A photovoltaic power generation system parameter identification method and system based on a swarm intelligence optimization algorithm

Abstract

The application discloses a photovoltaic power generation system parameter identification method and system based on a swarm intelligence optimization algorithm, and the method comprises the following steps: obtaining historical operation data of a photovoltaic power generation system, establishing a data set containing environmental factor influence, and constructing a photovoltaic power generation capacity prediction model; a particle swarm optimization algorithm is used to initialize a parameter search space, and the output power of the photovoltaic system is simulated and calculated; a nonlinear mapping relationship between the output power of the photovoltaic system and the environmental factor is established, and the predicted value of the system output power is calculated; the importance of each parameter to the system output power is evaluated through a random forest algorithm, and the key parameters are screened out; the optimized parameter values are substituted into the photovoltaic power generation capacity prediction model, the system output power is calculated, the output power is compared with the measured value, and if the error between the two is within the allowable range, the final parameter identification result is output. The application improves the accuracy and efficiency of photovoltaic power generation system parameter identification.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a photovoltaic power generation system parameter identification method and system based on a swarm intelligence optimization algorithm. Background Art

[0002] In photovoltaic power generation systems, photovoltaic cells and inverters are two key components, and their performance directly impacts the overall system's power generation efficiency. However, the actual operating parameters of these components often deviate from their theoretical design values, resulting in suboptimal system performance. To accurately describe the actual operating state of a photovoltaic power generation system, it is necessary to establish a precise mathematical model and identify the key parameters within the model. Traditional parameter identification methods, such as the least squares method, often perform poorly when dealing with nonlinear and multimodal problems, and are prone to falling into local optima. The strong nonlinearity and multivariable coupling of photovoltaic power generation systems complicate parameter identification. Furthermore, photovoltaic power generation systems are affected by environmental factors such as temperature and irradiance. Changes in these factors can cause system parameters to drift, further complicating parameter identification. Therefore, accurately identifying the key parameters of photovoltaic power generation systems under complex environmental conditions is an urgent technical challenge. This requires an optimization algorithm that can perform a global search in a high-dimensional, nonlinear space and exhibit a certain degree of robustness to adapt to changing environmental factors. Summary of the Invention

[0003] In order to solve the technical problems existing in the above-mentioned prior art, the present invention proposes a photovoltaic power generation system parameter identification method and system based on a swarm intelligence optimization algorithm to improve the accuracy and efficiency of photovoltaic power generation system parameter identification.

[0004] On the one hand, to achieve the above-mentioned purpose, the present invention provides a photovoltaic power generation system parameter identification method based on a swarm intelligence optimization algorithm, comprising:

[0005] Acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors, and construct a photovoltaic power generation prediction model based on the data set including the influence of environmental factors;

[0006] Initializing a parameter search space using a particle swarm optimization algorithm, setting hyperparameters, inputting the hyperparameters into the photovoltaic power generation prediction model, and simulating and calculating the output power of the photovoltaic system;

[0007] A nonlinear mapping relationship between the photovoltaic system output power and environmental factors is established through a support vector regression model. The parameter estimates obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted.

[0008] The importance of each parameter to the system output power is evaluated by the random forest algorithm, and key parameters are screened out. If the importance of a parameter among the key parameters is lower than a preset value, the key parameter is eliminated from the optimization process;

[0009] The optimized parameter values ​​are substituted into the photovoltaic power generation prediction model to calculate the system output power, and the output power is compared with the measured value. If the error between the two is within the allowable range, the final parameter identification result is output.

[0010] On the other hand, to achieve the above-mentioned purpose, the present invention also provides a photovoltaic power generation system parameter identification system based on a swarm intelligence optimization algorithm, comprising:

[0011] a data acquisition module, configured to acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors based on the historical operating data, and construct a photovoltaic power generation prediction model using the data set including the influence of environmental factors;

[0012] The parameter initialization module is used to initialize the parameter search space using the particle swarm optimization algorithm, set hyperparameters such as the particle swarm size, number of iterations, and learning factor, and define the mean square error between the system output power and the measured power as the fitness function;

[0013] The model building module uses a support vector regression model to establish a nonlinear mapping relationship between the photovoltaic system output power and environmental factors. The parameter estimation values ​​obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted.

