Preregulation-based photovoltaic support tracking method and device

Abstract

The invention provides a photovoltaic support tracking method and device based on pre-regulation, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring predicted environment data of a photovoltaic power station within a preset time, and inputting the predicted environment data into a pre-trained prediction model to obtain an initial support adjustment strategy; and based on the predicted environment data and the initial support adjustment strategy, optimizing a pre-constructed target function to obtain a target support adjustment strategy. According to the technical scheme, the globally optimal target support adjustment strategy can be planned in advance before support deployment, the generating capacity of the photovoltaic power station within the preset time is maximized, and the influence of the support initial adjustment stage on the generating capacity is reduced.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of photovoltaic cell technology, and in particular to a photovoltaic support control method, device, computer equipment, computer-readable storage medium, and computer program product based on pre-planning. Background Art

[0002] With the popularity of solar power generation around the world, photovoltaic power stations have increasingly higher requirements for the power generation of photovoltaic modules. Among them, photovoltaic modules can use fixed brackets and tracking brackets. The angle of the fixed bracket is pre-set according to the longitude and latitude of the photovoltaic power station and cannot be adjusted. The tracking bracket can dynamically adjust the angle according to the position of the sun to maximize the capture of solar radiation. Its power generation is usually 10~15% higher than that of the fixed bracket.

[0003] However, the algorithms used by tracking brackets have problems such as long initial learning cycle, poor adaptability and obvious lag, and perform poorly in complex and harsh environments, resulting in a short service life of the bracket and reduced power generation of photovoltaic modules.

[0004] It should be noted that the above content is not necessarily prior art, nor is it intended to limit the scope of patent protection of this application. Summary of the invention

[0005] The embodiments of the present application provide a photovoltaic support tracking method, apparatus, computer equipment, computer-readable storage medium, and computer program product based on pre-planning to solve or alleviate one or more of the technical problems raised above.

[0006] One aspect of an embodiment of the present application provides a photovoltaic support tracking method based on pre-regulation, the method comprising: Acquire predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes a plurality of time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit; Based on the predicted environmental data and the initial support adjustment strategy, optimizing the pre-constructed objective function to obtain a target support adjustment strategy; Among them, the objective function is used to calculate the power generation of the photovoltaic power station within the preset time, and the target bracket adjustment strategy is used to indicate: dividing the preset time into target time units and non-target time units, executing the cleaning strategy in the target time units, and executing the initial bracket adjustment strategy in the non-target time units.

[0007] Optionally, the predicted environmental data includes predicted meteorological data, and the predicted meteorological data includes predicted wind data and predicted precipitation data; The execution of the cleaning strategy includes: Based on the predicted environmental data, obtaining predicted meteorological data corresponding to the target time unit to determine predicted wind data and predicted precipitation data corresponding to the target time unit; Determining the wind direction and precipitation corresponding to the target time unit based on the predicted wind force data and the predicted precipitation data corresponding to the target time unit; determining a cleaning angle of photovoltaic components in the photovoltaic power station according to the wind direction and precipitation corresponding to the target time unit; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the cleaning angle to clean the photovoltaic components in the photovoltaic power station.

[0008] Optionally, the power generation of the photovoltaic power station within the preset time is calculated by the following operations: According to the predicted environmental data, obtaining a shielding factor of the photovoltaic power station in each of the time units; According to the predicted environmental data and the target bracket adjustment strategy, obtaining the solar radiation amount of the photovoltaic power station in each of the time units; Determining the power generation of the photovoltaic power station in each of the time units according to the shielding factor and the solar radiation of the photovoltaic power station in each of the time units; The power generation of the photovoltaic power station within the preset time is determined according to the power generation of the photovoltaic power station within each of the time units.

[0009] Optionally, the photovoltaic support tracking method based on pre-regulation further includes: Get the real-time environment data of the current time unit; Generate a support tracking strategy based on the real-time environmental data of the current time unit and the target support adjustment strategy; The support tracking strategy is used to adjust the support angle of the photovoltaic components in the photovoltaic power station in the current time unit.

[0010] Optionally, the photovoltaic support tracking method based on pre-regulation further includes: The support tracking strategy is corrected by using a pre-trained transmission error model to obtain a target support angle; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the target support angle.

