Adaptive photovoltaic tracking control method and system

Abstract

The application discloses a kind of adaptive photovoltaic tracking control method and system, method includes by collecting fusion multi-source data, the operating environment of photovoltaic system is detected in real time, and the real-time processing of the environment data collected;Adaptive photovoltaic tracking control model is constructed, adaptive photovoltaic tracking control model is trained, according to historical data and real-time environmental change, the tracking strategy of photovoltaic panel is dynamically adjusted;Dynamic error compensation mechanism is introduced, the error of tracking system is corrected in real time, and multi-objective optimization strategy is designed, according to different environmental conditions, automatically switch tracking mode.System includes multi-source data acquisition module, photovoltaic tracking control module and tracking mode switching module.The application improves the precision and efficiency of photovoltaic system operation monitoring;For photovoltaic tracking control model provides high-quality data input, helps to improve the decision-making ability of model;Realize the intelligentization and automation of photovoltaic panel tracking strategy, improve the power generation efficiency of photovoltaic system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation equipment, in particular to an adaptive photovoltaic tracking control method and system. BACKGROUND

[0002] The photovoltaic tracking control system adjusts the angle of the photovoltaic panel automatically to always face the sun, thereby maximizing the absorption and conversion efficiency of solar energy; this system can significantly improve the power generation efficiency of the photovoltaic system, reduce the unit power generation cost, and improve the return on investment; at present, the photovoltaic tracking control system is mainly divided into the following types: single-axis tracking, suitable for east-west tracking, with lower cost but limited accuracy; dual-axis tracking, with two degrees of rotational freedom, can more accurately track the movement of the sun, suitable for environments with more snow and dust; intelligent tracking system, combining sensor, microprocessor and fuzzy logic control technology, to realize efficient and accurate tracking control; although the photovoltaic tracking control technology has made significant progress, it still faces some challenges: high-concentration photovoltaic systems have very high tracking accuracy requirements, and existing tracker controllers still have errors in some cases; although the active tracking system can improve the power generation efficiency, the motor drive will consume a certain amount of power, and the control system needs to be optimized to reduce energy consumption; severe weather conditions pose higher requirements for the stability of the tracking system.

[0003] Prior art one, Chinese patent, application number 202110150065.4 discloses a photovoltaic tracking support control method, a photovoltaic tracking controller and a photovoltaic tracking system, acquires photovoltaic tracking related information, the photovoltaic tracking related information includes the ac power state, the electrical parameter and the photovoltaic tracking support tracking command of the photovoltaic inverter, when the ac power state is ac limited power state, and the photovoltaic tracking support tracking command is tracking support anti-tracking command, according to the photovoltaic tracking related information, the first control instruction of reducing the light incidence angle is generated, and the photovoltaic tracking support is controlled to track the reverse tracking rotation according to the first control instruction, and the light incidence angle is reduced. Although when the photovoltaic inverter is in the ac limited power state, and the direct current operating voltage of the photovoltaic inverter exceeds the threshold voltage, the light incidence angle is reduced by controlling the photovoltaic tracking support to track the reverse rotation, so that the power of the photovoltaic module installed on the photovoltaic tracking support is appropriately reduced, the direct current voltage of the photovoltaic inverter at this time is reduced, and the direct current voltage of the photovoltaic inverter is prevented from being too high; but the control strategy is not fine enough, which may cause the photovoltaic panel to have a short-term power generation interruption or fluctuation during the adjustment process, thereby affecting the stability of the system.

[0004] The prior art two, Chinese patent, application number 202311423764.7 discloses a photovoltaic tracking method and a photovoltaic tracking system, the photovoltaic tracking method comprises: obtaining a first solar radiation and a first photovoltaic power of a control group photovoltaic module; obtaining a user load power; determining an upper limit value of the radiation according to the first solar radiation, the first photovoltaic power and the user load power; obtaining a second solar radiation of an application group photovoltaic module; and adjusting the angle of the solar receiving surface of the application group photovoltaic module according to the upper limit value of the radiation and the second solar radiation. Although it solves the problem that when too much energy power is sent into the power grid at the same time, there is a reverse flow or a negative electricity price, which affects the stability of the power grid and the balance of electricity supply and demand; but the cost is relatively high, and the installation and maintenance are relatively complex.

[0005] The prior art three, Chinese patent, application number 202111049082.5 discloses a photovoltaic tracking control method, device, tracking controller and photovoltaic tracking system. The tracking controller does not adjust the angle of the tracking support at the last preset adjustment time and the next preset adjustment time, but adjusts the angle of the tracking support to the target tracking angle of the tracking support at the next preset adjustment time at the advance adjustment time between the last preset adjustment time and the next preset adjustment time, so that the mutation degree of the power generation of the photovoltaic module is reduced in the time period between the last preset adjustment time and the next preset adjustment time, and the continuity of the power generation is improved. Compared with the way of adjusting the tracking angle at each preset adjustment time to make the power generation of the photovoltaic module suddenly change, the power generation of the photovoltaic module in the time period between the last preset adjustment time and the next preset adjustment time can be improved. Although the power loss of photovoltaic power generation can be reduced; but the angle of the advance adjustment may not completely match the actual optimal tracking angle, thereby affecting the photovoltaic power generation efficiency.

[0006] At present, the prior art one, the prior art two and the prior art three have the problems that the control strategy is not fine enough, which may cause the photovoltaic panel to have a short power generation interruption or fluctuation during the adjustment process, thereby affecting the stability of the system, the cost is relatively high, and the installation and maintenance are relatively complex, and the angle of the advance adjustment may not completely match the actual optimal tracking angle, thereby affecting the photovoltaic power generation efficiency. Therefore, the present application provides a self-adaptive photovoltaic tracking control method and system. SUMMARY

[0007] The main purpose of the present application is to provide a self-adaptive photovoltaic tracking control method and system to solve the problems of the prior art that the control strategy is not fine enough, affects the stability of the system, the cost is relatively high, and the installation and maintenance are relatively complex, and affects the photovoltaic power generation efficiency.

[0008] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0009] An adaptive photovoltaic tracking control method, the adaptive photovoltaic tracking control method comprises:

[0010] By collecting and fusing multi-source data, the running environment of the photovoltaic system is detected in real time, and the collected environmental data is processed in real time;

[0011] An adaptive photovoltaic tracking control model is constructed, the adaptive photovoltaic tracking control model is trained, and the tracking strategy of the photovoltaic panel is dynamically adjusted according to historical data and real-time environmental changes;

[0012] A dynamic error compensation mechanism is introduced to correct the error of the tracking system in real time, and a multi-objective optimization strategy is designed to automatically switch the tracking mode according to different environmental conditions.

[0013] As a further improvement of the present application, the process of collecting and fusing multi-source data comprises the following steps:

[0014] Multi-sensor is used to collect meteorological data, environmental illumination and cloud coverage multi-source data in real time at the key nodes of the photovoltaic power station, and the collected data is preprocessed;

[0015] The preprocessed multi-source data is transmitted to a central data processing platform for edge computing processing such as hotspot detection, fault diagnosis and environmental adaptability adjustment;

[0016] The multi-source data processed by edge computing is fused and analyzed, and is input into the adaptive photovoltaic tracking control model to dynamically adjust the tracking angle of the photovoltaic panel according to environmental changes.