[0014] A parameter evaluation module is used to evaluate the importance of each parameter to the system output power through a random forest algorithm, screen out key parameters, and if the importance of a parameter among the key parameters is lower than a preset value, eliminate the key parameter from the optimization process;

[0015] The result output module is used to substitute the optimized parameter values ​​into the photovoltaic power generation prediction model, calculate the system output power, compare the output power with the measured value, and output the final parameter identification result if the error between the two is within the allowable range.

[0016] Compared with the prior art, the present invention has the following advantages and technical effects:

[0017] The method of the present invention acquires historical operating data to establish a dataset that includes the influence of environmental factors. It then uses a particle swarm optimization algorithm to initialize the parameter search space, and combines it with a support vector regression algorithm to establish a nonlinear mapping relationship between output power and environmental factors. Through an iterative optimization process, the particle swarm search strategy is continuously adjusted, while a random forest algorithm is used to assess parameter importance and screen key parameters, ultimately obtaining optimized parameter values. These parameters are then substituted into a mathematical model, the output power is calculated, and compared with the measured values. The final parameter identification results are output within the allowable error range. This method improves the accuracy and efficiency of photovoltaic power generation system parameter identification, providing a reliable basis for system performance optimization and fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0019] Figure 1 This is a flow chart of a photovoltaic power generation system parameter identification method based on a swarm intelligence optimization algorithm according to an embodiment of the present invention;

[0020] Figure 2 This is a structural diagram of a photovoltaic power generation system parameter identification system based on a swarm intelligence optimization algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0022] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0023] This embodiment proposes a photovoltaic power generation system parameter identification method based on a swarm intelligence optimization algorithm. Figure 1 ,include:

[0024] Acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors, and construct a photovoltaic power generation prediction model based on the data set including the influence of environmental factors;

[0025] Initializing a parameter search space using a particle swarm optimization algorithm, setting hyperparameters, inputting the hyperparameters into the photovoltaic power generation prediction model, and simulating and calculating the output power of the photovoltaic system;

[0026] A nonlinear mapping relationship between the photovoltaic system output power and environmental factors is established through a support vector regression model. The parameter estimates obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted.

[0027] The importance of each parameter to the system output power is evaluated by the random forest algorithm, and key parameters are screened out. If the importance of a parameter among the key parameters is lower than a preset value, the key parameter is eliminated from the optimization process;

[0028] The optimized parameter values ​​are substituted into the photovoltaic power generation prediction model to calculate the system output power, and the output power is compared with the measured value. If the error between the two is within the allowable range, the final parameter identification result is output.

[0029] Specifically, this embodiment acquires historical operating data to establish a dataset that includes environmental factors. A particle swarm optimization algorithm is used to initialize the parameter search space, and a support vector regression algorithm is combined to establish a nonlinear mapping relationship between output power and environmental factors. Through an iterative optimization process, the particle swarm search strategy is continuously adjusted, while a random forest algorithm is used to assess parameter importance and screen key parameters, ultimately obtaining optimized parameter values. These parameters are then substituted into a photovoltaic power generation prediction model, and the output power is calculated and compared with the measured value. The final parameter identification results are then output within the allowable error range.

[0030] Furthermore, establishing a data set that includes the impact of environmental factors includes:

[0031] Obtain historical operating data of the photovoltaic power generation system, using data acquisition equipment to collect and store voltage, current, temperature and irradiance parameters in real time to form a data set;

[0032] Preprocess the collected historical operating parameter data, including cleaning and normalization, removing outliers, and converting the data into a format suitable for modeling;

[0033] Based on the pre-processed historical operating parameter data, key characteristic parameters reflecting the operating status and environmental factors of the photovoltaic power generation system are extracted;

[0034] According to the key characteristic parameters, a correlation analysis method is adopted to analyze the correlation between the characteristic parameters, and a characteristic parameter subset is screened, that is, the data set containing the influence of environmental factors.

