A photovoltaic tracking support rotation control method and related device

Abstract

The application discloses a photovoltaic tracking support rotation control method and related device, and relates to the photovoltaic power generation field, in order to effectively filter instantaneous fluctuation (such as cloud layer fast sweeps), avoid photovoltaic tracking support frequent start-stop, only in the case where the first energy difference value of consecutive N control periods is greater than the first threshold value, the comparison of the power generation energy gain brought by the first tracking algorithm and the second tracking algorithm is triggered. And by calculating the second energy difference value representing the power generation energy gain brought by the first tracking algorithm relative to the second tracking algorithm, when the second energy difference value is greater than the second threshold value, the first tracking angle calculated by the more intelligent first tracking algorithm is used to maximize the photovoltaic array power generation energy. When the second energy difference value is less than or equal to the second threshold value, the second tracking angle calculated by the more stable second tracking algorithm is used to ensure the power generation robustness.

Description

Technical Field

[0001] This application relates to the field of photovoltaic power generation technology, and in particular to a method and device for controlling the rotation of a photovoltaic tracking bracket. Background Technology

[0002] To improve tracking accuracy, photovoltaic (PV) tracking systems are increasingly adopting optimization strategies based on irradiance prediction. In actual operation, PV power plants are frequently disturbed by complex cloud movements, and the tracking control system needs to assess the external environment at a high frequency. Therefore, there is an urgent need for sub-minute-scale ultra-short-term irradiance prediction technology. Furthermore, due to the vast area covered by power plants, the distribution and movement of clouds within the plant exhibit significant spatial heterogeneity. Relying solely on irradiance data from a single sampling point for prediction cannot accurately reflect the differentiated impact of clouds on the tracking systems in different areas. Achieving sub-minute-scale prediction with power plant-level accuracy requires integrating massive amounts of spatiotemporal feature data (such as ground-based / satellite cloud images, string-level power data, etc.) and utilizing complex deep learning models such as depth map convolutional networks and machine vision technology for modeling.

[0003] However, the aforementioned high-performance forecasting solutions face severe challenges in engineering implementation. Edge deployment environments are often limited by limited computing resources, difficulty in obtaining comprehensive meteorological data sources, and the high cost of high-precision sensors, resulting in insufficient feasibility and reliability of spatiotemporal forecasting algorithms in practical applications, and extremely high overall deployment and maintenance costs. Summary of the Invention

[0004] In view of the above problems, this application provides a method and related device for controlling the rotation of a photovoltaic tracking bracket. The specific solution is as follows:

[0005] The first aspect of this application provides a method for controlling the rotation of a photovoltaic tracking bracket, comprising:

[0006] Obtain the tracking angle of the photovoltaic tracking bracket and calculate the first power generation energy corresponding to the tracking angle;

[0007] The first tracking algorithm is used to calculate the first tracking angle of the photovoltaic array in the next control cycle, and the second power generation energy corresponding to the first tracking angle is calculated. The input data of the first tracking algorithm includes at least the solar position and irradiance data.

[0008] Calculate the first energy gain between the second generated energy and the first generated energy;

[0009] If the first energy gain is greater than the first threshold for N consecutive control cycles, the second tracking algorithm is used to calculate the second tracking angle for the next control cycle, and the third power generation energy corresponding to the second tracking angle is calculated. The input data of the second tracking algorithm includes at least the sun position, and N is an integer greater than 1.

[0010] Calculate the second energy gain between the second generated energy and the third generated energy;

[0011] When the second energy gain is greater than the second threshold, control the photovoltaic tracking bracket to rotate to the first tracking angle;

[0012] When the second energy gain is less than or equal to the second threshold, the photovoltaic tracking bracket is controlled to rotate to the second tracking angle.

[0013] In one possible implementation, the first threshold, the second threshold, and the count value N are determined based on the weather type of the current control cycle.

[0014] In one possible implementation, the method further includes:

[0015] Obtain statistical characteristics of irradiation data or power data within a sliding window, wherein the sliding window includes the current control cycle, and the statistical characteristics include at least one statistical indicator;

[0016] Based on the statistical indicators in the statistical features and the correspondence between the pre-configured statistical indicators and weather types, the weather type of the current control cycle is determined.

[0017] In one possible implementation, the method further includes: acquiring solar position information of the current control period and statistical characteristics of irradiance data or power data within a sliding window, wherein the sliding window includes the current control period, and the statistical characteristics include at least one of volatility index, abrupt change index, stability index, and clear sky index;

[0018] The solar position information and the statistical features are input into a pre-trained prediction model to obtain the weather type of the current control cycle output by the prediction model. The prediction model is trained in advance using training samples with labeled weather types.

[0019] In one possible implementation, the method for determining the first threshold, the second threshold, and the count value N corresponding to different weather types includes:

[0020] Obtain at least one historical sample dataset within the evaluation period, wherein the historical sample dataset includes data from multiple consecutive historical sample days, and the historical sample day data includes data sequences corresponding to one or more weather types;

[0021] Construct a parameter search space corresponding to each weather type, wherein the parameter search space includes: the value range of the first threshold, the value range of the second threshold, and the value range of the count value N;

[0022] The objective function is constructed with the goal of maximizing the difference between the power generation gain of the photovoltaic array and the rotation loss of the photovoltaic tracking bracket. The power generation gain of the photovoltaic array is the power generation energy gain relative to the second tracking algorithm and the first tracking algorithm controlling the rotation of the photovoltaic tracking bracket within the evaluation period.

