A Multivariable and Electrical Performance Prediction Method for Water-Based Photovoltaic Devices

Abstract

This invention discloses a method for predicting the multivariable and electrical performance of water-based photovoltaic (VPS) devices, belonging to the technical field of VPS device variable and performance prediction. The method includes: acquiring multivariable parameter values ​​of the VPS device and inputting them into a trained decision tree-optimized XGBOOST algorithm to predict the multivariable and electrical performance of the VPS device, obtaining the prediction results; training the decision tree-optimized XGBOOST algorithm includes: constructing a training set of multivariable and electrical performance data for VPS devices to train the decision tree-optimized XGBOOST algorithm; during training, adjusting the parameters of the decision tree-optimized XGBOOST algorithm according to a preset parameter tuning order to obtain a trained decision tree-optimized XGBOOST algorithm. This method overcomes the limitations of existing prediction methods that can only control a single variable through experimental means, enabling machine prediction and improving prediction efficiency.

Description

Technical Field

[0001] This invention relates to a method for predicting the multivariable and electrical performance of water-voltaic devices, belonging to the technical field of variable and performance prediction for water-voltaic devices. Background Technology

[0002] With the global fossil fuel crisis and its associated environmental pollution intensifying, the research and utilization of new clean energy sources have become an important aspect of sustainable development. Compared to other clean energy sources, hydroelectric power generation utilizes the evaporation of ambient water or moisture to generate electricity, offering the advantage of a wide range of applications.

[0003] Currently, in the fabrication of water-based photovoltaic (VPC) devices, multiple variables such as the material's zeta potential, three-dimensional dimensions, and the temperature of the testing environment synergistically affect the final electrical performance of the device. Therefore, technical means are needed to assist in exploring the relationship between the materials, structure, environment, and electrical performance of water-based VPC devices.

[0004] Existing research techniques are limited to controlling a single variable through experimental methods, and then iterating experiments on that single variable to obtain the independent relationship between each variable and electrical performance. For problems involving multiple variables, experimental research strategies suffer from problems such as cumbersome steps, high costs, limited correlation of conclusions, and difficulty in confirming subsequent experimental conditions. Summary of the Invention

[0005] The purpose of this invention is to provide a multivariable and electrical performance prediction method for water-based photovoltaic devices, which can overcome the limitations of existing prediction methods that can only control a single variable through experimental means, realize machine prediction, and improve prediction efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for predicting the multivariable and electrical performance of water-based photovoltaic devices, comprising:

[0008] Obtain multivariable parameter values ​​for water-based photovoltaic devices;

[0009] The multivariate parameter values ​​of the water-voltaic device are input into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and the prediction results of multivariate and electrical performance of the water-voltaic device are obtained.

[0010] The training methods for the decision tree optimization XGBOOST algorithm include:

[0011] Construct a training set for the multivariable and electrical performance of water-based photovoltaic devices;

[0012] The decision tree optimization XGBOOST algorithm was trained using a multivariable and electrical performance training set of water-voltaic devices. During the training process, the parameters of the decision tree optimization XGBOOST algorithm were tuned according to a preset parameter tuning order to obtain a well-trained decision tree optimization XGBOOST algorithm.

[0013] In conjunction with the first aspect, further, the construction of a multivariable and electrical performance training set for water-based photovoltaic devices includes:

[0014] The apparent dimensions and electrical performance of the water-voltaic device were tested to obtain multivariable and electrical performance data of the water-voltaic device;

[0015] The multivariable and electrical performance data of water-voltaic devices are extended to obtain a training set of multivariable and electrical performance data of water-voltaic devices;

[0016] Among them, the multivariables include the material parameters, size parameters and test environment parameters of the water-voltaic device. The material parameters include the zeta potential of the water-voltaic device, the size parameters include the length, width and thickness of the water-voltaic device, and the test environment parameters include the temperature and salt concentration of the test environment.

[0017] Electrical performance includes the electrical parameters of the water-volt device, which include the voltage, current, and power density of the water-volt device.

[0018] Each data point in the training set of multivariable and electrical performance data for water-voltaic devices includes the zeta potential, length, width, thickness, voltage, current, and power density of the water-voltaic device, as well as the temperature and salt concentration of the test environment.

[0019] In conjunction with the first aspect, furthermore, the decision tree optimization XGBOOST algorithm is trained using a multivariable and electrical performance training set of water-based photovoltaic devices. During the training process, the parameter tuning of the decision tree optimization XGBOOST algorithm is performed according to a preset parameter tuning order, including:

[0020] Standardize the training set of multivariable and electrical performance of water-voltaic devices;

[0021] The material parameters, dimensional parameters, and test environment parameters of the water-voltaic devices in the standardized multivariable and electrical performance training set were set as independent variables.

[0022] The electrical parameters of the water-voltaic devices in the training set of standardized multivariable and electrical performance data were set as dependent variables.

[0023] The data corresponding to the independent variables are input into the decision tree optimization XGBOOST algorithm for training. Through cross-validation, the predicted fitting curves of the independent and dependent variables are obtained, as well as the goodness of fit, mean absolute error and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves.