[0035] Specifically, historical operating data of the photovoltaic power generation system is collected in real time using data acquisition equipment for parameters such as voltage, current, temperature, and irradiance, and stored to form a data set. Data preprocessing techniques are used to clean and normalize the collected historical operating parameter data, remove outliers, and convert the data into a format suitable for modeling. Feature engineering techniques are then used to extract key characteristic parameters reflecting the operating status and environmental factors of the photovoltaic power generation system. Correlation analysis is then used to analyze the correlations between the extracted characteristic parameters, identifying a subset of characteristic parameters that are closely related to photovoltaic power generation. Machine learning algorithms, such as support vector machines and random forests, are then used to develop a photovoltaic power generation prediction model that incorporates environmental factors. This prediction model, fed with the real-time collected operating and environmental parameters of the photovoltaic power generation system, predicts power generation over a specific period of time, providing a basis for optimal control of the power generation system. Based on the predicted power generation and the actual operating status of the photovoltaic power generation system, an intelligent optimization algorithm is employed to dynamically optimize and adjust the system's operating parameters to improve power generation efficiency and ensure stable system operation.

[0036] Furthermore, the particle swarm optimization algorithm is used to initialize the parameter search space, and the hyperparameters are set including:

[0037] Initialize the particle swarm according to the preset particle swarm size, where each particle represents a set of algorithm parameter combinations to be optimized;

[0038] For each particle, the parameters carried by the particle are input into the system model, and the system output power under these parameters is obtained through simulation and calculation;

[0039] Obtaining measured power data under the same parameters, and calculating the mean square error between the system output power and the measured power as the fitness value of the particle;

[0040] Determine whether the current number of iterations has reached the preset maximum number of iterations. If not, update the particle speed and position according to the learning factor and enter the next round of iterative optimization;

[0041] If the maximum number of iterations has been reached, the particle with the best fitness value is selected from the particle swarm, and the parameters carried by the particle with the best fitness value are used as the optimal parameter combination;

[0042] The optimal parameter combination is input into the system model, and the optimized system output power is calculated through simulation. By comparing the mean square error of the system output power before and after optimization with the measured power, the effectiveness of the optimization algorithm is evaluated to determine whether further hyperparameter adjustments and re-optimization are needed.

[0043] Specifically, in photovoltaic power generation system parameter optimization, each particle represents a set of algorithm parameter combinations to be optimized, such as inverter efficiency and maximum power point tracking (MPPT) algorithm parameters. During initialization, in this embodiment, the particle swarm size is set to 50, and each particle is randomly distributed within the parameter space. For each particle, the parameters it carries are input into the system model for simulation. For example, a particle carries the following parameter combination: 95% inverter efficiency and 0.5V voltage step size for the MPPT algorithm. Simulation calculations show that the system output power under these parameters is 10kW. Simultaneously, measured power data for an actual photovoltaic system under the same conditions is obtained, assuming it is 9.8kW. The mean squared error between the simulated output power and the measured power is calculated, yielding a value of 0.04 as the particle's fitness value. During the iterative optimization process, the maximum number of iterations is set to 100. In each iteration, the particle's velocity and position are updated based on a learning factor. Learning factors typically include individual and social learning factors, which control the particle's tendency to move toward the individual and global optimal positions, respectively.