[0023] The photovoltaic tracking bracket rotation control method is simulated using at least one historical sample dataset within the evaluation period. With the objective function as the optimization objective, the first threshold, the second threshold, and the count value N corresponding to each weather type are solved.

[0024] In one possible implementation, the method further includes:

[0025] Given local running data of a preset duration, the local running data is processed to obtain an actual sample dataset;

[0026] The photovoltaic tracking bracket rotation control method is executed using the actual sample dataset. With the objective function as the optimization objective, the candidate first threshold, candidate second threshold, and candidate count value N corresponding to each weather type are solved.

[0027] The first control parameter combination and the second control parameter combination are simulated and compared using the historical sample dataset or the actual sample dataset to determine the optimal control parameter combination. The first control parameter combination includes: the first threshold, the second threshold, and the count value N. The second control parameter combination includes: the candidate first threshold, the candidate second threshold, and the candidate count value N.

[0028] The photovoltaic tracking bracket rotation control method is executed using the optimal combination of control parameters.

[0029] A second aspect of this application provides a controller comprising at least one processor and a memory connected to the processor, wherein:

[0030] The memory is used to store computer programs;

[0031] The processor is used to execute the computer program so that the controller can implement the photovoltaic tracking bracket rotation control method of the first aspect or any implementation thereof.

[0032] A third aspect of this application provides a photovoltaic tracking bracket rotation control system, comprising: a controller and a photovoltaic tracking bracket drive device;

[0033] The controller is communicatively connected to the photovoltaic tracking bracket drive device;

[0034] The photovoltaic tracking bracket driving device is used to drive the photovoltaic tracking bracket to rotate, and the photovoltaic tracking bracket is equipped with photovoltaic panels.

[0035] The controller is capable of implementing the photovoltaic tracking bracket rotation control method described in the first aspect or any implementation thereof.

[0036] The fourth aspect of this application provides a computer program product including computer-readable instructions that, when executed on a controller, cause the controller to implement the photovoltaic tracking bracket rotation control method described in the first aspect or any implementation thereof.

[0037] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by a controller, enable the controller to implement the photovoltaic tracking bracket rotation control method described in the first aspect or any implementation thereof.

[0038] By employing the above technical solution, this application provides a photovoltaic tracking bracket rotation control method. This method calculates the first power generation energy corresponding to the tracking angle of the photovoltaic tracking bracket, uses a first tracking algorithm to calculate the first tracking angle of the photovoltaic array in the next control cycle, and calculates the second power generation energy corresponding to the first tracking angle. By calculating the first energy gain between the second and first power generation energies, the power generation energy gain brought about by controlling the rotation of the photovoltaic tracking bracket using the first tracking algorithm is determined. Only when the first energy gain is greater than a first threshold for N consecutive control cycles is a comparison of the power generation energy gains brought about by the first and second tracking algorithms triggered, effectively filtering out instantaneous fluctuations (such as rapidly passing clouds) and avoiding frequent start-stop of the photovoltaic tracking bracket. By calculating the second energy gain, the power generation energy gain brought about by the first tracking algorithm is determined relative to the second tracking algorithm. When the second energy gain is greater than a second threshold, the power generation energy gain can cover the risks brought about by the uncertainty of the first tracking algorithm. The first tracking angle calculated using a more intelligent first tracking algorithm maximizes the power generation energy of the photovoltaic array. When the second energy gain is less than or equal to the second threshold, the power generation energy gain may not cover the risks brought about by the uncertainty of the first tracking algorithm. The second tracking angle calculated using a more stable second tracking algorithm ensures power generation robustness. Attached Figure Description

[0039] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0040] Figure 1A schematic flowchart illustrating a photovoltaic tracking bracket rotation control method provided in an embodiment of this application;

[0041] Figure 2 This is a schematic flowchart of a method for determining a first threshold, a second threshold, and a count value N corresponding to different weather types, provided in an embodiment of this application.

[0042] Figure 3 This is a schematic diagram illustrating model parameter training and application provided in the embodiments of this application;

[0043] Figure 4 This is a schematic diagram of the theoretical tracking angle of the first tracking algorithm before smoothing optimization, provided in an embodiment of this application.

[0044] Figure 5 A schematic diagram of the theoretical tracking angle using the first tracking algorithm after smoothing optimization, provided for an embodiment of this application;

[0045] Figure 6 This is a schematic diagram of the structure of a controller provided in an embodiment of this application. Detailed Implementation

[0046] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0047] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0048] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0049] This application provides a method for controlling the rotation of a photovoltaic tracking bracket. The method is described in detail below with reference to the accompanying drawings.

[0050] Reference Figure 1 , Figure 1This is a flowchart illustrating a photovoltaic tracking bracket rotation control method provided in an embodiment of this application, as shown below. Figure 1 As shown in the figure, the photovoltaic tracking bracket rotation control method provided in this application embodiment may include steps 101 to 109, which are described in detail below.

[0051] 101: Obtain the tracking angle of the photovoltaic tracking bracket and calculate the first power generation energy corresponding to the tracking angle;

[0052] Specifically, the tracking angle of the photovoltaic tracking bracket within the current control cycle is obtained. The control cycle is set according to the actual application scenario, for example, it can be set to 1 minute.