[0024] During the training process, based on the preset parameter tuning order, within the preset parameter tuning range, the XGBOOST decision tree optimization algorithm is tuned to underfitting with the goals of optimal goodness of fit, optimal mean absolute error and optimal mean variance, so as to obtain the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error and optimal mean variance.

[0025] The decision tree optimization XGBOOST algorithm with the optimal parameter combination is used as the trained decision tree optimization XGBOOST algorithm.

[0026] In conjunction with the first aspect, further, during the training process, based on the preset parameter tuning order, the cyclic comparison method is used to tune the XGBOOST decision tree optimization algorithm to underfit within the preset parameter tuning range, with the objectives of optimal goodness of fit, optimal mean absolute error, and optimal mean variance, thereby obtaining the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

[0027] In conjunction with the first aspect, further, during the training process, based on the preset parameter tuning order, the XGBOOST decision tree optimization algorithm is tuned to underfitting within the preset parameter tuning range using the traversal method, with the objectives of optimal goodness of fit, optimal mean absolute error, and optimal mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

[0028] In conjunction with the first aspect, further, during the training process, based on the preset parameter tuning order, the grid search method is used to tune the XGBOOST decision tree optimization algorithm to underfit within the preset parameter tuning range, with the objectives of optimal goodness of fit, optimal mean absolute error, and optimal mean variance, thereby obtaining the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

[0029] In conjunction with the first aspect, further methods for setting the parameter tuning order include:

[0030] The importance function is used to analyze the influence of multiple variables on the electrical performance of water-voltaic devices.

[0031] The parameter tuning order is set according to the importance of the influence of multiple variables on the electrical performance of the water-voltaic device;

[0032] The parameter tuning order is as follows: number of decision trees, minimum sum of leaf weights, subsample size, learning rate, maximum depth, regularization coefficient, and leaf node loss for the XGBOOST decision tree optimization algorithm.

[0033] The preset parameter range for the number of decision trees in the XGBOOST algorithm for decision tree optimization is 1 to 200.

[0034] The preset parameter range for the minimum leaf weight sum of the decision tree optimization XGBOOST algorithm is 1~200;

[0035] The preset parameter range for subsample size in the XGBOOST algorithm for decision tree optimization is 0.1~1;

[0036] The preset parameter range for the learning rate of the decision tree optimization XGBOOST algorithm is 0~1;

[0037] The preset parameter range for the maximum depth of the decision tree optimization XGBOOST algorithm is 1~50;

[0038] The preset tuning parameter range for the regularization coefficient of the decision tree optimization XGBOOST algorithm is 0~5;

[0039] The preset parameter range for the leaf node loss of the decision tree optimization XGBOOST algorithm is 0~5.

[0040] Secondly, the present invention provides a device for predicting the multivariable and electrical performance of water-based photovoltaic devices, comprising:

[0041] Data acquisition module: used to acquire multivariable parameter values ​​of water-based photovoltaic devices;

[0042] The training set construction module is used to construct a multivariable and electrical performance training set for water-voltaic devices;

[0043] The training module is used to train the decision tree optimization XGBOOST algorithm using a multivariable and electrical performance training set of water-voltaic devices. During the training process, the decision tree optimization XGBOOST algorithm is tuned according to a preset parameter tuning order to obtain the trained decision tree optimization XGBOOST algorithm.

[0044] Prediction module: This module is used to input the multivariate parameter values ​​of the water-voltaic device into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and obtain the prediction results of the multivariate and electrical performance of the water-voltaic device.

[0045] Thirdly, the present invention provides a computer system, comprising:

[0046] Storage medium: used to store computer programs;

[0047] Processor: Used to execute the computer program to implement the multivariable and electrical performance prediction method for water-voltaic devices described in the first aspect.

[0048] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multivariable and electrical performance prediction method for water-voltaic devices described in the first aspect.

[0049] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the multivariable and electrical performance prediction method for water-voltaic devices described in the first aspect.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] The present invention provides a method for predicting the multivariate and electrical performance of water-voltaic devices. It utilizes a training set of multivariate and electrical performance data for water-voltaic devices to train a decision tree-optimized XGBOOST algorithm. During training, the algorithm is optimized according to a preset parameter tuning order. Using the optimized XGBOOST algorithm tuned by this invention to predict the multivariate and electrical performance of water-voltaic devices yields accurate prediction results. This method overcomes the limitations of existing prediction methods that rely on experimental manipulation of a single variable, enabling machine prediction. It also solves the problems of low efficiency, cumbersome procedures, high cost, and difficulty in confirming experimental conditions caused by the need for iterative experiments based on a single variable in existing prediction methods, significantly improving the prediction efficiency of multivariate and electrical performance of water-voltaic devices. Attached Figure Description

[0052] Figure 1 This is a flowchart of the multivariable and electrical performance prediction method for water-based photovoltaic devices provided in this embodiment of the invention;