[0044] If the mean squared error (MSE) after optimization is significantly reduced, the optimization is successful. If the results are unsatisfactory, consider adjusting hyperparameters, such as increasing the particle swarm size, increasing the number of iterations, or adjusting the learning factor, and then re-optimizing. This optimization method's advantage lies in its ability to quickly find near-optimal solutions within complex parameter spaces, making it suitable for nonlinear, multivariable optimization problems such as photovoltaic power generation systems. Through continuous iteration and optimization, the system's power generation efficiency and stability can be significantly improved, enabling intelligent control and management of photovoltaic power generation systems.

[0045] Furthermore, establishing a nonlinear mapping relationship between the output power of the photovoltaic system and environmental factors includes:

[0046] Obtain a historical data set containing PV system output power, ambient temperature, and irradiance data, preprocess the data, remove outliers and missing values, and normalize the data;

[0047] Based on the preprocessed data set, a support vector regression model was used to construct a nonlinear regression model with ambient temperature and irradiance as input features and photovoltaic system output power as output target. The hyperparameters of the model were optimized through cross-validation.

[0048] The trained support vector regression model is applied to new ambient temperature and irradiance data to predict the output power of the photovoltaic system under the new environmental conditions.

[0049] Specifically, a historical dataset containing data on PV system output power, ambient temperature, and irradiance is obtained. The data is preprocessed to remove outliers and missing values ​​and normalize the data. Based on the preprocessed dataset, a nonlinear regression model is constructed using the support vector regression algorithm, with ambient temperature and irradiance as input features and the PV system output power as the output target. During the training of the support vector regression model, the model's hyperparameters, including the kernel function type, penalty coefficient, and tolerance error, are optimized through cross-validation to improve the model's generalization performance. The trained support vector regression model is applied to new ambient temperature and irradiance data to predict the PV system's output power under these conditions. The predicted results are compared with the actual output power to evaluate the model's prediction accuracy. If the model's prediction accuracy does not meet the preset threshold, the hyperparameters of the support vector regression model are adjusted and the model is retrained. Otherwise, the trained model is saved for subsequent PV system output power prediction. Forecasted ambient temperature and irradiance data for the region where the PV system is located for a certain period of time are obtained and input into the trained support vector regression model to obtain the predicted output power of the PV system for that period. According to the predicted value of the photovoltaic system output power, the operation strategy of the photovoltaic system is optimized, including battery charge and discharge control, load scheduling, etc., to improve the energy utilization efficiency and economic benefits of the photovoltaic system.

[0050] Furthermore, the parameter estimation value obtained by the particle swarm optimization algorithm is input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted, including:

[0051] Obtain a pre-established support vector regression model and a preset error threshold;

[0052] For the system to be predicted, the particle swarm optimization algorithm is used to estimate the parameters and obtain the parameter estimation value;

[0053] Inputting the parameter estimation value into the support vector regression model to calculate the predicted value of the system output power;

[0054] Obtain the measured value of the system output power and calculate the error between the predicted value and the measured value;

[0055] Determine whether the error exceeds the preset threshold. If so, adjust the search strategy of the particle swarm algorithm.

[0056] According to the adjusted search strategy, the particle swarm optimization algorithm is re-executed to obtain new parameter estimates;

[0057] The new parameter estimation value is input into the support vector regression model, and the updated power prediction value is calculated until the prediction value meets the error threshold requirement.