[0053] Specifically, any existing method for calculating power generation energy based on tracking angle can be used. For example, one method for calculating power generation energy is as follows:

[0054] ;

[0055] ;

[0056] In the formula, For tracking angle For the width of the photovoltaic array, For the spacing of the photovoltaic array, The height of the photovoltaic array above the ground. This refers to the direct irradiance on a horizontal plane. This refers to the horizontal diffuse irradiance. It is the surface albedo. For bifacial modules, the bifaciality factor is... These are the electrical performance parameters of photovoltaic modules, such as photoelectric conversion efficiency. The temperature of the photovoltaic module is used to calculate the power generation energy more accurately. It is an optional parameter. If the power station has temperature acquisition capabilities, the temperature parameter can be added for angle energy calculation.

[0057] 102: The first tracking angle of the photovoltaic array in the next control cycle is calculated using the first tracking algorithm, and the second power generation energy corresponding to the first tracking angle is calculated. The input data of the first tracking algorithm includes at least the solar position and irradiance data.

[0058] The first tracking algorithm can be a power generation optimization algorithm based on solar position algorithm, solar irradiance model, and photovoltaic module photoelectric characteristic model. It obtains the photovoltaic array power generation energy corresponding to different tracking angles through enumeration or optimization calculations, and finally determines the tracking angle that maximizes the photovoltaic array power generation energy, achieving a technological leap from "tracking the sun's position" to "tracking maximum power generation." The photovoltaic array power generation energy can be calculated by comprehensively considering direct irradiance, diffuse irradiance, and ground-reflected irradiance to determine the irradiance distribution on the front and back of the tracking array at the corresponding tracking angle. Then, based on the photovoltaic module photoelectric characteristic model, the photovoltaic array power generation power under this irradiance distribution is calculated in a refined manner, i.e., the photovoltaic array power generation energy.

[0059] The photovoltaic module photoelectric characteristic model can be configured according to the specific module model and support structure on site, and specific parameters such as bifacial factor, array ground clearance, and array spacing can be configured. The core algorithm framework is decoupled from the specific equipment parameters and is suitable for different tracking support forms such as single and bifacial modules and 1P / 2P.

[0060] For example, the formula for calculating the power generation of a photovoltaic array is as follows:

[0061] ;

[0062] In the formula, For tracking angle, other parameters are as described above. .

[0063] The first tracking angle is the tracking angle that maximizes the power generation of the photovoltaic array, and the calculation formula is as follows:

[0064] ;

[0065] In the formula, θ1 and θ2 are the upper and lower limits of the theoretical tracking angle of the photovoltaic array tracking bracket, respectively.

[0066] The formula for calculating the second power generation energy corresponding to the first tracking angle is as follows:

[0067] .

[0068] 103: Calculate the first energy gain between the second generated energy and the first generated energy;

[0069] First Energy Gain :

[0070] .

[0071] The first energy gain represents the power generation energy gain resulting from controlling the rotation of the photovoltaic tracking bracket using the first tracking algorithm.

[0072] 104: Whether the first energy gain of N consecutive control cycles is greater than the first threshold;

[0073] To effectively filter out instantaneous fluctuations (such as rapidly passing clouds) and avoid frequent start-ups and shutdowns of the photovoltaic tracking bracket, this embodiment introduces a continuous determination mechanism based on a state machine. The system maintains a state register to record the current tracking mode state of the photovoltaic tracking bracket (such as "normal tracking state" or "intelligent optimization state") and the number of times C the first energy gain ΔE1 continuously exceeds the first threshold T1.

[0074] When ΔE1 > T1 (the first threshold corresponding to the current weather type), the counter C is incremented by 1.

[0075] When ΔE ≤ T1, counter C is cleared to zero.

[0076] The system generates a valid rotation command only when the value of counter C reaches N. This mechanism ensures that only a sustained and significant energy gain will trigger actual rotation, effectively filtering out short-term noise interference.

[0077] If the first energy gain is greater than the first threshold for N consecutive control cycles, execute 105: use the second tracking algorithm to calculate the second tracking angle for the next control cycle, and calculate the third power generation energy corresponding to the second tracking angle;

[0078] The input data for the second tracking algorithm includes at least the position of the sun.

[0079] The second tracking algorithm combines apparent solar motion trajectory tracking with an inverse tracking strategy. This algorithm is a mature benchmark in the field of photovoltaic tracking. Apparent solar tracking calculates the theoretical position of the sun in real time based on astronomical formulas, ensuring that the tracking array is always perpendicular to the direction of solar incidence. The inverse tracking strategy reduces the tracking angle during periods of lower solar altitude angle in the early morning and late evening to prevent power generation losses caused by shading between arrays. This algorithm does not rely on real-time irradiance sensor data and features simple calculation and high reliability, providing a stable benchmark for power generation revenue for the first tracking algorithm.

[0080] For example, the calculation formula for the second tracking algorithm is as follows:

[0081] ;

[0082] In the formula, It is a time parameter vector containing timestamps (year, month, day, hour, minute, second) and time zone information; This is a vector of geographic parameters, including the station's latitude, longitude, and altitude. Width of the photovoltaic array; This refers to the spacing between photovoltaic arrays.

[0083] The third power generation energy corresponding to the second tracking angle is .

[0084] 106: Calculate the second energy gain between the second and third generated energy;

[0085] Second energy gain :

[0086] .

[0087] The second energy gain represents the power generation energy gain achieved by using the first tracking algorithm relative to the second tracking algorithm.

[0088] 107: Is the second energy gain greater than the second threshold?