[0017] As a further improvement of the present application, the process of constructing the adaptive photovoltaic tracking control model comprises the following steps:

[0018] The multi-source data after fusion analysis is obtained, the photovoltaic output historical data of the target photovoltaic system under different weather conditions is collected, and a data set is constructed;

[0019] The adaptive photovoltaic tracking control model is constructed, and the position of the maximum power point of the photovoltaic panel is predicted according to the current environmental conditions;

[0020] The trained adaptive photovoltaic tracking control model is deployed in the target photovoltaic system for testing, the effectiveness of the model is evaluated according to the test results, and the parameters of the adaptive photovoltaic tracking control model are further optimized.

[0021] As a further improvement of the present application, the process of constructing the adaptive photovoltaic tracking control model comprises the following steps:

[0022] The input features in the data set are analyzed for feature importance and principal component analysis, and the input features are selected and optimized;

[0023] An adaptive photovoltaic tracking control model was constructed using a neural network model. The model input variables included light intensity, temperature, current, and voltage parameters. The dataset was divided into a training set and a validation set. The adaptive photovoltaic tracking control model was trained using the training set, and the weights and biases were adjusted using an optimization algorithm.

[0024] The model performance is evaluated using a validation set, and the accuracy of the adaptive photovoltaic tracking control model is measured by calculating indicators such as prediction error and mean square error. A reward function is designed based on the prediction error, and rewards or penalties are given according to the actual power generation efficiency and power output of the photovoltaic panels.

[0025] As a further improvement of the present invention, the process of performing feature importance analysis and principal component analysis on the input features in the dataset includes the following steps:

[0026] The data in the dataset is standardized, the correlation coefficient matrix of the dataset is calculated, and the covariance matrix is ​​decomposed by eigenvalue decomposition to obtain eigenvalues ​​and corresponding eigenvectors.

[0027] Based on the magnitude of the obtained eigenvalues ​​or the cumulative contribution rate, the eigenvector corresponding to the largest eigenvalue is selected as the principal component. When the cumulative contribution rate reaches 80%, the principal component extraction is stopped.

[0028] The original data is dimensionality reduced using the selected extracted principal components to generate a compressed dataset.

[0029] As a further improvement of the present invention, the process of training the adaptive photovoltaic tracking control model includes the following steps:

[0030] The processed dataset was divided into a training set and a validation set in a 7:3 ratio, and the weights and biases of the adaptive photovoltaic tracking control model were initialized using a random initialization method.

[0031] The adaptive photovoltaic tracking control model is trained using training set data. During the training process, the weights and biases are updated through backpropagation algorithm, and the parameters of the adaptive photovoltaic tracking control model are continuously adjusted.

[0032] After training, the adaptive photovoltaic tracking control model is evaluated using a validation set; based on the performance evaluation results, the parameters of the adaptive photovoltaic tracking control model or the optimization algorithm are further adjusted.

[0033] As a further improvement of the present invention, the process of designing a reward function based on prediction error includes the following steps:

[0034] The performance of the adaptive photovoltaic tracking control model is evaluated using a validation set, and the prediction error of the adaptive photovoltaic tracking control model is calculated. A reward function is designed based on the prediction error.

[0035] Rewards or penalties are given based on the actual power generation efficiency and power output of the photovoltaic panels; if the power predicted by the adaptive photovoltaic tracking control model is close to the actual power output, a positive reward is given; if the prediction error is large, a negative penalty is given.

[0036] After each training step, the adaptive photovoltaic tracking control model is fine-tuned based on the results of the reward function. The training process is repeated until the prediction error of the adaptive photovoltaic tracking control model reaches the preset target.

[0037] As a further improvement of the present invention, the process of designing a multi-objective optimization strategy includes the following steps:

[0038] Sensors are used to monitor the angular error between the photovoltaic panel and the sun in real time and to collect environmental data on light intensity.

[0039] The system uses Kalman filtering or particle filtering algorithms to estimate and correct errors in real time. If the system detects that the error exceeds a certain threshold, the drive motor will make fine adjustments to ensure that the photovoltaic panel is always at the optimal angle.

[0040] Based on real-time environmental data, the tracking strategy is dynamically adjusted through fuzzy control or genetic algorithm, and different tracking strategies are set for different environments, including using a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather.

[0041] As a further improvement of the present invention, the process of setting up the usage environment for different tracking strategies includes the following steps:

[0042] The tracking strategy is dynamically adjusted based on real-time environmental data using a fuzzy logic controller; by fuzzifying weather condition variables, the output tracking strategy is determined using preset fuzzy rules.

[0043] The parameters of the fuzzy controller are optimized by a genetic algorithm, and the performance of the fuzzy controller is continuously optimized through iterative selection, crossover and mutation operations; so that it adopts a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather.

[0044] The control strategy is further optimized by monitoring the tracking effect in real time. The tracking strategy is dynamically adjusted according to changes in environmental data. When the weather and environmental conditions change, the system automatically switches the tracking mode.

[0045] To achieve the above objectives, the present invention also provides the following technical solution:

[0046] An adaptive photovoltaic tracking control system, applied to the aforementioned adaptive photovoltaic tracking control method, the adaptive photovoltaic tracking control system comprising:

[0047] The multi-source data acquisition module is used to collect and integrate multi-source data, monitor the operating environment of the photovoltaic system in real time, and process the collected environmental data in real time.

[0048] The photovoltaic tracking control module is used to build and train an adaptive photovoltaic tracking control model, and dynamically adjust the tracking strategy of the photovoltaic panel based on historical data and real-time environmental changes.

[0049] The tracking mode switching module is used to introduce a dynamic error compensation mechanism to correct the error of the tracking system in real time, and to design a multi-objective optimization strategy to automatically switch the tracking mode according to different environmental conditions.

[0050] This invention achieves a comprehensive and accurate understanding of the photovoltaic system's operating environment by collecting and fusing multi-source data; real-time processing of environmental data ensures its timeliness and accuracy, providing a reliable basis for decision-making; an adaptive photovoltaic tracking control model is designed, enabling the model to dynamically adjust its strategy based on historical data and real-time environmental changes; training improves the model's adaptability and robustness; a dynamic error compensation mechanism is introduced to correct tracking system errors in real time, improving tracking accuracy; and a multi-objective optimization strategy is designed to automatically switch tracking modes according to different environmental conditions, achieving flexibility and diversity in the tracking strategy. Attached Figure Description

[0051] Figure 1 This is a schematic flowchart of one embodiment of the adaptive photovoltaic tracking control method of the present invention;

[0052] Figure 2 This is a schematic flowchart illustrating the steps of acquiring and fusing multi-source data in one embodiment of the adaptive photovoltaic tracking control method of the present invention.

[0053] Figure 3 This is a schematic flowchart illustrating the steps involved in constructing an adaptive photovoltaic tracking control model according to an embodiment of the adaptive photovoltaic tracking control method of the present invention.

[0054] Figure 4 This is a flowchart illustrating the steps involved in designing a multi-objective optimization strategy for an embodiment of the adaptive photovoltaic tracking control method of the present invention.