[0058] Specifically, in this embodiment, particles are defined as a set of parameters of the support vector regression model, such as kernel function parameters, penalty factors, etc. Each particle represents a set of possible parameter values, and the optimal parameter combination is found through iterative optimization. Assuming that the initial particle swarm contains 50 particles, each particle contains 5 parameters, and the optimal parameter estimation value is obtained after 100 iterations. After the estimated parameter values ​​are input into the support vector regression model, the output power under specific environmental conditions can be predicted. For example, using the optimized parameters, the model predicts that the output power under specific environmental conditions is 30°C and the irradiance is 900W / m 2 When the system output power is 550 kW, the predicted value needs to be compared with the measured value. Assuming the actual measured output power is 545 kW, the prediction error is 5 kW. The setting of the preset error threshold depends on the specific application requirements. In this embodiment, the threshold is set to 10 kW or 2% of the output power, whichever is smaller. If the prediction error exceeds the threshold, the particle swarm optimization algorithm's search strategy needs to be adjusted. Adjustment methods include increasing the number of particles, modifying the inertia weight, or learning factor. For example, increasing the number of particles from 50 to 100 or adjusting the global learning factor from 2.0 to 2.2 can enhance the algorithm's global search capability. After adjusting the search strategy, the particle swarm optimization algorithm is re-executed. Assume that after these adjustments, the new parameter estimates result in a predicted output power of 547 kW, reducing the error from the measured value of 545 kW to 2 kW, meeting the preset error requirement. This iterative optimization process not only improves prediction accuracy but also enhances the model's adaptability to varying environmental conditions. This method, combining support vector regression and particle swarm optimization, has many advantages. First, it can handle nonlinear relationships in PV systems, improving prediction accuracy. Second, by dynamically adjusting parameters, the model can adapt to changes in system performance over time, such as efficiency fluctuations caused by equipment aging or seasonal changes. Finally, high-precision power forecasting helps optimize PV system operating strategies, such as battery charge and discharge management and load scheduling, thereby improving energy efficiency and economic benefits.

[0059] Furthermore, the importance of each parameter to the system output power is evaluated using the random forest algorithm, including:

[0060] Construct training and test datasets using different input parameters and corresponding output power data;

[0061] The random forest algorithm is used to train the training data set to obtain the random forest model, and the model is evaluated using the test data set to calculate the prediction accuracy of the model;

[0062] Using the random forest model to evaluate the importance of different input parameters, and obtain the importance value of each parameter;

[0063] According to a preset importance threshold, parameters whose importance values ​​are greater than or equal to the importance threshold are determined as key parameters;

[0064] Determine whether there is a parameter in the key parameters whose importance value is lower than the importance threshold; if so, remove the parameter from the key parameter list;

[0065] Based on the key parameters obtained by screening, an optimization model is constructed, and the key parameters are optimized and solved using a genetic algorithm to obtain the optimal values ​​of each key parameter;

[0066] The optimal values ​​of the key parameters are input into the photovoltaic power generation prediction model to obtain the output power of the system, and the output power is compared with the output power before optimization to evaluate the optimization effect.

[0067] Furthermore, genetic algorithms are used to optimize and solve key parameters, and the optimal values ​​of each key parameter are obtained, including:

[0068] Determine the optimization objective function, incorporate the screened key parameters into the objective function as optimization variables, set the value range and constraint conditions of the optimization variables, and build a complete optimization model;

[0069] Convert the constructed optimization model into a form that can be solved by genetic algorithm and determine the various parameter settings of genetic algorithm;

[0070] The initial population is randomly generated, each individual corresponds to a set of optimized variable values, and the individual fitness is calculated according to the optimization objective function;

[0071] Perform selection, crossover, and mutation genetic operations on the current population to generate new offspring populations and continuously iterate and optimize;

[0072] When the iteration termination condition is met, the individual with the highest fitness is selected from the final population, and the value of the optimization variable corresponding to the individual with the highest fitness is the optimal solution;

[0073] The optimal solution obtained by the genetic algorithm is used as the optimal value of the key parameters.

[0074] Specifically, the selection operation can adopt the roulette wheel method, and the individuals with higher output power have a greater probability of being selected. The crossover operation can use arithmetic crossover, such as taking the average of the two parent generation inclination angles as the offspring inclination angle. The mutation operation can randomly adjust the parameters in small increments near the current value, such as increasing or decreasing the inclination angle by 1-2 degrees. The iteration termination condition can be reaching the maximum number of iterations (such as 100 generations) or there is no significant improvement in the optimal solution for multiple consecutive generations. The final optimal solution may be an inclination of 32° and an azimuth of 175°, at which point the system output power reaches the maximum. The advantage of this optimization method is that it can quickly find an approximate optimal solution in a complex parameter space, and is particularly suitable for nonlinear and multi-peak optimization problems. The optimal parameters obtained by the genetic algorithm can directly guide the installation of solar panels and improve the overall efficiency of the system. At the same time, this method can also be applied to other fields, such as the layout optimization of wind turbines and parameter tuning of chemical production processes, reflecting its wide applicability and practical value.