[0089] If the second energy gain is greater than the second threshold, proceed to step 108: control the photovoltaic tracking bracket to rotate to the first tracking angle;

[0090] When ΔE1 > T2 (the second threshold corresponding to the current weather type), it indicates that the theoretical advantage of the first tracking algorithm is relatively high, and its benefits can cover the risks brought about by the uncertainty of the first tracking algorithm under the current conditions. The target angle of the photovoltaic tracking bracket is then set to θ. max This maximizes the power generation of the photovoltaic array.

[0091] If the second energy gain is less than or equal to the second threshold, proceed to step 109: control the photovoltaic tracking bracket to rotate to the second tracking angle.

[0092] When ΔE ≤ T2, it indicates that the theoretical advantage of the first tracking algorithm is negligible, and its benefits may not outweigh the risks caused by the uncertainty of the first tracking algorithm under the current circumstances. In this case, the system will forcibly override the decision of the current first tracking algorithm and set the target angle of the photovoltaic tracking bracket to θ. con This ensures that the system operates on a mature and reliable benchmark, fundamentally avoiding the risk of "negative gain." This mechanism is key to guaranteeing the robustness of the system's power generation.

[0093] This embodiment provides a photovoltaic tracking bracket rotation control method. Only when the first energy gain is greater than a first threshold for N consecutive control cycles is a comparison of the power generation gain brought by the first tracking algorithm and the second tracking algorithm triggered. This effectively filters out instantaneous fluctuations (such as rapidly passing clouds) and avoids frequent start-stop cycles of the photovoltaic tracking bracket. By calculating the second energy gain, the power generation gain brought by the first tracking algorithm is determined relative to the second tracking algorithm. When the second energy gain is greater than the second threshold, the power generation gain can cover the risks caused by the uncertainty of the first tracking algorithm. The first tracking angle calculated using the more intelligent first tracking algorithm is then used to maximize the power generation energy of the photovoltaic array. When the second energy gain is less than or equal to the second threshold, the power generation gain may not be able to cover the risks caused by the uncertainty of the first tracking algorithm. The second tracking angle calculated using the more stable second tracking algorithm is then used to ensure power generation robustness.

[0094] In one possible implementation, the first threshold, the second threshold, and the count value N in the above embodiments can be fixed or determined according to the weather type of the current control cycle. That is, the first threshold may be different for different weather types, the second threshold may be different for different weather types, and the count value N may be different for different weather types.

[0095] For example, weather types include: stable sunny, uniformly cloudy, fluctuating slightly cloudy, fluctuating moderately cloudy, and fluctuating heavily cloudy.

[0096] The following sections introduce two methods for obtaining weather types.

[0097] Method 1: Empirical Threshold Model

[0098] Statistical characteristics of irradiance or power data within a sliding window are obtained. The sliding window includes the current control period. The statistical characteristics include at least one statistical indicator. Based on the statistical indicators in the above statistical characteristics and the correspondence between pre-configured statistical indicators and weather types, the weather type of the current control period is determined.

[0099] The aforementioned sliding window includes the current control period, which is a moving observation period. Every control period, the statistical characteristics of multiple past control periods are statistically analyzed. Each time the window slides for a control period, the oldest control period is discarded and a new control period is added. For example, if the control period is 1 minute, the sliding window can be 30 minutes, which is used to statistically analyze the irradiance data or power data within the past 30 minutes.

[0100] For example, statistical characteristics include statistical indicators such as the average clear sky index. and coefficient of variation C v .

[0101] ;

[0102] G represents the actual irradiance, which can be converted into the actual irradiance of the current control period by combining the installed capacity of the photovoltaic array and the power generation of the reference surface corresponding to the current control period.

[0103] ;

[0104] Where P is the power generation of the photovoltaic array in the current control cycle, and W is the installed capacity of the photovoltaic array.

[0105] G sky The irradiance value under the theoretical clearing conditions of the current control cycle can be calculated based on the power plant's geographical location information (latitude, longitude, and altitude) and the current time, according to the reference plane where the irradiance meter is installed and the photovoltaic array plane. For example, G... sky It can be calculated based on the clear-sky irradiance model. For the same input parameters, the model output is uniquely determined and there is no randomness.

[0106] ;

[0107] In the formula, It is a time parameter vector containing timestamps (year, month, day, hour, minute, second) and time zone information; This is a vector of geographic parameters, including the station's latitude, longitude, and altitude. This is a vector of geometric parameters, containing the tilt angle and azimuth angle of the photovoltaic array reference plane; This can be a vector of optional meteorological parameters, such as aerosol optical depth (AOD) and atmospheric precipitable water. If the selected station has optional meteorological parameters, they can be used to improve the calculation accuracy of clear-sky irradiance.

[0108] ;

[0109] in, The standard deviation of the irradiance or power data within the sliding window. This represents the average value of the irradiance or power data within the sliding window.

[0110] when >0.7 and C v When the value is less than 0.1, the weather type is stable and sunny.

[0111] when <0.3 and C v When the value is less than 0.15, the weather type is uniformly cloudy.

[0112] When the coefficient of variation C v <0.3 and C v When the value is ≥0.15, the weather type is slightly fluctuating and cloudy.

[0113] When the coefficient of variation C v <0.5 and C v When the value is ≥0.3, the weather type is fluctuating moderately cloudy.

[0114] When the coefficient of variation C v When the value is ≥0.5, the weather type is fluctuating and cloudy.