[0053] Figure 2These are schematic diagrams illustrating the accuracy of the predicted values ​​of the dependent variables output by the decision tree optimization XGBOOST algorithm before and after parameter tuning for the training and test sets, respectively, provided in this embodiment of the invention. Specifically, (a) shows the accuracy of the predicted values ​​of the voltage of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm before parameter tuning for the training and test sets, (b) shows the accuracy of the predicted values ​​of the voltage of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm after parameter tuning for the training and test sets, (c) shows the accuracy of the predicted values ​​of the current of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm before parameter tuning for the training and test sets, (d) shows the accuracy of the predicted values ​​of the current of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm after parameter tuning for the training and test sets, (e) shows the accuracy of the predicted values ​​of the power density of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm before parameter tuning for the training and test sets, and (f) shows the accuracy of the predicted values ​​of the power density of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm after parameter tuning for the training and test sets.

[0054] Figure 3 This is a schematic diagram comparing the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables output by the decision tree optimization XGBOOST algorithm before and after parameter tuning, compared with the actual fitting curves, provided in the embodiments of the present invention. In this diagram, (a) represents the voltage of the water-volt device as the dependent variable, (b) represents the current of the water-volt device as the dependent variable, and (c) represents the power density of the water-volt device as the dependent variable.

[0055] Figure 4 This is a schematic diagram of the multivariate and electrical performance prediction results of a water-voltaic device obtained by using the decision tree optimization XGBOOST algorithm provided in this embodiment of the invention to predict the multivariate and electrical performance of a prediction array consisting of 80,000 unknown data points.

[0056] Figure 5 This is a schematic diagram of the error curves of the three error output variables—goodness of fit, mean absolute error, and mean variance—in the ergodic method provided in this embodiment of the invention.

[0057] Figure 6 This is a schematic diagram comparing the goodness of fit obtained by training in the traversal method provided in this embodiment of the invention with the goodness of fit under cross-validation. (a) represents a schematic diagram comparing the goodness of fit obtained by training before parameter tuning with the goodness of fit under cross-validation, and (b) represents a schematic diagram comparing the goodness of fit obtained by training after parameter tuning with the goodness of fit under cross-validation.

[0058] Figure 7This is a schematic diagram comparing the goodness of fit obtained by training in the grid search method provided in this embodiment of the invention with the goodness of fit under cross-validation. In this diagram, (a) represents the comparison of the goodness of fit obtained by training before parameter tuning with the goodness of fit under cross-validation, and (b) represents the comparison of the goodness of fit obtained by training after parameter tuning with the goodness of fit under cross-validation. Detailed Implementation

[0059] The technical solution of the present invention will be further described in detail below with reference to specific embodiments.

[0060] Embodiments of the present invention are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Unless otherwise specified, embodiments of the present invention and the technical features thereof can be combined with each other.

[0061] This invention provides a method for predicting the multivariable and electrical performance of water-based photovoltaic devices, comprising:

[0062] Obtain multivariable parameter values ​​for water-based photovoltaic devices;

[0063] The multivariate parameter values ​​of the water-voltaic device are input into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and the prediction results of multivariate and electrical performance of the water-voltaic device are obtained.

[0064] The training methods for the decision tree optimization XGBOOST algorithm include:

[0065] Construct a training set for the multivariable and electrical performance of water-based photovoltaic devices;

[0066] The decision tree optimization XGBOOST algorithm was trained using a multivariable and electrical performance training set of water-voltaic devices. During the training process, the parameters of the decision tree optimization XGBOOST algorithm were tuned according to a preset parameter tuning order to obtain a well-trained decision tree optimization XGBOOST algorithm.

[0067] The multivariate prediction method for the electrical performance of water-voltaic devices provided in this invention utilizes a training set of multivariate and electrical performance data for water-voltaic devices to train the decision tree optimization XGBOOST algorithm. During the training process, the XGBOOST algorithm is tuned according to a preset parameter tuning order. Using the XGBOOST algorithm tuned by this invention to predict the multivariate and electrical performance of water-voltaic devices yields accurate prediction results. This overcomes the limitations of existing prediction methods that rely on experimental manipulation of a single variable, enabling machine prediction. It also solves the problems of low efficiency, cumbersome steps, high cost, and difficulty in confirming experimental conditions caused by the need for iterative experiments on a single variable in existing prediction methods, significantly improving the prediction efficiency of multivariate and electrical performance of water-voltaic devices.

[0068] Figure 1 This is a flowchart illustrating a multivariable and electrical performance prediction method for water-based photovoltaic devices provided in this embodiment. This flowchart only shows the logical sequence of the method in this embodiment; however, different methods may be used without conflict. Figure 1 Complete the steps shown or described in the order indicated.

[0069] The multivariable and electrical performance prediction method for water-voltaic devices provided in this embodiment can be applied to a terminal and can be executed by a multivariable and electrical performance prediction device for water-voltaic devices. This device can be implemented by software and / or hardware and can be integrated into the terminal, such as any tablet computer or computer device with communication function.