[0075] Furthermore, the final parameter identification results are output including:

[0076] According to the optimized parameter values, the photovoltaic power generation prediction model is used to calculate the output power estimation value of the photovoltaic system;

[0077] Obtain the measured output power data of the photovoltaic system under the same conditions as a reference value;

[0078] Determine whether the error between the estimated output power and the measured output power is within a preset allowable range. If so, proceed to the next step; otherwise, return to the first step to re-optimize the parameters.

[0079] Based on the comparison results of the estimated output power and the measured output power, it is determined whether the current optimized parameter value is reliable. If it is reliable, it is output as the final parameter identification result.

[0080] This embodiment also provides a photovoltaic power generation system parameter identification system based on swarm intelligence optimization algorithm, such as Figure 2 ,include:

[0081] a data acquisition module, configured to acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors based on the historical operating data, and construct a photovoltaic power generation prediction model using the data set including the influence of environmental factors;

[0082] The parameter initialization module is used to initialize the parameter search space using the particle swarm optimization algorithm, set hyperparameters such as the particle swarm size, number of iterations, and learning factor, and define the mean square error between the system output power and the measured power as the fitness function;

[0083] The model building module uses a support vector regression model to establish a nonlinear mapping relationship between the photovoltaic system output power and environmental factors. The parameter estimation values ​​obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted.

[0084] A parameter evaluation module is used to evaluate the importance of each parameter to the system output power through a random forest algorithm, screen out key parameters, and if the importance of a parameter among the key parameters is lower than a preset value, eliminate the key parameter from the optimization process;

[0085] The result output module is used to substitute the optimized parameter values ​​into the photovoltaic power generation prediction model, calculate the system output power, compare the output power with the measured value, and output the final parameter identification result if the error between the two is within the allowable range.

[0086] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm, characterized in that: include: Acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors, and construct a photovoltaic power generation prediction model based on the data set including the influence of environmental factors; Initializing a parameter search space using a particle swarm optimization algorithm, setting hyperparameters, inputting the hyperparameters into the photovoltaic power generation prediction model, and simulating and calculating the output power of the photovoltaic system; A nonlinear mapping relationship between the photovoltaic system output power and environmental factors is established through a support vector regression model. The parameter estimates obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted. The importance of each parameter to the system output power is evaluated by the random forest algorithm, and key parameters are screened out. If the importance of a parameter among the key parameters is lower than a preset value, the key parameter is eliminated from the optimization process; Substituting the optimized parameter values ​​into the photovoltaic power generation prediction model, calculating the system output power, and comparing the output power with the measured value. If the error between the two is within the allowable range, outputting the final parameter identification result; The parameter estimation value obtained by the particle swarm optimization algorithm is input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted, including: Obtain a pre-established support vector regression model and a preset error threshold; For the system to be predicted, the particle swarm optimization algorithm is used to estimate the parameters and obtain the parameter estimation value; Inputting the parameter estimation value into the support vector regression model to calculate the predicted value of the system output power; Obtain the measured value of the system output power and calculate the error between the predicted value and the measured value; Determine whether the error exceeds the preset threshold. If so, adjust the search strategy of the particle swarm algorithm. According to the adjusted search strategy, the particle swarm optimization algorithm is re-executed to obtain new parameter estimates; Inputting the new parameter estimation value into the support vector regression model and calculating the updated power prediction value until the prediction value meets the error threshold requirement; The importance of each parameter to the system output power is evaluated using the random forest algorithm, including: Construct training and test datasets using different input parameters and corresponding output power data; The random forest algorithm is used to train the training data set to obtain the random forest model, and the model is evaluated using the test data set to calculate the prediction accuracy of the model; Using the random forest model to evaluate the importance of different input parameters, and obtain the importance value of each parameter; According to a preset importance threshold, parameters whose importance values ​​are greater than or equal to the importance threshold are determined as key parameters; Determine whether there is a parameter in the key parameters whose importance value is lower than the importance threshold; if so, remove the parameter from the key parameter list; Based on the key parameters obtained by screening, an optimization model is constructed, and the key parameters are optimized and solved using a genetic algorithm to obtain the optimal values ​​of each key parameter; The optimal values ​​of the key parameters are input into the photovoltaic power generation prediction model to obtain the output power of the system, and the output power is compared with the output power before optimization to evaluate the optimization effect.

2. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 1 is characterized in that: Establishing the data set including the impact of environmental factors includes: Obtain historical operating data of the photovoltaic power generation system, using data acquisition equipment to collect and store voltage, current, temperature and irradiance parameters in real time to form a data set; Preprocess the collected historical operating parameter data, including cleaning and normalization, removing outliers, and converting the data into a format suitable for modeling; Based on the pre-processed historical operating parameter data, key characteristic parameters reflecting the operating status and environmental factors of the photovoltaic power generation system are extracted; According to the key characteristic parameters, a correlation analysis method is adopted to analyze the correlation between the characteristic parameters, and a characteristic parameter subset is screened, that is, the data set containing the influence of environmental factors.

3. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 1 is characterized in that: The particle swarm optimization algorithm is used to initialize the parameter search space, and the hyperparameters are set as follows: Initialize the particle swarm according to the preset particle swarm size, where each particle represents a set of algorithm parameter combinations to be optimized; For each particle, the parameters carried by the particle are input into the system model, and the system output power under these parameters is obtained through simulation and calculation; Obtaining measured power data under the same parameters, and calculating the mean square error between the system output power and the measured power as the fitness value of the particle; Determine whether the current number of iterations has reached the preset maximum number of iterations. If not, update the particle speed and position according to the learning factor and enter the next round of iterative optimization; If the maximum number of iterations has been reached, a particle with the best fitness value is selected from the particle swarm, and the parameters carried by the particle with the best fitness value are used as the optimal parameter combination.

4. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 3 is characterized in that: Inputting the optimal parameter combination into the photovoltaic power generation prediction model includes: By comparing the mean square error between the system output power before and after optimization and the measured power, the effect of the optimization algorithm can be evaluated to determine whether hyperparameters need to be adjusted and re-optimized.

5. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 1 is characterized in that: Establishing a nonlinear mapping relationship between the photovoltaic system output power and environmental factors includes: Obtain a historical data set containing PV system output power, ambient temperature, and irradiance data, preprocess the data, remove outliers and missing values, and normalize the data; Based on the preprocessed data set, a support vector regression model was used to construct a nonlinear regression model with ambient temperature and irradiance as input features and photovoltaic system output power as output target. The hyperparameters of the model were optimized through cross-validation. The trained support vector regression model is applied to the new ambient temperature and irradiance data to predict the output power of the photovoltaic system under the new environmental conditions.

6. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 1 is characterized in that: The key parameters are optimized and solved using a genetic algorithm to obtain the optimal values ​​of each key parameter, including: Determine the optimization objective function, incorporate the screened key parameters into the objective function as optimization variables, set the value range and constraint conditions of the optimization variables, and build a complete optimization model; Convert the constructed optimization model into a form that can be solved by genetic algorithm and determine the various parameter settings of genetic algorithm; The initial population is randomly generated, each individual corresponds to a set of optimized variable values, and the individual fitness is calculated according to the optimization objective function; Perform selection, crossover, and mutation genetic operations on the current population to generate new offspring populations and continuously iterate and optimize; When the iteration termination condition is met, the individual with the highest fitness is selected from the final population, and the value of the optimization variable corresponding to the individual with the highest fitness is the optimal solution; The optimal solution obtained by the genetic algorithm is used as the optimal value of the key parameters.