[0115] The above threshold classification method is only a preferred classification method. The specific threshold selection and number of categories can be obtained through various empirical analysis and data analysis methods, such as automatically discovering the characteristic distribution patterns of different weather types through cluster analysis (such as K-means), and obtaining specific weather categories and corresponding classification indicators based on the classification results.

[0116] Method 2: Threshold classification using supervised learning

[0117] Obtain the solar position information and statistical characteristics of irradiance or power data within a sliding window during the current control period. The sliding window includes the current control period, and the statistical characteristics include at least one of the following: volatility index, abrupt change index, stability index, and clear sky index. Input the solar position information and statistical characteristics into a pre-trained prediction model to obtain the weather type output by the prediction model.

[0118] The solar position information for the current control cycle includes solar altitude angle and solar azimuth angle.

[0119] The statistical characteristics of irradiance or power data within a sliding window include, but are not limited to, at least one of the following:

[0120] Volatility indicators: standard deviation, mean absolute deviation, coefficient of variation (standard deviation / mean);

[0121] Mutation index: the maximum value and standard deviation of the difference between adjacent control periods;

[0122] Stability metric: Range within the sliding window (maximum value - minimum value);

[0123] Clear Sky Index: .

[0124] The prediction model is trained in advance using training samples with labeled weather types. The training samples include: solar position information of the sample control period and statistical features of irradiance data or power data within a sliding window containing the sample control period. The training samples can be combined with auxiliary information such as satellite cloud images and ground-based cloud images, and can be manually or semi-automatically labeled using various clustering analysis methods. The labeled weather type labels include, but are not limited to: stable sunny, uniformly cloudy, highly fluctuating cloudy, fluctuating moderately cloudy, and fluctuating slightly cloudy.

[0125] During model training, the solar position information of the sample control period and the statistical features of the irradiance or power data within the sliding window containing the sample control period in each training sample are used as the input feature vector X, and the corresponding weather type label is used as the output Y. The model can be used for lightweight, edge-deployable supervised learning models, such as the following shallow decision trees and lightweight neural networks:

[0126] (1) Shallow decision tree

[0127] The tree depth is limited to 3-5 layers to ensure fast inference speed. It is constructed using the CART algorithm, supports multi-classification tasks, has low computational complexity, and is suitable for real-time operation on edge devices.

[0128] It can select the best splitting features based on information gain or Gini coefficient according to feature indicators, reduce the number of features required for inference on edge devices, recursively build a binary tree until the stopping condition is met (such as depth limit, minimum number of samples in leaf nodes), and use cost complexity pruning to prevent overfitting. Finally, the trained decision tree is transformed into an if-else rule set, which is convenient for edge deployment.

[0129] (2) Lightweight Neural Networks

[0130] The system uses a single hidden layer structure with ≤50 neurons, employs the ReLU activation function to avoid gradient vanishing, and uses Softmax for multi-class classification in the output layer. During the quantization training phase, low-precision numerical values ​​are used to accelerate computation, and redundant connections and neurons are removed through network pruning.

[0131] Supervised learning methods can better learn various features of large-scale meteorological and power generation datasets across multiple climate zones globally compared to empirical threshold models, effectively improving the accuracy of weather classification.

[0132] To adapt to the irradiance fluctuation characteristics of different weather types, the first threshold, the second threshold, and the count value N are different for different weather types. The following provides a method for determining the first threshold, the second threshold, and the count value N for different weather types.

[0133] Please see Figure 2 The flowchart shown illustrates the method for determining the first threshold, the second threshold, and the count value N corresponding to different weather types, including the following steps 201-204:

[0134] 201: Obtain at least one historical sample dataset within the evaluation period. The historical sample dataset includes data from multiple consecutive historical sample days, and the historical sample day data includes data sequences corresponding to more than one weather type.

[0135] To ensure the model with the aforementioned parameters exhibits good generalization ability, the evaluation period must maintain a balanced distribution of sample sizes across seasons. For example, selecting equal amounts of historical sample datasets from each of the four seasons (spring, summer, autumn, and winter) effectively prevents seasonal bias in the parameters. Furthermore, the historical sample dataset must maintain temporal series integrity. When partitioning the historical sample dataset, the temporal continuity of natural weather sequences must be preserved, avoiding random partitioning that could disrupt the inherent patterns of weather evolution. For instance, 30 consecutive days of data can be considered a complete historical sample dataset, ensuring that the natural transitions and continuity of various weather patterns are fully preserved.

[0136] It should be noted that if the photovoltaic site in the actual application scenario only contains data from a specific climate zone, it can be used to train a dedicated model for that climate zone to ensure a high degree of fit between the model and the actual application scenario.

[0137] The above data sequence can be collected using devices such as a Supervisory Control and Data Acquisition (SCADA) system or a tracking bracket communicator (NCU). The collection frequency should be consistent with the control cycle in the above embodiment. The data sequence includes the following:

[0138] Irradiation data: total horizontal irradiance; optionally, scattered irradiance and direct irradiance can also be collected.

[0139] Operational data: Current tracking angle of the photovoltaic tracking bracket; optional, it can also collect the temperature of the module backsheet.

[0140] Power generation data: Power output of the photovoltaic array.

[0141] Optionally, meteorological data such as ambient temperature, wind speed, and relative humidity can be further collected and combined with the module backsheet temperature. Through steady-state or non-steady-state photovoltaic module thermal models and photovoltaic module photoelectric characteristic models, the power generation energy of the photovoltaic module can be calculated more precisely.