[0011] Optionally, the prediction model is obtained by the following operations: Acquiring historical environmental data, wherein the historical environmental data includes historical meteorological data, historical astronomical data and historical topographic data; Based on the historical meteorological data, the historical astronomical data and the historical terrain data, constructing a first prediction model, a second prediction model and a third prediction model in a one-to-one correspondence; The first prediction model, the second prediction model and the third prediction model are integrated to obtain the prediction model.

[0012] Optionally, based on the historical meteorological data, the astronomical meteorological data and the historical terrain data, a first prediction model, a second prediction model and a third prediction model are constructed in a one-to-one correspondence, including: Acquire multiple pieces of the historical meteorological data, and merge the multiple pieces of the historical meteorological data into a wide table, wherein each column of the wide table represents a feature, each row of the wide table represents a sample, the sample includes feature values ​​corresponding to multiple features, and each sample corresponds to one piece of the historical meteorological data; Preprocessing the wide table, the preprocessing comprising: deleting outlier eigenvalues, filling missing eigenvalues, synthesizing samples and adding the synthesized samples to the wide table; Encode the feature values ​​of the preprocessed wide table and perform feature screening; Based on the screened features and corresponding feature values, the first prediction model is constructed.

[0013] Optionally, the wide table further includes a label column, and the label column has multiple label values; The synthesizing sample and adding the synthesized sample to the wide table comprises: Obtaining the number of samples corresponding to different label values, determining a first type of label value and a second type of label value, wherein the number of samples corresponding to the first type of label value is less than the number of samples corresponding to the second type of label value; When the number of samples corresponding to the first category label value is less than the number of samples corresponding to the second category label value: determine a first sample from the samples corresponding to the first category label value, and determine a plurality of candidate samples whose similarity to the first sample is within a preset range; determine a second sample from the plurality of candidate samples; and generate a synthetic sample corresponding to the first category label value based on the second sample and the first sample.

[0014] Another aspect of the embodiments of the present application provides a photovoltaic support tracking device based on pre-regulation, the device comprising: A first pre-regularization module is used to obtain predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes a plurality of time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit; A second pre-regularization module is used to optimize the pre-constructed objective function based on the predicted environmental data and the initial support adjustment strategy to obtain a target support adjustment strategy; Among them, the objective function is used to calculate the power generation of the photovoltaic power station within the preset time; the target bracket adjustment strategy is used to indicate: dividing the preset time into target time units and non-target time units, executing the cleaning strategy in the target time units, and executing the initial bracket adjustment strategy in the non-target time units.

[0015] Another aspect of an embodiment of the present application provides a computer device, including: at least one processor; and a memory communicatively coupled to the at least one processor; Wherein: the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0016] Another aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method described above is implemented.

[0017] Another aspect of an embodiment of the present application provides a computer program product, including a computer program, which implements the method described above when executed by a processor.

[0018] The above technical solution adopted in the embodiment of the present application may have the following advantages: By predicting environmental data and pre-trained prediction models, the initial bracket adjustment strategy of the photovoltaic power station is generated. This strategy can plan the bracket angle of the photovoltaic components in each time unit within the preset time in advance. Based on the predicted environmental data and the initial bracket adjustment strategy, with the goal of maximizing the power generation of the photovoltaic power station within the preset time, the target bracket adjustment strategy is further optimized. This strategy is used to indicate: divide the preset time into target time units and non-target time units, execute the cleaning strategy in the target time unit, and execute the initial bracket adjustment strategy in the non-target time unit.