[0070] This invention provides a method for predicting the multivariable and electrical performance of water-based photovoltaic devices, such as... Figure 1 As shown, the specific steps include the following:

[0071] Step 1: Obtain the multivariable parameter values ​​of the water-based photovoltaic device;

[0072] Step 2: Construct a training set for the multivariable and electrical performance of water-based photovoltaic devices;

[0073] In this embodiment, constructing the multivariable and electrical performance training set for water-voltaic devices specifically includes the following steps:

[0074] Step 1: Perform apparent size and electrical performance tests on the water-voltaic device to obtain multivariable and electrical performance data of the water-voltaic device;

[0075] In this embodiment, the multiple variables include the material parameters, size parameters, and test environment parameters of the water-voltaic device. The material parameters include the zeta potential of the water-voltaic device, the size parameters include the length, width, and thickness of the water-voltaic device, and the test environment parameters include the temperature and salt concentration of the test environment.

[0076] Electrical performance includes the electrical parameters of the water-voltaic device, including its voltage, current, and power density.

[0077] Step 2: Expand the multivariable and electrical performance data of the water-voltaic device to obtain the training set of multivariable and electrical performance of the water-voltaic device.

[0078] In this embodiment, multivariable and electrical performance data of water-voltaic devices are obtained based on apparent size and electrical performance tests. The multivariable and electrical performance data of water-voltaic devices are expanded through literature review to enhance the adaptability of the multivariable and electrical performance data of water-voltaic devices in different material structure systems, thus obtaining a training set of multivariable and electrical performance data of water-voltaic devices.

[0079] Specifically, the training set for the multivariable and electrical performance of water-voltaic devices includes 500 data points. Each data point in the training set includes the zeta potential, length, width, thickness, voltage, current, and power density of the water-voltaic device, as well as the temperature and salt concentration of the test environment.

[0080] Step 3: Use the multivariable and electrical performance training set of water-voltaic devices to train the decision tree optimization XGBOOST algorithm. During the training process, the decision tree optimization XGBOOST algorithm is tuned according to the preset parameter tuning order to obtain the trained decision tree optimization XGBOOST algorithm.

[0081] In this embodiment, the decision tree optimization XGBOOST algorithm is trained using a multivariable and electrical performance training set of water-voltaic devices. During the training process, the parameter tuning of the decision tree optimization XGBOOST algorithm based on a preset parameter tuning order specifically includes the following steps:

[0082] Step 1: Standardize the training set of multivariable and electrical performance of water-voltaic devices; set the material parameters, size parameters, and test environment parameters of water-voltaic devices in the standardized training set as independent variables; set the electrical parameters of water-voltaic devices in the standardized training set as dependent variables.

[0083] In this embodiment, the sklearn database is used to call the decision tree to optimize the XGBOOST algorithm, and the training set of multivariable and electrical performance of water-voltaic devices is imported to form an array matrix.

[0084] Specifically, based on the standardization of the training set for the multivariable and electrical performance of water-voltaic devices, the combined array matrix of material parameters, size parameters, and test environment parameters of water-voltaic devices in the training set is set as the independent variable, and the array matrix of electrical parameters of water-voltaic devices is set as the dependent variable. The array matrix is ​​then randomly divided, with 20% of the array matrix allocated to the test set and 80% allocated to the training set.

[0085] Step 2: Input the data corresponding to the independent variables into the decision tree optimization XGBOOST algorithm for training, and obtain the predicted fitting curves of the independent and dependent variables through cross-validation, as well as the goodness of fit, mean absolute error and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves.

[0086] In this embodiment, the data corresponding to the independent variables are input into the decision tree optimization XGBOOST algorithm for training, and through 5-fold cross-validation, the predicted fitting curves of the independent and dependent variables are obtained, as well as the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves.

[0087] Specifically, the data corresponding to the independent variables are input into the decision tree optimization XGBOOST algorithm for training. The decision tree optimization XGBOOST algorithm outputs the predicted values ​​of the dependent variables corresponding to each independent variable. Based on the predicted values ​​of the independent variables and their corresponding dependent variables, the predicted fitting curves of the independent and dependent variables are obtained. The predicted fitting curves of the independent and dependent variables are then compared with the actual fitting curves to obtain the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves.

[0088] Among the dependent variables, for the voltage of the water-based photovoltaic device, before parameter tuning, the accuracy of the predicted voltage values ​​of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm for the training and test sets is as follows: Figure 2 As shown in Figure (a), after parameter tuning, the accuracy of the predicted voltage values ​​of the water-based photovoltaic devices output by the decision tree optimization XGBOOST algorithm for the training and test sets is as follows: Figure 2 As shown in (b).

[0089] Among the dependent variables, for the current of the water-based photovoltaic device, before parameter tuning, the accuracy of the predicted current of the water-based photovoltaic device output by the decision tree optimization XGBOOST algorithm for the training set and test set is as follows: Figure 2 As shown in (c), after parameter tuning, the accuracy of the predicted current values ​​of the water-based photovoltaic devices output by the decision tree optimization XGBOOST algorithm for the training and test sets is as follows: Figure 2 As shown in (d).

