metallization of silicon heterojunction solar cells is divided into a training set and a test set, wherein the historical data set of copper metallization of silicon heterojunction solar cells includes production data with a large correlation with TP value, electroplating uniformity and hole copper electroplating efficiency, and an ideal condition value of copper metallization; Obtaining the importance of each silicon heterojunction solar cell copper metallization historical data variable in the training set; According to the importance of the data variables and the dimensional factors with greater correlation with TP value, electroplating uniformity and hole copper electroplating efficiency, consistency sorting is performed; The RMSE mean and R-square mean of the importance of the historical dataset variables of copper metallization of silicon heterojunction solar cells in the three sequences are calculated; the two historical dataset variables of copper metallization of silicon heterojunction solar cells with the best and second-best RMSE mean and R-square mean are retained to obtain the optimal variable feature set.

8. The method for optimizing copper metallization of silicon heterojunction solar cells according to claim 7, characterized in that: The obtaining of the importance of the historical data variables of copper metallization of each silicon heterojunction solar cell in the training set includes: in: Im represents the importance value of the historical data variable of copper metallization of silicon heterojunction solar cells, and the important value represents the importance of the historical data variable of copper metallization of silicon heterojunction solar cells; G j Represents the cumulative sum of the first-order partial derivatives of the samples contained in each XGBoost tree leaf node j; H j Represents the cumulative sum of the second-order partial derivatives of the samples contained in each XGBoost tree leaf node j; λ represents the value of randomly changing the training set at a variable in a silicon heterojunction solar cell copper metallization historical data set, which is used to control the size of the leaf node weight.

9. The method for optimizing copper metallization of a silicon heterojunction solar cell according to claim 6, characterized in that: The method of using the SVM algorithm to perform model training on the optimal variable feature set to obtain a trained silicon heterojunction solar cell copper metallization optimization model includes: Using the Laplace kernel function, the grid search method was used to obtain the optimal model parameters; According to the optimal model parameters and the optimal variable feature set, a trained silicon heterojunction solar cell copper metallization optimization model is obtained.

Images