[0019] It can be seen that the embodiment of the present application can pre-plan the future long-term bracket adjustment strategy before the bracket is deployed, and select the bracket adjustment strategy based on the predicted environmental data to obtain the global optimal target bracket adjustment strategy. This can maximize the power generation of the photovoltaic power station within the preset time, reduce the impact of the initial bracket adjustment stage on the power generation, and adapt to various complex and harsh environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The embodiments shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0021] Figure 1 A flowchart of a photovoltaic support tracking method based on pre-regulation according to Embodiment 1 of the present application is schematically shown; Figure 2 The flowchart of constructing a prediction model according to the first embodiment of the present application is schematically shown; Figure 3 The modeling process of the prediction model according to the first embodiment of the present application is schematically shown; Figure 4 Schematically shows Figure 2 Flow chart of sub-steps of step S202; Figure 5 A flowchart of executing a cleaning strategy according to Embodiment 1 of the present application is schematically shown; Figure 6 A flowchart for calculating the estimated power generation of a photovoltaic power station within a preset period according to the first embodiment of the present application is schematically shown; Figure 7 This is an application example diagram of the photovoltaic support tracking method based on pre-regulation according to the first embodiment of the present application; Figure 8 A block diagram of a photovoltaic support tracking device based on pre-regulation according to the second embodiment of the present application is schematically shown; and Fig. 9 The hardware architecture diagram of the computer device according to the third embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0023] It should be noted that the descriptions involving "first", "second", etc. in the embodiments of the present application are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0024] In the description of the present application, it should be understood that the numerical labels before the steps do not indicate the order in which the steps are executed, but are only used to facilitate the description of the present application and to distinguish each step, and therefore should not be understood as a limitation on the present application.

[0025] First, the following terms are explained: JSON (JavaScript Object Notation): A data exchange format commonly used in network transmission.

[0026] MQTT (Message Queuing Telemetry Transport): A lightweight IoT communication protocol.

[0027] OpenCV (Open Source Computer Vision Library): A lightweight and efficient cross-platform computer vision and machine learning software library.

[0028] Python: An object-oriented computer programming language.

[0029] PVLIB: An open source library for solar photovoltaic system design and performance simulation.

[0030] Optuna: An automatic hyperparameter optimization framework.

[0031] The embodiment of the present application provides a photovoltaic support tracking technology solution based on pre-planning. In this technical solution: (1) A complete closed loop of collection, training, and updating is set. According to the multi-dimensional input (historical environmental data, predicted environmental data) provided by the big data platform, the spacing of photovoltaic modules is planned in advance to maximize the power generation of the power station. The support adjustment strategy is calculated and set in advance to reduce the impact of the initial support adjustment stage on the power generation. (2) A monitoring module is designed to monitor the power generation power and the environment of the current photovoltaic module, so as to dynamically track and adjust the support. The power of photovoltaic modules is positively correlated with the power generation of the power station. Based on the pre-planned support adjustment strategy, the power generation can be estimated in the early stage of power station construction, and the actual power generation can be monitored during the operation of the power station to determine the deviation between the estimated power generation and the actual power generation. The deviation can be used to optimize the subsequent support adjustment strategy. (3) It has a wide range of applications and strong generalization ability, and can adapt to different weather, terrain, altitude, time, support spacing and support types. (4) It has high precision, relying on the fusion of multiple models and the transmission error model as a correction tool for the execution step. Targeted data collection in different scenarios is used to organize and proportion data reasonably, and this data is used as training input for multiple models, which greatly improves the adaptability and accuracy of the models in multiple scenarios. (5) Before the deployment of the bracket, the long-term illumination distribution is statistically calculated to preview the lighting conditions (equivalent to predicting environmental data), and the bracket adjustment angle for the long term in the future is pre-calculated, which greatly improves the pre-regulation capability of the bracket adjustment and reduces the computing and communication costs. (6) Extensive collection of astronomical, meteorological, topographic and elevation information is used to establish meteorological models, astronomical models and terrain models using machine learning and deep learning technologies. According to the bracket spacing and current bracket angle of different power stations, the current longitude and latitude of the bracket, the earth's rotation inclination and the orbit of revolution are collected to establish a sun-bracket position relationship model, which significantly improves the response effect of the bracket adjustment algorithm, as well as the adaptability and accuracy in different time, climate and terrain scenarios. At the same time, by calculating the distribution of mechanical transmission errors, a compensation algorithm for mechanical errors is implemented to improve the accuracy of the overall photovoltaic bracket tracking algorithm when the bracket is rotated. See below for details.

[0032] The technical solutions of the present application are described below through multiple embodiments. It should be noted that these embodiments can be implemented in a variety of different forms and should not be construed as being limited to the embodiments described here.

[0033] Embodiment 1 Figure 1 A flowchart of a photovoltaic support tracking method based on pre-planning according to Embodiment 1 of the present application is schematically shown.