[0090] Among the dependent variables, for the power density of the hydrovoltaic device, before parameter tuning, the accuracy of the predicted power density values ​​of the hydrovoltaic device output by the decision tree optimization XGBOOST algorithm for the training and test sets is as follows: Figure 2 As shown in (e), after parameter tuning, the accuracy of the predicted power density of the hydrovoltaic device output by the decision tree optimization XGBOOST algorithm for the training and test sets is as follows: Figure 2 As shown in (f).

[0091] Depend on Figure 2 As shown in (a) to (f), the predicted value of the dependent variable output by the optimized XGBOOST algorithm after parameter tuning is closer to the actual value than that before parameter tuning.

[0092] Among the dependent variables, for the voltage of the water-based photovoltaic device, the comparison of the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables obtained by the decision tree optimization XGBOOST algorithm before and after parameter tuning compared to the actual fitting curves is as follows: Figure 3 As shown in (a).

[0093] Among the dependent variables, for the current of the water-based photovoltaic device, the comparison of the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables obtained by the decision tree optimization XGBOOST algorithm before and after parameter tuning compared with the actual fitting curves is as follows: Figure 3 As shown in (b).

[0094] Among the dependent variables, for the power density of the water-based photovoltaic device, the comparison of the goodness of fit, mean absolute error, and mean variance of the predicted fitting curves of the independent and dependent variables obtained by the decision tree optimization XGBOOST algorithm before and after parameter tuning compared to the actual fitting curves is as follows: Figure 3 As shown in (c).

[0095] Depend on Figure 3 As shown in (a) to (c), the goodness of fit of the predicted fitting curves of the independent and dependent variables obtained by the optimized XGBOOST algorithm after parameter tuning is significantly higher than that of the actual fitting curves before parameter tuning, while the mean absolute error and mean variance are significantly lower.

[0096] Step 3: During the training process, based on the preset parameter tuning order, the XGBOOST decision tree optimization algorithm is tuned to underfitting within the preset parameter tuning range using the cyclic comparison method, traversal method, or grid search method, with the goal of achieving the best fit, the best mean absolute error, and the best mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under the conditions of best fit, best mean absolute error, and best mean variance.

[0097] In this embodiment, training begins initially based on the default parameters of the decision tree optimization XGBOOST algorithm.

[0098] During the training process, based on the preset parameter tuning order, the XGBOOST decision tree optimization algorithm is tuned to underfitting within the preset parameter tuning range using the cyclic comparison method, traversal method, or grid search method, with the goal of achieving the best fit, the best mean absolute error, and the best mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under the conditions of the best fit, the best mean absolute error, and the best mean variance.

[0099] In this embodiment, the parameter tuning of the XGBOOST algorithm for decision tree optimization using the traversal method specifically includes the following steps:

[0100] Step i: Based on the preset parameter tuning order, set each parameter as an input variable, select an interval based on the preset parameter tuning range, and set a reasonable step size;

[0101] Step ii: Define three error output variables: mean absolute error, mean variance, and generalization error;

[0102] Step iii: Run the decision tree optimization XGBOOST algorithm, import the input variables, traverse each value within the selected interval, and output the three error output variables corresponding to each training iteration;

[0103] Step iv: Plot the error curves for the three error output variables corresponding to each training iteration, and automatically compare them to obtain the input variable with the lowest error value.

[0104] In this embodiment, the error curves of the three error output variables—mean absolute error, mean variance, and generalization error—are as follows: Figure 5 As shown, the comparison between the goodness of fit obtained from training before parameter tuning and the goodness of fit under cross-validation is as follows: Figure 6 As shown in (a), the comparison between the goodness of fit obtained after parameter tuning and the goodness of fit under cross-validation is as follows: Figure 6 As shown in (b), Figure 6 In this process, the closer the goodness of fit obtained from training is to the goodness of fit obtained from cross-validation, the better. Figure 6 As shown in (a) to (b), the goodness of fit obtained after parameter tuning is closer to the goodness of fit under cross-validation than that before parameter tuning.

[0105] Taking the number of decision trees in the XGBOOST algorithm as an example, when the number of decision trees in the XGBOOST algorithm is adjusted from the default value of 100 to 4, the corresponding goodness of fit increases from 0.78 to 0.79.

[0106] In this embodiment, the parameter tuning of the decision tree optimization XGBOOST algorithm using the grid search method specifically includes the following steps:

[0107] Step i: For multiple parameters with strong internal correlation, they cannot be optimized one by one in sequence. Instead, both are traversed through the intervals simultaneously, and the fit of the calculation results of each variable combination is compared cross-referenced.

[0108] Step ii: Define the interval and step size of the variable group, and import the decision tree to optimize the XGBOOST algorithm;

[0109] Step iii: Output the optimal combination of variables for best fit, best mean absolute error, and best mean variance.