[0142] In other words, the above data sequence should include all the data required for the photovoltaic tracking bracket rotation control method provided in the above embodiments.

[0143] Furthermore, a sliding window (e.g., 30 minutes) data buffer can be maintained to update and store the preprocessed data sequence in real time. Data preprocessing includes outlier removal, data smoothing, and feature extraction, providing high-quality input for subsequent analysis.

[0144] 202: Construct the parameter search space corresponding to each weather type. The parameter search space includes: the value range of the first threshold, the value range of the second threshold, and the value range of the count value N.

[0145] For example, the parameter search space for each weather type is shown in the table below:

[0146]

[0147] 203: The objective function is constructed with the goal of maximizing the difference between the power generation gain of the photovoltaic array and the rotational loss of the photovoltaic tracking bracket.

[0148] The power generation gain of the photovoltaic array is the power generation energy gain relative to the second tracking algorithm used, which controls the rotation of the photovoltaic tracking bracket using the first tracking algorithm, within the evaluation period.

[0149] For example, the objective function is as follows:

[0150] ;

[0151] in, For the power generation gain of the photovoltaic array;

[0152] This refers to the rotational loss of the photovoltaic tracking bracket.

[0153] Specifically, the evaluation period refers to the power generation gain brought about by using the first tracking algorithm to control the rotation of the photovoltaic tracking bracket, compared to using the second tracking algorithm. The specific quantification process is as follows:

[0154] ;

[0155] In the formula, The power generation energy of the first tracking algorithm after using the smooth rotation control method in the above embodiments in time period t. Let T be the power generation energy of the second tracking algorithm in time period t, where T is the total number of time periods within the evaluation period.

[0156] It can be a comprehensive index composed of data such as the number of rotations, total rotation angle, and motor energy consumption (motor energy consumption can be modeled and calculated based on existing load models, real-time angles, and rotation angles).

[0157] For example, Calculations can be performed by fusing multiple indicators:

[0158] ;

[0159] In the formula, The number of rotations, The total rotation angle, For motor energy consumption, , , The normalized weights for each indicator are given, and each weight coefficient reflects the relative importance of the corresponding indicator in the assessment of rotational loss.

[0160] Each weight coefficient satisfies the normalization condition: ,and , .

[0161] In addition to power generation gain and rotational losses, the objective function can be further optimized to improve its overall performance. For example, system stability indicators and equipment lifespan factors can be incorporated into the objective function through weighted combination.

[0162] ;

[0163] in, As a system stability indicator, it can be constructed based on parameters such as mode switching frequency and standard deviation of angle fluctuation, and is used to evaluate the smooth operation of the control system and avoid frequent state jumps.

[0164] The equipment life factor can be constructed based on at least one parameter, such as the cumulative wear of the rotating mechanism and the influence parameters of motor temperature rise. The cumulative wear of the rotating mechanism can be estimated based on the rotation angle and load model. The equipment life factor is used to evaluate long-term operational reliability.

[0165] The above rotational loss weighting coefficient ( , , The objective function weight coefficients (α, β, γ, δ) can be determined through, but are not limited to, one of the following methods:

[0166] 1) Subjective weighting method: Based on expert experience, it reflects the relative importance of each indicator in mechanical loss;

[0167] 2) Objective weighting method: Based on statistical analysis of historical data, it can automatically allocate weights according to the degree of variation of each indicator;

[0168] 3) Subjective and objective weighting method: The subjective weights and objective weights are weighted and integrated, and the optimal combination coefficients are determined through an optimization algorithm;

[0169] 4) Determine the method by debugging in a simulation environment through trial and error.

[0170] 204: Using at least one historical sample dataset within the evaluation period, simulate the photovoltaic tracking bracket rotation control method, and with the objective function maximization as the optimization objective, solve for the first threshold, the second threshold, and the count value N corresponding to each weather type.

[0171] After determining the objective function, weighting coefficients, and the parameter search space for each weather type, a suitable optimization algorithm can be selected to determine the control parameters (i.e., the first threshold, the second threshold, and the count value N) that maximize the objective function under different weather types. Specific optimization algorithms that can be used, but are not limited to, are as follows:

[0172] Grid search method: This method establishes a three-dimensional parameter space and systematically tests the three key control parameters—the first threshold, the second threshold, and the count value N. In practice, firstly, a reasonable parameter value step size is set for each weather type based on the parameter search space for each control parameter. Then, the comprehensive score of each combination of control parameters is calculated point-by-point within the parameter search space, and finally, the optimal parameter combination is selected. Although this method involves relatively large computational loads, it ensures that the optimal solution is found within the parameter search space, and the process is intuitive and easy to understand.

[0173] Genetic Algorithm: This method simulates the biological evolution process to optimize parameters. First, the control parameter combination [T1, T2, N] is encoded as a binary string or real-number vector. The objective function is used as the fitness value, and through selection, crossover, and mutation, the algorithm gradually evolves to the optimal control parameter combination. This adaptive optimization method is particularly suitable for handling complex nonlinear optimization problems, effectively avoiding getting trapped in local optima and reducing computational cost.

[0174] The parameter optimization process can employ a full-cycle simulation method. For each candidate parameter combination, the smooth control process is reproduced daily on the training set, and the objective function value is calculated based on the simulation results. By comparing the average performance of each parameter combination on the training set, a preliminary optimal parameter solution is determined.