[0055] Figure 5 This is a functional module diagram of an embodiment of the adaptive photovoltaic tracking control system of the present invention;

[0056] Figure 6 This is a schematic diagram of the structure of an embodiment of the electronic device of the present invention;

[0057] Figure 7 This is a schematic diagram of the structure of one embodiment of the storage medium of the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0059] The terms "first," "second," and "third" used in this invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this invention are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the accompanying drawings). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0060] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0061] like Figure 1 As shown, this embodiment provides an example of an adaptive photovoltaic tracking control method. In this embodiment, the adaptive photovoltaic tracking control method specifically includes the following steps:

[0062] Step S100: By collecting and integrating multi-source data, the operating environment of the photovoltaic system is detected in real time, and the collected environmental data is processed in real time.

[0063] Step S200: Construct an adaptive photovoltaic tracking control model, train the adaptive photovoltaic tracking control model, and dynamically adjust the tracking strategy of the photovoltaic panel based on historical data and real-time environmental changes;

[0064] Step S300: Introduce a dynamic error compensation mechanism to correct the error of the tracking system in real time, and design a multi-objective optimization strategy to automatically switch the tracking mode according to different environmental conditions.

[0065] Preferably, in this embodiment, step S100, by collecting and fusing multi-source data, can comprehensively and accurately understand the operating environment of the photovoltaic system; real-time processing of environmental data ensures the timeliness and accuracy of the data, providing a reliable basis for decision-making; step S200 designs an adaptive photovoltaic tracking control model, enabling the model to dynamically adjust its strategy based on historical data and real-time environmental changes; training improves the model's adaptability and robustness; step S300 introduces a dynamic error compensation mechanism, which can correct the tracking system's errors in real time and improve tracking accuracy; and a multi-objective optimization strategy is designed to automatically switch tracking modes according to different environmental conditions, realizing the flexibility and diversity of the tracking strategy.

[0066] In summary, this embodiment improves the accuracy and efficiency of photovoltaic system operation monitoring; provides high-quality data input for the photovoltaic tracking control model, which helps improve the model's decision-making ability; realizes the intelligentization and automation of photovoltaic panel tracking strategies, improving the power generation efficiency of the photovoltaic system; reduces power generation losses caused by environmental changes, improving the economy and reliability of the photovoltaic system; further enhances the performance and stability of the photovoltaic tracking system; adapts to tracking needs under different environmental conditions, and improves the overall power generation capacity and adaptability of the photovoltaic system.

[0067] Furthermore, such as Figure 2 As shown, in the adaptive photovoltaic tracking control method of this embodiment, the process of collecting and fusing multi-source data in step S100 specifically includes the following steps:

[0068] Step S101: At key nodes of the photovoltaic power station, multiple sensors are used to collect real-time data from multiple sources, such as meteorological data, ambient light, and cloud cover, and the collected data is preprocessed.

[0069] Step S102: Transmit the preprocessed multi-source data to the central data processing platform, and perform edge computing processing such as hotspot detection, fault diagnosis and environmental adaptability adjustment on the preprocessed multi-source data;

[0070] Step S103: The multi-source data processed by edge computing is fused and analyzed, and then input into the adaptive photovoltaic tracking control model to dynamically adjust the tracking angle of the photovoltaic panel according to environmental changes.

[0071] Preferably, in step S101 of this embodiment, multiple sensors are deployed at key nodes of the photovoltaic power station to collect multi-source data such as meteorological data, ambient light intensity, and cloud cover in real time and comprehensively. Preprocessing the collected data removes noise and fills in missing values, improving data accuracy and usability. Step S102 transmits the preprocessed data to a central data processing platform, achieving centralized data management and processing. Edge computing processing, such as hotspot detection, fault diagnosis, and environmental adaptability adjustment, is performed on the platform, enabling timely detection and resolution of problems, improving the operating efficiency and safety of the photovoltaic power station. Step S103 fuses and analyzes the multi-source data after edge computing, comprehensively considering the impact of various factors on the photovoltaic power station. The results of the fusion analysis are input into an adaptive photovoltaic tracking control model, which dynamically adjusts the tracking angle of the photovoltaic panels according to environmental changes, thereby improving the power generation efficiency of the photovoltaic power station.

[0072] In summary, the real-time and accurate data acquisition in this embodiment provides a solid foundation for subsequent data processing and analysis; the preprocessing steps ensure data quality, providing a reliable data source for subsequent edge computing and fusion analysis; edge computing reduces data transmission latency and improves processing speed, enabling photovoltaic power plants to respond to environmental changes more quickly; hotspot detection and fault diagnosis can promptly identify potential problems, prevent fault escalation, and reduce downtime; fusion analysis makes photovoltaic power plant decision-making more intelligent and precise; dynamically adjusting the tracking angle of photovoltaic panels can maximize the utilization of solar energy resources, improve power generation efficiency, and reduce operating costs.

[0073] Furthermore, such as Figure 3 As shown, in the adaptive photovoltaic tracking control method of this embodiment, the process of constructing the adaptive photovoltaic tracking control model in step S200 specifically includes the following steps:

[0074] Step S201: Obtain multi-source data after fusion analysis, collect historical photovoltaic output data of the target photovoltaic system under different weather conditions, and construct a dataset;

[0075] Step S202: Construct an adaptive photovoltaic tracking control model to predict the location of the maximum power point of the photovoltaic panel based on the current environmental conditions;

[0076] Step S203: Deploy the trained adaptive photovoltaic tracking control model to the target photovoltaic system for testing. Based on the test results, evaluate the effectiveness of the model and further optimize the parameters of the adaptive photovoltaic tracking control model.

[0077] Preferably, in this embodiment, step S201 involves acquiring multi-source data after fusion analysis to construct a dataset containing photovoltaic system output data under different weather conditions. This dataset forms the basis for training the adaptive photovoltaic tracking control model and is crucial for the model to accurately predict the location of the photovoltaic panel's maximum power point under different environments. This provides a rich data environment, enabling the model to learn how to control the photovoltaic system to maximize output power in varying environments. Step S202 involves constructing the adaptive photovoltaic tracking control model, creating a model capable of predicting the location of the photovoltaic panel's maximum power point based on current environmental conditions. The model can dynamically adjust the photovoltaic system's tracking strategy to adapt to constantly changing environmental conditions, such as weather, light intensity, and temperature. This improves the efficiency and reliability of the photovoltaic system by dynamically adjusting the tracking strategy, ensuring the photovoltaic panel remains in the optimal position to obtain maximum solar energy. Step S203 involves deploying and testing the adaptive photovoltaic tracking control model, conducting practical tests on the model, and evaluating its effectiveness based on the test results. This includes verifying the model's performance and further optimizing the model parameters. This ensures the model achieves the expected results in practical applications and can be further adjusted and optimized based on actual operating conditions to achieve the best tracking control effect.

[0078] In summary, the three steps of this embodiment together constitute a closed-loop system, which not only improves the efficiency of the photovoltaic system but also possesses self-learning and self-optimization capabilities. This is key to achieving a high-efficiency, low-cost, and environmentally friendly energy system. Through the adaptive control model, solar energy resources can be better utilized, promoting the development of renewable energy technologies.