[0110] In this embodiment, the comparison between the goodness of fit obtained from training before parameter tuning and the goodness of fit under cross-validation is as follows: Figure 7 As shown in (a), the comparison between the goodness of fit obtained after parameter tuning and the goodness of fit under cross-validation is as follows: Figure 7 As shown in (b), Figure 7 In this process, the closer the goodness of fit obtained from training is to the goodness of fit obtained from cross-validation, the better. Figure 7 As shown in (a) to (b), the goodness of fit obtained after parameter tuning is closer to the goodness of fit under cross-validation than that before parameter tuning.

[0111] Taking the regularization coefficient of the decision tree optimization XGBOOST algorithm as an example, when the alpha and lambda values ​​in the regularization coefficient of the decision tree optimization XGBOOST algorithm are adjusted from the default value to 0.9 and 0.3, the corresponding goodness of fit is improved from 0.83 to 0.98.

[0112] In this embodiment, the method for setting the parameter tuning order specifically includes the following steps:

[0113] Step ①: Use the importance function to analyze the significance of the multivariable influence of water-voltaic devices on their electrical performance;

[0114] In this implementation, the importance function built into the decision tree optimization XGBOOST algorithm is called to analyze the importance of the multivariate influence of the water-voltaic device on the electrical performance of the water-voltaic device.

[0115] Step 2: Set the parameter tuning sequence according to the importance of the influence of multiple variables of the water-voltaic device on the electrical performance of the water-voltaic device.

[0116] In this embodiment, the parameter tuning order is as follows: number of decision trees, minimum sum of leaf weights, subsample size, learning rate, maximum depth, regularization coefficient, and leaf node loss for the decision tree optimization XGBOOST algorithm.

[0117] The preset parameter range for the number of decision trees in the XGBOOST algorithm for decision tree optimization is 1~200;

[0118] The preset parameter range for the minimum leaf weight sum of the decision tree optimization XGBOOST algorithm is 1~200;

[0119] The preset parameter range for subsample size in the XGBOOST algorithm for decision tree optimization is 0.1~1;

[0120] The preset parameter range for the learning rate of the decision tree optimization XGBOOST algorithm is 0~1;

[0121] The preset parameter range for the maximum depth of the decision tree optimization XGBOOST algorithm is 1~50;

[0122] The preset tuning parameter range for the regularization coefficient of the decision tree optimization XGBOOST algorithm is 0~5;

[0123] The preset parameter range for the leaf node loss of the decision tree optimization XGBOOST algorithm is 0~5.

[0124] In this embodiment, the decision tree optimization XGBOOST algorithm with the optimal parameter combination is used as the trained decision tree optimization XGBOOST algorithm.

[0125] Specifically, in the underfitting state, the output decision tree optimization XGBOOST algorithm has the optimal parameter combination under the optimal goodness of fit, optimal mean absolute error and optimal mean variance. At this time, the decision tree optimization XGBOOST algorithm has been tuned to the optimal state, that is, the trained decision tree optimization XGBOOST algorithm has been obtained.

[0126] By referring to the optimal parameter combination output in this embodiment, other decision tree optimization XGBOOST algorithms that have not been tuned can be optimized, enabling them to be well applied to the prediction of multivariate and electrical performance of water-voltaic devices.

[0127] Step 4: Input the multivariate parameter values ​​of the water-voltaic device into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and obtain the multivariate and electrical performance prediction results of the water-voltaic device.

[0128] In this embodiment, the multivariate parameter values ​​of the water-voltaic device are obtained, and the multivariate parameter values ​​of the water-voltaic device are input into the trained decision tree optimization XGBOOST algorithm to predict the multivariate and electrical performance of the water-voltaic device, thereby obtaining the prediction results of the multivariate and electrical performance of the water-voltaic device.

[0129] By using the decision tree-optimized XGBOOST algorithm, which has been parameter-tuned in this embodiment, to predict the multivariate and electrical performance of water-voltaic devices, the prediction results of the multivariate and electrical performance of water-voltaic devices can be obtained. This overcomes the limitations of existing prediction methods that rely on experimental methods to control a single variable, and enables machine prediction. It also solves the problems of low efficiency, cumbersome steps, high cost, and difficulty in confirming experimental conditions caused by the need to iterate experiments on a single variable in existing prediction methods, thus significantly improving the prediction efficiency of the multivariate and electrical performance of water-voltaic devices.

[0130] To verify the effectiveness of the multivariate and electrical performance prediction method for water-voltaic devices provided in this embodiment, 80,000 unknown data points were collected to form a prediction array. Under unknown conditions, the multivariate and electrical performance prediction method for water-voltaic devices provided in this embodiment was used to predict the multivariate and electrical performance of water-voltaic devices. Each data point only contains independent variable information and does not contain dependent variable information.

[0131] The XGBOOST algorithm, optimized using a decision tree with parameter tuning in this embodiment, is used to predict the multivariate and electrical performance of a prediction array consisting of 80,000 unknown data points. The prediction results for the multivariate and electrical performance of the water-voltaic device are as follows: Figure 4 As shown.