[0175] A rigorous generalization capability verification mechanism is established, and the initially optimal parameters are applied to an independent validation set for performance evaluation. If the validation set performance meets expectations, the parameter combination is confirmed as the final optimization result; if the performance does not meet expectations, the training process is reviewed or the parameter search strategy is adjusted to ensure the reliability of the optimization result.

[0176] The test set is specifically used to evaluate the actual effectiveness of the optimal control parameters and verify their performance on completely unknown data. Through this complete optimization process, the optimal combination of control parameters [T1, T2, N] for each weather condition is finally determined, providing a fully validated parameter configuration scheme for the photovoltaic tracking system.

[0177] Furthermore, once sufficient local operational data has been accumulated (e.g., data covering 1-3 months of the typical local season), an offline optimization process is initiated to make the parameter tables more closely match the actual climatic conditions of the power plant. Please refer to [link / reference]. Figure 3 Weather type prediction models and control parameters can be trained offline in the cloud and then deployed on edge devices, such as controllers, for online application.

[0178] Specifically, after acquiring local operating data for a preset duration, the local operating data is processed to obtain an actual sample dataset. The photovoltaic tracking bracket rotation control method is then executed using this actual sample dataset, with the objective function maximization as the optimization goal. Candidate first thresholds, candidate second thresholds, and candidate count values ​​N are solved for each weather type. Simulations are performed using historical sample datasets or the actual sample dataset to compare the first and second control parameter combinations and determine the optimal control parameter combination. The first control parameter combination includes a first threshold, a second threshold, and a count value N; the second control parameter combination includes candidate first thresholds, candidate second thresholds, and candidate count values ​​N. Finally, the photovoltaic tracking bracket rotation control method provided in this application embodiment is executed using the optimal control parameter combination.

[0179] When switching control parameters, in order to ensure the stability of power generation before and after the switch, the system maintains two parameter tables, one old and one new. After observing the short-term stable operation, the new table is officially put into use to ensure a disturbance-free update.

[0180] Furthermore, version control can be implemented on the parameter table, recording the time and data basis of each optimization, which facilitates subsequent tracking and rollback.

[0181] This process can be repeated periodically (e.g., every six months or every year), using the ever-accumulating new data to continuously evolve the system's control strategies and adapt to the long-term changes and subtle characteristics of the local climate.

[0182] Compared to conventional tracking algorithms based on a fixed schedule, this solution is expected to reduce the overall number of rotations by 40% to 60% while maintaining continuous power generation gain (>95%). On cloudy days, the number of rotations can be reduced to less than 10% of that of conventional tracking algorithms, thereby effectively increasing the power generation gain of the photovoltaic tracking system and significantly extending its mechanical life.

[0183] Based on historical operating data of a power plant during a typical cloudy day in January 2025, the comprehensive performance of the photovoltaic tracking bracket rotation control method provided in this application embodiment under cloudy weather conditions was verified using a digital twin simulation platform.

[0184] Typical cloudy weather data covers various cloudy weather types with varying degrees of fluctuation. The theoretical tracking angle of the first tracking algorithm of the photovoltaic tracking bracket rotation control method provided in this application embodiment was not used; that is, the theoretical tracking angle of the first tracking algorithm before smoothing optimization was not used. Figure 4 As shown, the theoretical tracking angle of the first tracking algorithm using the photovoltaic tracking bracket rotation control method provided in this application embodiment, that is, the theoretical tracking angle of the first tracking algorithm after smoothing optimization, is as follows: Figure 5 As shown.

[0185] The second tracking algorithm on a typical cloudy day rotates 150 times (the tracking accuracy of the tracking bracket is 1°). The algorithm angle change is affected by drastic fluctuations in cloudy weather. Before smoothing optimization, the number of rotations was 273. After optimization using the photovoltaic tracking bracket rotation control method provided in this application embodiment, the number of rotations decreased to 75 times, which is 50% of the second tracking algorithm and 27.5% of the first tracking algorithm before smoothing optimization. Furthermore, using the photovoltaic tracking bracket rotation control method provided in this application embodiment, the power generation gain of the first tracking algorithm after smoothing optimization is 96.51% of that before optimization.

[0186] This example effectively verifies the technical advantages of the photovoltaic tracking bracket rotation control method provided in this application embodiment under complex weather conditions (such as turbulent cloudy weather). While maintaining the power generation gain, it effectively reduces the number of rotations and shows good adaptability to cloudy weather with different fluctuation intensities.

[0187] This application also provides a controller in its embodiments. (See reference...) Figure 6 As shown, it illustrates a structural schematic diagram suitable for implementing the controller in the embodiments of this application. Figure 6 The controller shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0188] like Figure 6 As shown, the controller may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. When the controller is powered on, the RAM 603 also stores various programs and data required for controller operation. The processing device 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0189] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 607 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 608 including, for example, memory card, hard disk, etc.; and communication devices 609. Communication device 609 allows the controller to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 A controller with various devices is shown; however, it should be understood that implementation or possession of all the devices shown is not required. More or fewer devices may be implemented alternatively.

[0190] This application embodiment also provides a photovoltaic tracking bracket rotation control system, including: a controller and a photovoltaic tracking bracket drive device;

[0191] The controller is communicatively connected to the photovoltaic tracking bracket drive device;

[0192] The photovoltaic tracking bracket driving device is used to drive the photovoltaic tracking bracket to rotate, and the photovoltaic tracking bracket is equipped with photovoltaic panels.

[0193] The controller can implement any of the photovoltaic tracking bracket rotation control methods provided in the embodiments of this application.