[0034] like Figure 1 As shown, the photovoltaic support tracking method based on pre-planning may include steps S100 to S102, wherein: Step S100, obtaining predicted environmental data of the photovoltaic power station within a preset time, inputting the predicted environmental data into a pre-trained prediction model, and obtaining an initial bracket adjustment strategy; wherein the preset time includes multiple time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit.

[0035] Step S102: Based on the predicted environmental data and the initial support adjustment strategy, the pre-constructed objective function is optimized to obtain a target support adjustment strategy.

[0036] The objective function is used to calculate the power generation of the photovoltaic power station within the preset time. The target bracket adjustment strategy is used to indicate: the preset time is divided into a target time unit and a non-target time unit, the cleaning strategy is executed in the target time unit, and the initial bracket adjustment strategy is executed in the non-target time unit.

[0037] The photovoltaic bracket tracking method based on pre-planning provided in this embodiment generates an initial bracket adjustment strategy for the photovoltaic power station by predicting environmental data and a pre-trained prediction model. This strategy can plan in advance the bracket angle of the photovoltaic module in each time unit within the preset time. Based on the predicted environmental data and the initial bracket adjustment strategy, with the goal of maximizing the power generation of the photovoltaic power station within the preset time, the target bracket adjustment strategy is further optimized. The strategy is used to indicate: divide the preset time into target time units and non-target time units, execute the cleaning strategy in the target time unit, and execute the initial bracket adjustment strategy in the non-target time unit. It can be seen that the embodiment of the present application can pre-plan the future long-term bracket adjustment strategy before the bracket is deployed, and select the bracket adjustment strategy based on the predicted environmental data to obtain the globally optimal target bracket adjustment strategy. In this way, the power generation of the photovoltaic power station within the preset time can be maximized, the impact of the initial bracket adjustment stage on the power generation can be reduced, and various complex and harsh environments can be adapted.

[0038] The following combination Figure 1 , each step in steps S100~S102 and other optional steps are explained in detail.

[0039] Step S100 , obtain predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes multiple time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit.

[0040] The preset time can be the entire life cycle of the photovoltaic power station, or in units of years, months, weeks, or days, such as ten years, one year, half a year, one month, etc. The preset time and time unit can be set according to the actual situation. For example, if the preset time is set to one year, the time unit can be every day of the year. If the preset time is set to one day, the time unit can be every hour of the day.

[0041] Environmental data may include meteorological data, topographic data, and astronomical data. Meteorological data may include cloud cover, wind force, precipitation, humidity, air pressure, surrounding sunshine, weather, etc. Topographic data may include elevation (altitude), slope, occlusion, etc. Astronomical data may include sun angle, longitude and latitude, rotation angle, etc. In practical applications, topographic data can be obtained by using drones to map the location of photovoltaic power stations. Meteorological data can be collected by pre-arranged meteorological sensor arrays and sky cameras. Astronomical data can be obtained from the National Comprehensive Earth Observation Data Sharing Platform (CGEO). These collected environmental data can be saved to a preset database or big data platform. Historical environmental data may be real environmental data from the past, which can be obtained through laboratories, field measurement equipment, or other built photovoltaic power stations. Predicted environmental data may be environmental data that may appear in the future based on historical environmental data. For example, statistics on the illumination distribution of the location of the light-emitting power station to be built in the past year can be used to preview the illumination conditions in the next year.

[0042] The prediction model can be trained with a large amount of historical environmental data to predict the optimal bracket angle of photovoltaic modules in different environments (meteorological, astronomical, and topographical). There are many ways to build a prediction model, such as: single modeling, combined modeling, hierarchical modeling, hybrid modeling, etc. An exemplary solution is provided below.

[0043] In an optional embodiment, if Figure 2 As shown, the prediction model can be obtained by the following operations: Step S200, obtaining historical environmental data, wherein the historical environmental data includes historical meteorological data, historical astronomical data and historical terrain data.

[0044] Step S202, based on the historical meteorological data, the historical astronomical data and the historical terrain data, construct a first prediction model, a second prediction model and a third prediction model in a one-to-one correspondence.

[0045] Step S204: fusing the first prediction model, the second prediction model and the third prediction model to obtain the prediction model.

[0046] In this embodiment, each type of environmental data is modeled separately, which can better learn different types of data features. Fusion and integration of multiple prediction models can effectively reduce the overfitting risk of a single prediction model, enhance the robustness and accuracy of the prediction model, and improve the overall prediction performance.