[0079] Furthermore, in the adaptive photovoltaic tracking control method of this embodiment, the process of constructing the adaptive photovoltaic tracking control model in step S202 specifically includes the following steps:

[0080] Step S2021: Perform feature importance analysis and principal component analysis on the input features in the dataset, and select and optimize the input features; where the input features are the photovoltaic output of the target photovoltaic system under different weather conditions;

[0081] Step S2022: Construct an adaptive photovoltaic tracking control model using a neural network model. The model input variables include light intensity, temperature, current, and voltage parameters. Divide the dataset into a training set and a validation set. Use the training set to train the adaptive photovoltaic tracking control model and use an optimization algorithm to adjust the weights and biases.

[0082] Step S2023: Use the validation set to evaluate the model performance. Measure the accuracy of the adaptive photovoltaic tracking control model by calculating indicators such as prediction error and mean square error. Design a reward function based on the prediction error and give rewards or penalties according to the actual power generation efficiency and power output of the photovoltaic panels.

[0083] Preferably, in this embodiment, step S2021 performs feature importance analysis and principal component analysis on the input features in the dataset to select and optimize the input features. Feature importance analysis helps identify which features have the greatest impact on the model's prediction results, thereby retaining important features and removing irrelevant or redundant features. Principal component analysis can further reduce feature dimensionality, retain the main information in the data, and reduce computational complexity. Step S2022 uses a neural network model to construct an adaptive photovoltaic tracking control model, trains the model using a training set, and uses an optimization algorithm to adjust the weights and biases. The neural network model can handle complex nonlinear relationships. Suitable for building adaptive photovoltaic tracking control models; training with a training set allows the model to learn the relationship between light intensity, temperature, current, voltage parameters, and photovoltaic panel power generation efficiency; the optimization algorithm can continuously adjust the model's weights and biases to minimize prediction errors; step S2023 uses a validation set to evaluate model performance and designs a reward function to reward or penalize based on the actual power generation efficiency and power output of the photovoltaic panel; by calculating indicators such as prediction error and mean square error, the model's prediction accuracy can be quantified; the design of the reward function allows the model to continuously optimize its performance during training to pursue higher power generation efficiency and power output.

[0084] In summary, this embodiment optimizes input features to improve model training efficiency and prediction accuracy; by reducing the number of features, model complexity is reduced, making it easier to train and deploy; the constructed adaptive photovoltaic tracking control model can adjust the working state of photovoltaic panels in real time according to current environmental conditions, thereby improving power generation efficiency; through training and optimization, the model can gradually approximate the real photovoltaic power generation process, improving prediction accuracy and reliability; validation set evaluation is a key step to ensure the model performs well in practical applications; the introduction of the reward function enhances the model's self-learning ability, enabling it to adaptively adjust according to actual power generation efficiency and power output, thereby improving overall performance.

[0085] Furthermore, in the adaptive photovoltaic tracking control method of this embodiment, the process of performing feature importance analysis and principal component analysis on the input features in the dataset in step S2021 specifically includes the following steps:

[0086] Step S20211: Standardize the data in the dataset, calculate the correlation coefficient matrix of the dataset, and decompose the covariance matrix by eigenvalue decomposition to obtain eigenvalues ​​and corresponding eigenvectors;

[0087] Step S20212: Based on the magnitude of the obtained eigenvalues ​​or the cumulative contribution rate, select the eigenvector corresponding to the largest eigenvalue as the principal component. When the cumulative contribution rate reaches 80%, stop the principal component extraction.

[0088] Step S20213: Use the selected extracted principal components to perform dimensionality reduction on the original data to generate a compressed dataset.

[0089] Step S20211 involves standardizing the data in the dataset:

[0090]

[0091] In the formula, After standardization, the first i The first sample j One eigenvalue; Represents the first in the original dataset i The first sample j One eigenvalue; Indicates the first of all samples j The average of the eigenvalues ​​represents the central tendency of that eigenvalue across all samples. Indicates the first j The standard deviation of each eigenvalue measures the dispersion of that eigenvalue; the formula converts the eigenvalues ​​into a distribution with a mean of 1 and a standard deviation of 1, which helps the algorithm process the eigenvalues ​​more efficiently because it eliminates the influence of different units; the sample represents photovoltaic output data under different weather conditions (such as photovoltaic output on a certain day); the feature represents the weather factors that affect photovoltaic output (such as temperature, light intensity, humidity, etc.); the eigenvalue represents the specific value of a certain weather factor in a specific sample (such as a temperature of 25℃).

[0092] Calculate the correlation coefficient matrix and perform eigenvalue decomposition:

[0093]

[0094] In the formula, Representation of features and characteristics The correlation coefficient between two features measures the strength of the linear relationship between them. Indicates the first a The feature in the first k The values ​​of each sample; Indicates the first a The average of each feature across all samples; Indicates the first a Standard deviation of each feature; Indicates the total sample size;

[0095] Eigenvalue decomposition:

[0096]

[0097] In the formula, Represents the covariance matrix or correlation coefficient matrix; The eigenvector matrix contains A eigenvectors of a matrix; This represents an eigenvalue diagonal matrix, where the values ​​on the diagonal are... A Eigenvalues ​​of a matrix; extracting the main direction of change in data through eigenvalue decomposition is a key step in principal component analysis.

[0098] Step S20212: Principal Component Contribution Rate Calculation:

[0099]

[0100] In the formula, Represents the cumulative contribution rate, representing the previous... The cumulative contribution of each principal component to the total variance of the dataset; The eigenvalues ​​of the covariance matrix are represented by the eigenvalues ​​of the first eigenvalue. Variance contribution in each principal component direction; This indicates the number of principal components selected, i.e., the number of principal components considered important. This represents the total number of feature values, i.e., the number of features in the original dataset;

[0101] Step S20213: Data Dimensionality Reduction:

[0102]

[0103] In the formula, This indicates the dimensionality-reduced dataset, where the number of features per sample is reduced. This represents the original dataset, containing all samples and features; This represents a matrix composed of the eigenvectors corresponding to the selected principal components, used to project the original data into a lower-dimensional space. Through the above steps, the original dataset can be transformed into a lower-dimensional version while retaining as much key information as possible from the original data.

[0104] Preferably, in this embodiment, step S20211 scales each feature in the dataset to the same scale to ensure that the contribution of each feature to the result is fair and avoids result bias caused by different units; measures the linear correlation between the features in the dataset to provide a basis for eigenvalue decomposition; decomposes the covariance matrix to obtain eigenvalues ​​and eigenvectors; eigenvalues ​​reflect the importance of eigenvectors, while eigenvectors indicate the main direction of the data; step S20212 selects principal components based on the magnitude of the eigenvalues ​​or the cumulative contribution rate; the larger the eigenvalue, the greater the variance in the direction represented by the corresponding eigenvector, that is, the higher the dispersion of the data in this direction; extraction stops when the cumulative contribution rate reaches 80%, which means that the selected principal components can explain 80% of the variance in the original data, achieving better data compression and information preservation; step S20213 projects the original data using the selected principal components to generate a compressed dataset; maps the original data from a high-dimensional space to a low-dimensional space, while preserving the structure and information of the original data as much as possible.