[0132] Figure 4 middle, , This indicates the temperature and salt concentration of the testing environment. This represents the zeta potential of a water-voltaic device. , , This indicates the length, width, and thickness of the water-based photovoltaic device. , , This indicates the voltage, current, and power density of the water-based photovoltaic device.

[0133] Depend on Figure 4 It is evident that, under unknown conditions, using the decision tree-optimized XGBOOST algorithm (tuned in this embodiment) to predict the multivariate and electrical performance of a prediction array consisting of 80,000 unknown data points can yield the predicted results. Compared to existing prediction methods, this approach overcomes the limitations of single-variable control through experimental means, eliminating the need for iterative experiments based on a single variable, thereby improving the prediction efficiency of the multivariate and electrical performance of water-voltaic devices.

[0134] This invention provides a device for predicting the multivariable and electrical performance of water-based photovoltaic devices, comprising:

[0135] Data acquisition module: used to acquire multivariable parameter values ​​of water-based photovoltaic devices;

[0136] The training set construction module is used to construct a multivariable and electrical performance training set for water-voltaic devices;

[0137] The training module is used to train the decision tree optimization XGBOOST algorithm using a multivariable and electrical performance training set of water-voltaic devices. During the training process, the decision tree optimization XGBOOST algorithm is tuned according to a preset parameter tuning order to obtain the trained decision tree optimization XGBOOST algorithm.

[0138] Prediction module: This module is used to input the multivariate parameter values ​​of the water-voltaic device into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and obtain the prediction results of the multivariate and electrical performance of the water-voltaic device.

[0139] The multivariable and electrical performance prediction device for water-based photovoltaic devices provided in this embodiment of the invention can execute the multivariable and electrical performance prediction method for water-based photovoltaic devices provided in this embodiment of the invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0140] This invention provides a computer system, comprising:

[0141] Storage medium: used to store computer programs;

[0142] Processor: Used to execute computer programs to implement the multivariable and electrical performance prediction method for water-voltaic devices provided in the embodiments of the present invention.

[0143] This invention provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the multivariable and electrical performance prediction method for water-voltaic devices provided in this invention.

[0144] This invention provides a computer program product, including a computer program that, when executed by a processor, implements the multivariable and electrical performance prediction method for water-voltaic devices provided in this invention.

[0145] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0147] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0148] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0149] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for predicting the multivariable and electrical performance of water-based photovoltaic devices, characterized in that, include: Obtain multivariable parameter values ​​for water-based photovoltaic devices; The multivariate parameter values ​​of the water-voltaic device are input into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and the prediction results of multivariate and electrical performance of the water-voltaic device are obtained. The training methods for the decision tree optimization XGBOOST algorithm include: Construct a training set for the multivariable and electrical performance of water-based photovoltaic devices; The decision tree optimization XGBOOST algorithm was trained using a multivariable and electrical performance training set of water-based photovoltaic devices. During the training process, the XGBOOST algorithm was tuned according to a preset parameter tuning order to obtain a trained decision tree optimization XGBOOST algorithm, including: Standardize the training set of multivariable and electrical performance of water-voltaic devices; The material parameters, dimensional parameters, and test environment parameters of the water-voltaic devices in the standardized multivariable and electrical performance training set were set as independent variables. The electrical parameters of the water-voltaic devices in the training set after standardization were set as dependent variables. The data corresponding to the independent variables are input into the decision tree optimization XGBOOST algorithm for training. Through cross-validation, the predicted fitting curves of the independent and dependent variables are obtained, as well as the goodness of fit, mean absolute error and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves. During the training process, based on the preset parameter tuning order, within the preset parameter tuning range, the XGBOOST decision tree optimization algorithm is tuned to underfitting with the goals of optimal goodness of fit, optimal mean absolute error and optimal mean variance, so as to obtain the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error and optimal mean variance. The decision tree optimization XGBOOST algorithm with the optimal parameter combination is used as the trained decision tree optimization XGBOOST algorithm. The methods for setting the parameter tuning order include: The importance function is used to analyze the influence of multiple variables on the electrical performance of water-voltaic devices. The parameter tuning order is set according to the importance of the influence of multiple variables on the electrical performance of the water-voltaic device; The parameter settings are configured in the following order: number of decision trees, minimum sum of leaf weights, subsample size, learning rate, maximum depth, regularization coefficient, and leaf node loss for the XGBOOST decision tree optimization algorithm.

2. The method for predicting the multivariable and electrical performance of water-based photovoltaic devices according to claim 1, characterized in that, The training set for constructing multivariable and electrical performance of water-based photovoltaic devices includes: The apparent dimensions and electrical performance of the water-voltaic device were tested to obtain multivariable and electrical performance data of the water-voltaic device; The multivariable and electrical performance data of water-voltaic devices are extended to obtain a training set of multivariable and electrical performance data of water-voltaic devices; Among them, the multivariables include the material parameters, size parameters and test environment parameters of the water-voltaic device. The material parameters include the zeta potential of the water-voltaic device, the size parameters include the length, width and thickness of the water-voltaic device, and the test environment parameters include the temperature and salt concentration of the test environment. Electrical performance includes the electrical parameters of the water-volt device, which include the voltage, current, and power density of the water-volt device. Each data point in the training set of multivariable and electrical performance data for water-voltaic devices includes the zeta potential, length, width, thickness, voltage, current, and power density of the water-voltaic device, as well as the temperature and salt concentration of the test environment.