[0194] This application also provides a computer program product including computer-readable instructions, which, when executed on a controller, cause the controller to implement any of the photovoltaic tracking bracket rotation control methods provided in this application.

[0195] This application also provides a computer-readable storage medium carrying one or more computer programs. When the one or more computer programs are executed by a controller, the controller can implement any of the photovoltaic tracking bracket rotation control methods provided in this application.

[0196] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0197] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0198] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0199] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for controlling the rotation of a photovoltaic tracking bracket, characterized in that, include: Obtain the tracking angle of the photovoltaic tracking bracket and calculate the first power generation energy corresponding to the tracking angle; The first tracking algorithm is used to calculate the first tracking angle of the photovoltaic array in the next control cycle, and the second power generation energy corresponding to the first tracking angle is calculated. The input data of the first tracking algorithm includes at least the solar position and irradiance data. Calculate the first energy difference between the second generated energy and the first generated energy; If the first energy difference is greater than the first threshold for N consecutive control cycles, the second tracking algorithm is used to calculate the second tracking angle for the next control cycle, and the third power generation energy corresponding to the second tracking angle is calculated. The input data of the second tracking algorithm includes at least the sun position, and N is an integer greater than 1. Calculate the second energy difference between the second generated energy and the third generated energy; If the second energy difference is greater than the second threshold, control the photovoltaic tracking bracket to rotate to the first tracking angle; When the second energy difference is less than or equal to the second threshold, the photovoltaic tracking bracket is controlled to rotate to the second tracking angle.

2. The photovoltaic tracking bracket rotation control method according to claim 1, characterized in that, The first threshold, the second threshold, and the count value N are determined based on the weather type of the current control cycle.

3. The photovoltaic tracking bracket rotation control method according to claim 2, characterized in that, The method further includes: Obtain statistical characteristics of irradiation data or power data within a sliding window, wherein the sliding window includes the current control cycle, and the statistical characteristics include at least one statistical indicator; Based on the statistical indicators in the statistical features and the correspondence between the pre-configured statistical indicators and weather types, the weather type of the current control cycle is determined.

4. The photovoltaic tracking bracket rotation control method according to claim 2, characterized in that, The method further includes: Acquire the solar position information and statistical characteristics of irradiance data or power data within the sliding window of the current control cycle. The sliding window includes the current control cycle, and the statistical characteristics include at least one of the following: volatility index, abrupt change index, stability index, and clear sky index. The solar position information and the statistical features are input into a pre-trained prediction model to obtain the weather type of the current control cycle output by the prediction model. The prediction model is trained in advance using training samples with labeled weather types.

5. The photovoltaic tracking bracket rotation control method according to claim 2, characterized in that, The method for determining the first threshold, the second threshold, and the count value N corresponding to different weather types includes: Obtain at least one historical sample dataset within the evaluation period, wherein the historical sample dataset includes data from multiple consecutive historical sample days, and the historical sample day data includes data sequences corresponding to one or more weather types; Construct a parameter search space corresponding to each weather type, wherein the parameter search space includes: the value range of the first threshold, the value range of the second threshold, and the value range of the count value N; The objective function is constructed with the goal of maximizing the difference between the power generation gain of the photovoltaic array and the rotation loss of the photovoltaic tracking bracket. The power generation gain of the photovoltaic array is the power generation energy gain relative to the second tracking algorithm and the first tracking algorithm controlling the rotation of the photovoltaic tracking bracket within the evaluation period. The photovoltaic tracking bracket rotation control method is simulated using at least one historical sample dataset within the evaluation period. With the objective function as the optimization objective, the first threshold, the second threshold, and the count value N corresponding to each weather type are solved.

6. The photovoltaic tracking bracket rotation control method according to claim 5, characterized in that, The method further includes: Given local running data of a preset duration, the local running data is processed to obtain an actual sample dataset; The photovoltaic tracking bracket rotation control method is executed using the actual sample dataset. With the objective function as the optimization objective, the candidate first threshold, candidate second threshold, and candidate count value N corresponding to each weather type are solved. The first control parameter combination and the second control parameter combination are compared using the historical sample dataset or the actual sample dataset to determine the optimal control parameter combination. The first control parameter combination includes: the first threshold, the second threshold, and the count value N. The second control parameter combination includes: the candidate first threshold, the candidate second threshold, and the candidate count value N. The photovoltaic tracking bracket rotation control method is executed using the optimal combination of control parameters.

7. A controller, characterized in that, include: It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the controller can implement the photovoltaic tracking bracket rotation control method as described in any one of claims 1 to 6.

8. A photovoltaic tracking bracket rotation control system, characterized in that, include: Controller and photovoltaic tracking bracket drive equipment; The controller is communicatively connected to the photovoltaic tracking bracket drive device; The photovoltaic tracking bracket driving device is used to drive the photovoltaic tracking bracket to rotate, and the photovoltaic tracking bracket is equipped with photovoltaic panels. The controller is capable of implementing the photovoltaic tracking bracket rotation control method as described in any one of claims 1 to 6.

9. A computer program product, characterized in that, Includes computer-readable instructions that, when executed on a controller, cause the controller to implement the photovoltaic tracking bracket rotation control method as described in any one of claims 1 to 6.

10. A computer storage medium, characterized in that, The storage medium carries one or more computer programs that, when executed by the controller, enable the controller to implement the photovoltaic tracking bracket rotation control method as described in any one of claims 1 to 6.

Images