[0105] In summary, this embodiment provides the necessary mathematical foundation for principal component selection, ensuring the accuracy and effectiveness of the analysis; by selecting the most important principal components, it achieves dimensionality reduction of the data, simplifies the data model, and improves computational efficiency; it ensures that the dimensionality-reduced data still retains most of the information of the original data, which is helpful for subsequent data analysis and modeling; it reduces the dimensionality of the data, thereby reducing data complexity and computational load, making data processing and analysis easier; it improves the visualization effect of the data, making the distribution of data in the low-dimensional space clearer and easier to understand; and it helps to discover potential patterns and trends in the data, providing more valuable information for data mining and machine learning.

[0106] Furthermore, in the adaptive photovoltaic tracking control method of this embodiment, the process of training the model in step S2022 specifically includes the following steps:

[0107] Step S20221: Divide the processed dataset into a training set and a validation set in a 7:3 ratio, and initialize the weights and biases of the adaptive photovoltaic tracking control model using a random initialization method;

[0108] Step S20222: Train the adaptive photovoltaic tracking control model using the training set data. During the training process, update the weights and biases using the backpropagation algorithm and continuously adjust the parameters of the adaptive photovoltaic tracking control model. The backpropagation algorithm is based on the chain rule and uses gradient descent or other optimization algorithms to update the network weights and biases.

[0109] Step S20223: After training is completed, the adaptive photovoltaic tracking control model is evaluated using the validation set; based on the performance evaluation results, the parameters of the adaptive photovoltaic tracking control model or the optimization algorithm are further adjusted.

[0110] Preferably, in step S20221 of this embodiment, the processed dataset is divided into a training set and a validation set in a 7:3 ratio. This helps ensure that the model can fully learn the data features during training, while the validation set is used to evaluate the model's generalization ability and avoid overfitting. The weights and biases of the adaptive photovoltaic tracking control model are set using a random initialization method, providing a starting point for model training. Random initialization ensures diversity in the model at the beginning of training, which helps explore a broader parameter space. In step S20222, the adaptive photovoltaic tracking control model is trained using the training set data. The weights and biases are continuously updated using the backpropagation algorithm, allowing the model to gradually approximate the distribution of the real data. During training, model parameters, including learning rate and batch size, are continuously adjusted based on the model's performance to ensure that the model can learn data features efficiently and accurately. In step S20223, the trained adaptive photovoltaic tracking control model is evaluated using the validation set, including metrics such as accuracy and recall, to measure the model's generalization ability on unknown data. Based on the validation results, model parameters are further adjusted or the algorithm is optimized to improve model performance.

[0111] In summary, this embodiment provides a reasonable data foundation for model training and validation, ensuring that the model can be effectively trained on real data; random initialization provides initial conditions for optimizing model parameters, laying the foundation for the training process; through training, the model can learn key features in the data, improving its performance in photovoltaic tracking and control tasks; parameter adjustment helps optimize the training process, improving the model's convergence speed and accuracy; validation allows for objective evaluation of model performance, providing direction for model optimization; and model optimization can further enhance model performance, making it more stable and reliable in practical applications.

[0112] Furthermore, in the adaptive photovoltaic tracking control method of this embodiment, the process of designing the reward function based on the prediction error in step S2023 specifically includes the following steps:

[0113] Step S20231: Use the validation set to evaluate the performance of the adaptive photovoltaic tracking control model, calculate the prediction error of the adaptive photovoltaic tracking control model, and design a reward function based on the prediction error;

[0114] Step S20232: Reward or penalize based on the actual power generation efficiency and power output of the photovoltaic panel; if the power predicted by the adaptive photovoltaic tracking control model is close to the actual power output, a positive reward is given; if the prediction error is large, a negative penalty is given.

[0115] Step S20233: After each training step, fine-tune the adaptive photovoltaic tracking control model based on the result of the reward function, and repeat the training process until the prediction error of the adaptive photovoltaic tracking control model reaches the preset target.

[0116] The rewards are divided into two parts: intermediate rewards and final rewards. Intermediate rewards are given dynamically during the training process based on the model's performance to help the model find optimization directions more quickly. Final rewards are given at the end of training based on the model's overall performance to ensure that the model can achieve the preset goals.

[0117] The intermediate reward design dynamically adjusts the reward value based on the model's performance during training; intermediate rewards are based on power bias.

[0118] Assume power deviation Intermediate Rewards Designed as follows:

[0119]

[0120] In the formula, and It is the intensity coefficient of the reward and punishment; and It is a smoothing parameter to avoid the denominator being zero; It is the deviation threshold, used to distinguish between positive rewards and negative penalties;

[0121] Based on the trend of prediction error, the intermediate reward can be referenced by the trend of prediction error; let the change in prediction error be... Intermediate Rewards It can be designed as follows:

[0122]

[0123] In the formula, and It is the intensity coefficient of the reward and punishment; This indicates that the prediction error is decreasing, and a positive reward is given. This indicates that the prediction error is increasing, and a negative penalty is imposed.

[0124] The final reward design needs to consider the overall performance of the model to ensure that the model can achieve the preset goals:

[0125] The final reward is based on the prediction error, let the prediction error be... Root mean square error RMSE Final reward Designed as follows:

[0126]

[0127] In the formula, and These are weighting coefficients used to balance the linear and nonlinear effects of errors; It is the attenuation coefficient, which controls the attenuation rate of the exponential part; This represents the final reward value; the smaller the error, the larger the reward value.

[0128] To further optimize the model, power generation efficiency is introduced as a reference for the final reward, based on power generation efficiency. Let power generation efficiency be... The final reward is the ratio of actual power generation to theoretical maximum power generation. Designed as follows:

[0129]

[0130] In the formula, Indicates power generation efficiency; It is the reward intensity coefficient;

[0131] Combine intermediate and final rewards to form a comprehensive reward function. It can be designed as follows:

[0132]

[0133] In the formula, and Used to guide the model to optimize gradually during training; and This mechanism ensures the model reaches a preset target at the end of training. It sets initial model parameters and reward function parameters; in each training step, it calculates intermediate rewards based on power deviation and prediction error trends; updates model parameters based on the comprehensive reward function; and at the end of training, it calculates the final reward based on prediction error and power generation efficiency. This process is repeated until the prediction error meets the preset target. By introducing intermediate and final rewards, and combining them with multi-dimensional indicators such as power deviation, prediction error trends, and power generation efficiency, the sparse reward problem can be effectively alleviated, improving the model's learning efficiency and performance.

[0134] Preferably, in this embodiment, step S20231 uses a validation set to objectively evaluate the performance of the adaptive photovoltaic tracking control model in the current state, which helps to understand the model's generalization ability and performance in practical applications; the reward function designed based on the prediction error provides a clear optimization objective for the reinforcement learning process; the smaller the prediction error, the higher the reward, and vice versa. Step S20232 gives a positive reward when the model's predicted power is close to the actual power output, which encourages the model to continue to maintain or improve this accuracy; when the prediction error is large, a negative penalty is given, which prompts the model to adjust its parameters to reduce errors in future predictions; Step S20233 after each training step, the model is fine-tuned according to the result of the reward function, ensuring that the model can gradually approach the optimal solution; the training process is repeated until the model's prediction error reaches the preset target, ensuring that the model can reach the required performance level after sufficient training.