7. The photovoltaic power generation system parameter identification method based on swarm intelligence optimization algorithm according to claim 1 is characterized in that: The final parameter identification results are output as follows: According to the optimized parameter values, the photovoltaic power generation prediction model is used to calculate the output power estimation value of the photovoltaic system; Obtain the measured output power data of the photovoltaic system under the same conditions as a reference value; Determine whether the error between the estimated output power and the measured output power is within a preset allowable range. If so, proceed to the next step; otherwise, return to the first step to re-optimize the parameters. Based on the comparison results of the estimated output power and the measured output power, it is determined whether the current optimized parameter value is reliable. If it is reliable, it is output as the final parameter identification result.

8. A photovoltaic power generation system parameter identification system based on swarm intelligence optimization algorithm, characterized in that: include: a data acquisition module, configured to acquire historical operating data of the photovoltaic power generation system, establish a data set including the influence of environmental factors based on the historical operating data, and construct a photovoltaic power generation prediction model using the data set including the influence of environmental factors; The parameter initialization module is used to initialize the parameter search space using the particle swarm optimization algorithm, set hyperparameters such as the particle swarm size, number of iterations, and learning factor, and define the mean square error between the system output power and the measured power as the fitness function; The model building module uses a support vector regression model to establish a nonlinear mapping relationship between the photovoltaic system output power and environmental factors. The parameter estimation values ​​obtained by the particle swarm optimization algorithm are input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted. A parameter evaluation module is used to evaluate the importance of each parameter to the system output power through a random forest algorithm, screen out key parameters, and if the importance of a parameter among the key parameters is lower than a preset value, eliminate the key parameter from the optimization process; A result output module is used to substitute the optimized parameter values ​​into the photovoltaic power generation prediction model, calculate the system output power, compare the output power with the measured value, and output the final parameter identification result if the error between the two is within the allowable range; The parameter estimation value obtained by the particle swarm optimization algorithm is input into the support vector regression model to calculate the system output power prediction value. If the error between the predicted value and the measured value exceeds a preset threshold, the particle swarm search strategy is readjusted, including: Obtain a pre-established support vector regression model and a preset error threshold; For the system to be predicted, the particle swarm optimization algorithm is used to estimate the parameters and obtain the parameter estimation value; Inputting the parameter estimation value into the support vector regression model to calculate the predicted value of the system output power; Obtain the measured value of the system output power and calculate the error between the predicted value and the measured value; Determine whether the error exceeds the preset threshold. If so, adjust the search strategy of the particle swarm algorithm. According to the adjusted search strategy, the particle swarm optimization algorithm is re-executed to obtain new parameter estimates; Inputting the new parameter estimation value into the support vector regression model and calculating the updated power prediction value until the prediction value meets the error threshold requirement; The importance of each parameter to the system output power is evaluated using the random forest algorithm, including: Construct training and test datasets using different input parameters and corresponding output power data; The random forest algorithm is used to train the training data set to obtain the random forest model, and the model is evaluated using the test data set to calculate the prediction accuracy of the model; Using the random forest model to evaluate the importance of different input parameters, and obtain the importance value of each parameter; According to a preset importance threshold, parameters whose importance values ​​are greater than or equal to the importance threshold are determined as key parameters; Determine whether there is a parameter in the key parameters whose importance value is lower than the importance threshold; if so, remove the parameter from the key parameter list; Based on the key parameters obtained by screening, an optimization model is constructed, and the key parameters are optimized and solved using a genetic algorithm to obtain the optimal values ​​of each key parameter; The optimal values ​​of the key parameters are input into the photovoltaic power generation prediction model to obtain the output power of the system, and the output power is compared with the output power before optimization to evaluate the optimization effect.

Images