3. The method for predicting the multivariable and electrical performance of water-based photovoltaic devices according to claim 1, characterized in that, During the training process, based on the preset parameter tuning order, the cyclic comparison method is used to tune the XGBOOST decision tree optimization algorithm to underfit within the preset parameter tuning range, with the goals of optimal goodness of fit, optimal mean absolute error, and optimal mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

4. The method for predicting the multivariable and electrical performance of water-based photovoltaic devices according to claim 1, characterized in that, During the training process, based on the preset parameter tuning order, the XGBOOST decision tree optimization algorithm is tuned to underfitting within the preset parameter tuning range using the traversal method, with the goals of optimal goodness of fit, optimal mean absolute error, and optimal mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

5. The method for predicting the multivariable and electrical performance of water-based photovoltaic devices according to claim 1, characterized in that, During the training process, based on the preset parameter tuning order, the grid search method is used to tune the XGBOOST decision tree optimization algorithm to underfit within the preset parameter tuning range, with the goals of optimal goodness of fit, optimal mean absolute error, and optimal mean variance. This yields the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error, and optimal mean variance.

6. The method for predicting the multivariable and electrical performance of water-based photovoltaic devices according to claim 1, characterized in that, The preset parameter range for the number of decision trees in the XGBOOST algorithm for decision tree optimization is 1~200; The preset parameter range for the minimum leaf weight sum of the decision tree optimization XGBOOST algorithm is 1~200; The preset parameter range for subsample size in the XGBOOST algorithm for decision tree optimization is 0.1~1; The preset parameter range for the learning rate of the decision tree optimization XGBOOST algorithm is 0~1; The preset parameter range for the maximum depth of the decision tree optimization XGBOOST algorithm is 1~50; The preset tuning parameter range for the regularization coefficient of the decision tree optimization XGBOOST algorithm is 0~5; The preset parameter range for the leaf node loss of the decision tree optimization XGBOOST algorithm is 0~5.

7. A device for predicting the multivariable and electrical performance of a water-based photovoltaic device, characterized in that, include: Data acquisition module: used to acquire multivariable parameter values ​​of water-based photovoltaic devices; The training set construction module is used to construct a multivariable and electrical performance training set for water-voltaic devices; The training module is used to train the decision tree optimization XGBOOST algorithm using a multivariate and electrical performance training set of hydrovoltaic devices. During the training process, the XGBOOST algorithm is tuned according to a preset parameter tuning order to obtain a trained XGBOOST algorithm, including: Standardize the training set of multivariable and electrical performance of water-voltaic devices; The material parameters, dimensional parameters, and test environment parameters of the water-voltaic devices in the standardized multivariable and electrical performance training set were set as independent variables. The electrical parameters of the water-voltaic devices in the training set after standardization were set as dependent variables. The data corresponding to the independent variables are input into the decision tree optimization XGBOOST algorithm for training. Through cross-validation, the predicted fitting curves of the independent and dependent variables are obtained, as well as the goodness of fit, mean absolute error and mean variance of the predicted fitting curves of the independent and dependent variables compared with the actual fitting curves. During the training process, based on the preset parameter tuning order, within the preset parameter tuning range, the XGBOOST decision tree optimization algorithm is tuned to underfitting with the goals of optimal goodness of fit, optimal mean absolute error and optimal mean variance, so as to obtain the optimal parameter combination of the XGBOOST decision tree optimization algorithm under optimal goodness of fit, optimal mean absolute error and optimal mean variance. The decision tree optimization XGBOOST algorithm with the optimal parameter combination is used as the trained decision tree optimization XGBOOST algorithm. The methods for setting the parameter tuning order include: The importance function is used to analyze the influence of multiple variables on the electrical performance of water-voltaic devices. The parameter tuning order is set according to the importance of the influence of multiple variables on the electrical performance of the water-voltaic device; The parameter tuning order is as follows: number of decision trees, minimum sum of leaf weights, subsample size, learning rate, maximum depth, regularization coefficient, and leaf node loss for the XGBOOST decision tree optimization algorithm. Prediction module: This module is used to input the multivariate parameter values ​​of the water-voltaic device into the trained decision tree to optimize the XGBOOST algorithm for multivariate and electrical performance prediction of the water-voltaic device, and obtain the prediction results of the multivariate and electrical performance of the water-voltaic device.

8. A computer system, characterized in that, include: Storage medium: used to store computer programs; Processor: Used to execute the computer program to implement the multivariable and electrical performance prediction method for water-voltaic devices according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multivariable and electrical performance prediction method for water-voltaic devices as described in any one of claims 1 to 6.

Images