[0135] In summary, this embodiment uses a reward function to guide the model to evolve towards improving prediction accuracy. Through this real-time feedback mechanism, the model can continuously learn and adjust during training, thereby gradually improving its predictive ability. This is crucial for improving the efficiency and stability of photovoltaic tracking systems. Through continuous iterative training and adjustment, the model can gradually adapt to various complex scenarios, improving its performance in practical applications. This is of great significance for improving the overall efficiency and economic benefits of photovoltaic power generation systems.

[0136] Furthermore, such as Figure 4 As shown, in the adaptive photovoltaic tracking control method of this embodiment, the process of designing a multi-objective optimization strategy in step S300 specifically includes the following steps:

[0137] Step S301: Use sensors to monitor the angle error between the photovoltaic panel and the sun in real time, and collect environmental data such as light intensity;

[0138] Step S302: Apply particle filtering algorithm to estimate and correct errors in real time; if the system detects that the error exceeds a certain threshold, drive the motor to make fine adjustments so that the photovoltaic panel is always at the optimal angle;

[0139] Step S303: Based on real-time environmental data, dynamically adjust the tracking strategy through fuzzy control or genetic algorithm, and set the usage environment for different tracking strategies, including using a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather.

[0140] The particle filter algorithm includes:

[0141] Nonlinear state prediction formula:

[0142]

[0143] In the formula, Indicates the first Each particle at time The state vector. The state vector typically contains information such as the angle and angular velocity of the photovoltaic panel; The basic state transition function describes the main dynamic characteristics of the system; for example, it may represent the motion law of a photovoltaic panel under undisturbed conditions. This represents a higher-order nonlinear function used to describe the secondary dynamic characteristics of a system, such as wind disturbances and mechanical vibrations, which often lead to the nonlinear behavior of the system. These represent the weighting coefficients for the nonlinear terms, used to adjust the impact of each nonlinear term on the state prediction. These coefficients can be determined experimentally or through optimization algorithms. The number of nonlinear terms indicates the types of minor dynamic characteristics that need to be considered in the system. By introducing higher-order nonlinear terms, the state prediction formula can more accurately describe the complex dynamic behavior of the photovoltaic system, thereby improving the accuracy of state estimation.

[0144] Higher-order observation likelihood function:

[0145]

[0146] In the formula, Indicates time The observation vector typically contains information such as the angle error and light intensity measured by the sensor. Indicates the first l An observation model describes the mapping relationship from the state vector to the observation vector; for example, it represents the angle error that the sensor should measure in a certain state. This represents a Gaussian distribution with a mean of 1 / 2. The covariance matrix is , indicating the state Below, observed The probability of; Indicates the first l The weights are distributed according to a Gaussian distribution, satisfying... , representing the importance of each observation model in the overall observation likelihood function; L The number of Gaussian distributions indicates the types of observation models that need to be considered. By introducing a mixture of Gaussian distributions, the observation likelihood function can more accurately describe the uncertainty of the observation data, thereby improving the robustness of the state estimation.

[0147] Adaptive resampling formula:

[0148]

[0149] In the formula, This represents the number of effective particles, used to measure the degree of particle degeneration; if the number of effective particles is too small, it means that the weight of most particles is concentrated on a few particles, resulting in particle degeneration. This represents the normalized particle weights, indicating the degree of matching between each particle and the observed data. This represents the resampling threshold, typically set to N / 2. If the number of effective particles is lower than this threshold, resampling is performed. The resampling strategy indicates that resampling can employ hierarchical resampling or systematic resampling methods to improve the efficiency and accuracy of resampling. Through an adaptive resampling strategy, particle degradation can be effectively avoided, thereby improving the accuracy and stability of state estimation.

[0150] The optimized formula for error correction is as follows:

[0151]

[0152] In the formula, Indicates time The angle error correction amount is used to adjust the angle of the photovoltaic panel; The desired angle of the photovoltaic panel is usually calculated using a solar position model; Indicates the first Each particle at time Angle estimate; This represents the regularization coefficient, used to balance error correction and the smoothness of the control signal; This represents a regularization term used to prevent excessive correction. By introducing an optimization objective function, the error correction process can calculate the correction amount more accurately, thereby improving the tracking accuracy of the photovoltaic panel.

[0153] Advanced formula for generating motor control signals:

[0154]

[0155] In the formula, Indicates time The motor control signal is used to drive the motor to adjust the angle of the photovoltaic panel; This represents the proportional gain coefficient, used to adjust the system's response speed. This represents the differential gain coefficient, used to adjust the damping characteristics of the system. This represents the integral gain coefficient, used to eliminate steady-state error; This represents the sliding mode gain coefficient, used to enhance the robustness of the system; This represents a symbolic function used in sliding mode control; The integral term representing the error correction is used to eliminate steady-state error; by introducing a high-order controller, the motor control signal can more accurately adjust the angle of the photovoltaic panel, thereby improving the tracking performance and stability of the system.

[0156] Multi-objective optimization objective function:

[0157]

[0158] In the formula, This represents a multi-objective optimization objective function, used to comprehensively consider angle error, energy consumption, and system stability. , , These represent the weighting coefficients, corresponding to angle error, energy consumption, and system stability, respectively. This indicates the optimization time range and the number of time steps that need to be considered. The stability term is used to smooth the control signal. By introducing a multi-objective optimization objective function, efficient operation can be achieved under different environmental conditions, while ensuring stability and energy efficiency. Through the above complex formulas and optimization strategies, the particle filter algorithm can more accurately estimate and correct the angle error of the photovoltaic panel, while achieving multi-objective optimization and ensuring the efficient operation of the system under different environmental conditions.

[0159] Preferably, in this embodiment, step S301 uses a sensor to monitor the angle error between the photovoltaic panel and the sun in real time and collects environmental data such as light intensity. Through real-time monitoring, the system can obtain accurate angle error data and key environmental information such as light intensity. Step S302 applies Kalman filtering or particle filtering algorithms to estimate and correct the error in real time. If the system detects that the error exceeds a certain threshold, the drive motor is fine-tuned to keep the photovoltaic panel at the optimal angle. Through advanced filtering algorithms, the system can estimate and correct the error in real time and accurately, ensuring that the photovoltaic panel always maintains the optimal angle, thereby improving the light energy conversion efficiency. Step S303 dynamically adjusts the tracking strategy according to real-time environmental data through fuzzy control or genetic algorithms, and sets the usage environment of different tracking strategies, including using a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather. The system can dynamically adjust the tracking strategy according to environmental changes to adapt to different weather conditions, thereby improving the system's adaptability and flexibility.

[0160] In summary, this embodiment ensures that the system can operate precisely according to actual conditions, improving the system's response speed and accuracy. Through real-time correction, the system can adaptively adjust the angle of the photovoltaic panels to adapt to changes in the sun's position, thereby maximizing power generation efficiency. This enables the system to more intelligently adapt to complex and changing environmental conditions. By selecting appropriate tracking strategies, the system can maintain high power generation efficiency under different weather conditions, thereby improving the overall performance and stability of the system.

[0161] Furthermore, in the adaptive photovoltaic tracking control method of this embodiment, the process of setting the usage environment for different tracking strategies in step S303 specifically includes the following steps:

[0162] Step S3031: Utilize a fuzzy logic controller to dynamically adjust the tracking strategy based on real-time environmental data; by fuzzifying weather condition variables, use preset fuzzy rules to determine the output tracking strategy;

[0163] Step S3032: Optimize the parameters of the fuzzy controller through a genetic algorithm. Continuously optimize the performance of the fuzzy controller through iterative selection, crossover, and mutation operations; enable it to adopt a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather.

[0164] Step S3033: Further optimize the control strategy by monitoring the tracking effect in real time, so that the system can dynamically adjust the tracking strategy according to changes in environmental data. When the weather and environmental conditions change, the system automatically switches the tracking mode.

[0165] Preferably, in this embodiment, step S3031 utilizes a fuzzy logic controller to dynamically adjust the tracking strategy based on real-time environmental data. By fuzzifying weather condition variables and using preset fuzzy rules to determine the output tracking strategy, the flexibility and adaptability of the tracking strategy are achieved. Fuzzy logic can handle uncertainty and fuzziness, enabling the system to make decisions based on incomplete or fuzzy information. Step S3032 optimizes the parameters of the fuzzy controller using a genetic algorithm. The parameters are continuously optimized through iterative selection, crossover, and mutation operations. This improves the performance of the fuzzy controller, enabling it to adopt a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather. Step S3033 further optimizes the control strategy by monitoring the tracking effect in real time. This allows the system to dynamically adjust the tracking strategy according to changes in environmental data and automatically switch the tracking mode when weather conditions change, ensuring continuous optimization and adaptability of the system.

[0166] In summary, this embodiment can automatically adjust its tracking behavior under different weather conditions, thereby improving tracking efficiency and accuracy; it can automatically adjust its tracking strategy according to changes in weather conditions while maintaining high efficiency and stability. The optimization process of the genetic algorithm ensures that the fuzzy controller can achieve optimal performance for specific application scenarios; it continuously learns and adapts to new environmental conditions, improving its long-term performance and reliability; and it can automatically adjust to adapt to constantly changing environmental conditions.

[0167] like Figure 5 As shown, this embodiment also provides an embodiment of an adaptive photovoltaic tracking control system. In this embodiment, the adaptive photovoltaic tracking control system is applied to the adaptive photovoltaic tracking control method as described in the above embodiments. The adaptive photovoltaic tracking control system specifically includes:

[0168] Multi-source data acquisition module 1 is used to collect and integrate multi-source data, detect the operating environment of the photovoltaic system in real time, and process the collected environmental data in real time.

[0169] Photovoltaic tracking control module 2 is used to build an adaptive photovoltaic tracking control model, train the adaptive photovoltaic tracking control model, and dynamically adjust the tracking strategy of the photovoltaic panel based on historical data and real-time environmental changes.

[0170] Tracking mode switching module 3 is used to introduce a dynamic error compensation mechanism to correct the error of the tracking system in real time, and to design a multi-objective optimization strategy to automatically switch the tracking mode according to different environmental conditions.

[0171] Preferably, in this embodiment, the multi-source data acquisition module 1 can comprehensively and accurately understand the operating environment of the photovoltaic system by collecting and fusing multi-source data; real-time processing of environmental data ensures the timeliness and accuracy of the data, providing a reliable basis for decision-making; the photovoltaic tracking control module 2 designs an adaptive photovoltaic tracking control model based on machine learning, enabling the model to dynamically adjust its strategy according to historical data and real-time environmental changes; training using reinforcement learning or deep learning improves the model's adaptability and robustness; the tracking mode switching module 3 introduces a dynamic error compensation mechanism, which can correct the tracking system's errors in real time and improve tracking accuracy; a multi-objective optimization strategy is designed to automatically switch tracking modes according to different environmental conditions, realizing the flexibility and diversity of the tracking strategy.

[0172] In summary, this embodiment improves the accuracy and efficiency of photovoltaic system operation monitoring; provides high-quality data input for the photovoltaic tracking control model, which helps improve the model's decision-making ability; realizes the intelligentization and automation of photovoltaic panel tracking strategies, improving the power generation efficiency of the photovoltaic system; reduces power generation losses caused by environmental changes, improving the economy and reliability of the photovoltaic system; further enhances the performance and stability of the photovoltaic tracking system; adapts to tracking needs under different environmental conditions, and improves the overall power generation capacity and adaptability of the photovoltaic system.

[0173] like Figure 6 As shown, this embodiment provides an embodiment of an electronic device 4, which includes a processor 41 and a memory 42 coupled to the processor 41.

[0174] The memory 42 stores program instructions for implementing the adaptive photovoltaic tracking control system of any of the above embodiments.

[0175] The processor 41 is used to execute program instructions stored in the memory 42 to perform an adaptive photovoltaic tracking control system.

[0176] The processor 41 can also be referred to as a CPU (Central Processing Unit). The processor 41 may be an integrated circuit chip with signal processing capabilities. The processor 41 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.

[0177] Furthermore, Figure 7 This is a schematic diagram of the structure of a storage medium according to an embodiment of this application. The storage medium 5 of this embodiment stores program instructions 51 capable of implementing all the methods described above. These program instructions 51 can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0178] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0179] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

[0180] The specific embodiments of the invention have been described in detail above, but these are merely examples, and the invention is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this invention. Therefore, all equivalent transformations, modifications, and improvements made without departing from the spirit and principles of this invention should be included within the scope of this invention.

Claims

1. An adaptive photovoltaic tracking control method, characterized in that, The adaptive photovoltaic tracking control method includes: By collecting and integrating multi-source data, the operating environment of the photovoltaic system can be monitored in real time, and the collected environmental data can be processed in real time. An adaptive photovoltaic tracking control model is constructed and trained. Based on historical data and real-time environmental changes, the tracking strategy of the photovoltaic panel is dynamically adjusted. The construction of the adaptive photovoltaic tracking control model includes: After obtaining multi-source data through fusion analysis, historical photovoltaic output data of the target photovoltaic system under different weather conditions are collected to construct a dataset; feature importance analysis and principal component analysis are performed on the input features in the dataset, and the input features are selected and optimized. An adaptive photovoltaic tracking control model is constructed using a neural network model to predict the location of the maximum power point of the photovoltaic panel based on the current environmental conditions. The model input variables include light intensity, temperature, current, and voltage parameters. The dataset is divided into a training set and a validation set. The adaptive photovoltaic tracking control model is trained using the training set, and the weights and biases are adjusted using an optimization algorithm. The trained adaptive photovoltaic tracking control model is deployed to the target photovoltaic system for testing. The effectiveness of the model is evaluated using a validation set, and the parameters of the adaptive photovoltaic tracking control model are optimized. The accuracy of the adaptive photovoltaic tracking control model is measured by calculating the prediction error and mean square error. A reward function is designed based on the prediction error, and rewards or penalties are given according to the actual power generation efficiency and power output of the photovoltaic panels. The reward function designed based on prediction error includes: If the power predicted by the adaptive photovoltaic tracking control model is close to the actual power output, a positive reward is given; if the prediction error is large, a negative penalty is given. After each training step, the adaptive photovoltaic tracking control model is fine-tuned based on the result of the reward function. The training process is repeated until the prediction error of the adaptive photovoltaic tracking control model reaches the preset target. A dynamic error compensation mechanism is introduced to correct the tracking system error in real time, and a multi-objective optimization strategy is designed to automatically switch the tracking mode according to different environmental conditions.

2. The adaptive photovoltaic tracking control method according to claim 1, characterized in that, The process of collecting and fusing multi-source data includes the following steps: At key nodes of the photovoltaic power station, multiple sensors are used to collect meteorological data, ambient light and cloud cover data in real time, and the collected data is preprocessed. The pre-processed multi-source data is transmitted to the central data processing platform, where edge computing processing is performed on the pre-processed multi-source data, including hotspot detection, fault diagnosis, and environmental adaptability adjustment. Multi-source data processed by edge computing is fused and analyzed, and then input into an adaptive photovoltaic tracking control model to dynamically adjust the tracking angle of the photovoltaic panel according to environmental changes.

3. The adaptive photovoltaic tracking control method according to claim 1, characterized in that, The process of performing feature importance analysis and principal component analysis on the input features in the dataset includes the following steps: The data in the dataset is standardized, the correlation coefficient matrix of the dataset is calculated, and the covariance matrix is ​​decomposed by eigenvalue decomposition to obtain eigenvalues ​​and corresponding eigenvectors. Based on the magnitude of the obtained eigenvalues ​​or the cumulative contribution rate, the eigenvector corresponding to the largest eigenvalue is selected as the principal component. When the cumulative contribution rate reaches 80%, the principal component extraction is stopped. The original data is dimensionality reduced using the selected extracted principal components to generate a compressed dataset.

4. The adaptive photovoltaic tracking control method according to claim 1, characterized in that, The process of training an adaptive photovoltaic tracking control model includes the following steps: The processed dataset was divided into a training set and a validation set in a 7:3 ratio, and the weights and biases of the adaptive photovoltaic tracking control model were initialized using a random initialization method. The adaptive photovoltaic tracking control model is trained using training set data. During the training process, the weights and biases are updated through backpropagation algorithm, and the parameters of the adaptive photovoltaic tracking control model are continuously adjusted. After training, the adaptive photovoltaic tracking control model is evaluated using a validation set; based on the performance evaluation results, the parameters of the adaptive photovoltaic tracking control model or the optimization algorithm are adjusted.

5. The adaptive photovoltaic tracking control method according to claim 1, characterized in that, The process of designing a multi-objective optimization strategy includes the following steps: Sensors are used to monitor the angular error between the photovoltaic panel and the sun in real time and to collect environmental data on light intensity. The system uses Kalman filtering or particle filtering algorithms to estimate and correct errors in real time. If the system detects that the error exceeds the threshold, it drives the motor to make fine adjustments so that the photovoltaic panel is always at the optimal angle. Particle filtering algorithms include: Nonlinear state prediction formula: In the formula, Indicates the first Each particle at time The state vector contains information about the angle and angular velocity of the photovoltaic panel. The basic state transition function describes the dynamic characteristics and represents the motion law of the photovoltaic panel under undisturbed conditions. This represents a higher-order nonlinear function used to describe minor dynamic characteristics such as wind disturbance and mechanical vibration. The weighting coefficients represent the nonlinear terms and are used to adjust the impact of each nonlinear term on state prediction. They are determined through experiments or optimization algorithms. It represents the number of nonlinear terms and the types of minor dynamic characteristics that need to be considered in the system. Higher-order observation likelihood function: In the formula, Indicates time The observation vector contains the angle error and light intensity information measured by the sensor; Indicates the first l An observation model describes the mapping relationship from the state vector to the observation vector; This represents a Gaussian distribution with a mean of 1 / 2. The covariance matrix is , indicating the state Below, observed The probability of; Indicates the first l The weights are distributed according to a Gaussian distribution, satisfying... , representing the importance of each observation model in the overall observation likelihood function; L This indicates the number of Gaussian distributions and the types of observation models that need to be considered. Adaptive resampling formula: In the formula, This represents the number of effective particles, used to measure the degree of particle degeneration; if the number of effective particles is too small, it means that the weight of most particles is concentrated on a few particles, resulting in particle degeneration. This represents the normalized particle weights, indicating the degree of matching between each particle and the observed data. This represents the resampling threshold, set to N / 2; if the number of valid particles is lower than this threshold, resampling will be performed. The optimized formula for error correction is as follows: In the formula, Indicates time The angle error correction amount is used to adjust the angle of the photovoltaic panel; The desired angle of the photovoltaic panel is calculated using a solar position model. Indicates the first Each particle at time Angle estimate; This represents the regularization coefficient, used to balance error correction and the smoothness of the control signal; This represents a regularization term, used to prevent excessive correction. Advanced formula for generating motor control signals: In the formula, Indicates time The motor control signal is used to drive the motor to adjust the angle of the photovoltaic panel; This represents the proportional gain coefficient, used to adjust the system's response speed. This represents the differential gain coefficient, used to adjust the damping characteristics of the system. This represents the integral gain coefficient, used to eliminate steady-state error; This represents the sliding mode gain coefficient, used to enhance the robustness of the system; This represents a symbolic function used in sliding mode control; The integral term representing the error correction amount is used to eliminate steady-state error; Multi-objective optimization objective function: In the formula, This represents a multi-objective optimization objective function, used to comprehensively consider angle error, energy consumption, and system stability. , , These represent the weighting coefficients, corresponding to angle error, energy consumption, and system stability, respectively. This indicates the optimization time range and the number of time steps that need to be considered. This represents a stability term, used to smooth control signals. Based on real-time environmental data, the tracking strategy is dynamically adjusted through fuzzy control or genetic algorithm, and different tracking strategies are set for different environments, including using a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather.

6. The adaptive photovoltaic tracking control method according to claim 5, characterized in that, The process of setting up the usage environment for different tracking strategies includes the following steps: The tracking strategy is dynamically adjusted based on real-time environmental data using a fuzzy logic controller; by fuzzifying weather condition variables, the output tracking strategy is determined using preset fuzzy rules. The parameters of the fuzzy controller are optimized by a genetic algorithm, and the performance of the fuzzy controller is continuously optimized through iterative selection, crossover and mutation operations; so that it adopts a fast tracking mode in cloudy weather and a stable tracking mode in sunny weather. The system optimizes the control strategy by monitoring the tracking effect in real time, dynamically adjusts the tracking strategy according to changes in environmental data, and automatically switches the tracking mode when the weather and environmental conditions change.

7. An adaptive photovoltaic tracking control system, applied to the adaptive photovoltaic tracking control method as described in any one of claims 1 to 6, characterized in that, The adaptive photovoltaic tracking control system includes: The multi-source data acquisition module is used to collect and integrate multi-source data, monitor the operating environment of the photovoltaic system in real time, and process the collected environmental data in real time. The photovoltaic tracking control module is used to build and train an adaptive photovoltaic tracking control model, and dynamically adjust the tracking strategy of the photovoltaic panel based on historical data and real-time environmental changes. The tracking mode switching module is used to introduce a dynamic error compensation mechanism to correct the error of the tracking system in real time, and to design a multi-objective optimization strategy to automatically switch the tracking mode according to different environmental conditions.

Images