[0047] The following will be combined Figure 3 The modeling process of the prediction model is further introduced.

[0048] For example, some national comprehensive earth observation data (historical astronomical data), historical terrain data, and historical meteorological data can be extracted from the big data platform. This embodiment performs separate modeling of different types of environmental data, and the modeling method and process are reusable. The following will take historical meteorological data as an example to exemplify the modeling method and process.

[0049] In an optional embodiment, if Figure 4 As shown, step S202 may include: Step S400, obtain multiple pieces of the historical meteorological data, and merge the multiple pieces of the historical meteorological data into a wide table, each column of the wide table represents a feature, each row of the wide table represents a sample, the sample includes feature values ​​corresponding to multiple features, and each sample corresponds to one piece of the historical meteorological data.

[0050] Step S402, preprocessing the wide table, the preprocessing comprising: deleting outlier eigenvalues, filling missing eigenvalues, synthesizing samples and adding the synthesized samples to the wide table.

[0051] Step S404, encoding the preprocessed feature values ​​of the wide table and performing feature screening.

[0052] Step S406: construct the first prediction model based on the screened features and corresponding feature values.

[0053] Exemplarily, multiple pieces of historical meteorological data may be acquired and merged to form a wide table, as shown in Table 1.

[0054]

[0055] Table 1 According to Table 1, each column of the wide table can represent a (meteorological) feature, and each row of the wide table can represent a sample. The sample is composed of eigenvalues ​​of different features, and each sample corresponds to a piece of historical meteorological data. In practical applications, the wide table can also be preprocessed, for example: deleting outlier eigenvalues ​​(outliers), filling missing eigenvalues, synthesizing samples and adding the synthesized samples to the wide table, etc., to clean the data, filter noise, avoid sample skew, improve the quality and reliability of the data, and thus improve the prediction ability of the model.

[0056] There are many ways to fill missing eigenvalues, such as forward filling, that is, filling with the previous non-missing eigenvalue of the missing eigenvalue, and the previous non-missing eigenvalue can come from the previous sample. Forward filling is simple and easy, reducing data processing time. Another example: average value filling, the application example is as follows: If the eigenvalues ​​of a column of features show a bimodal distribution, the data can be binned, and the outlier eigenvalues ​​(outliers) can be reassigned using the average values ​​of each box. Average value filling can effectively reduce data deviation. There is also: hot card filling, that is, extracting the eigenvalues ​​of the corresponding features in similar samples to fill the missing eigenvalues. Hot card filling can make the filled data close to the original data distribution. In this embodiment, different filling methods can be selected according to the distribution characteristics of other eigenvalues ​​in the same column as the missing eigenvalues ​​to better improve the data and help model building.

[0057] Outlier eigenvalues ​​can be determined by standard deviation, density-based outlier detection, distance-based outlier detection, machine learning, etc. For example, the KNN model can be used to clean outliers in the data (eigenvalues), and the formula is as follows: , where x is used to represent the sample. This method does not rely on specific distribution assumptions of the data and can effectively identify outlier eigenvalues ​​in wide tables. First, determine the number K of the nearest neighbors of the target defect value. Then, for each eigenvalue in the same column, calculate its distance to all other eigenvalues ​​and find the K nearest points. Finally, based on the distance relationship between these K eigenvalues ​​and the target eigenvalue, determine whether the target eigenvalue is an outlier eigenvalue. After finding the outlier eigenvalue, you can determine the preprocessing method according to actual needs, such as deleting, correcting, or replacing the outlier eigenvalue to improve the data distribution and reduce the risk of overfitting.

[0058] In some embodiments, preprocessing may also include: converting date-type feature values ​​into month values ​​from now, which can make discrete values ​​continuous and significantly reduce dimensions to reduce computational complexity and simplify the model.

[0059] In an optional embodiment, as shown in Table 1, the wide table may further include a label column, and the label column may have multiple label values. The synthesized sample and adding the synthesized sample to the wide table may include: obtaining the number of samples corresponding to different label values, determining the first category label value and the second category label value, the number of samples corresponding to the first category label value is less than the number of samples corresponding to the second category label value; when the number of samples corresponding to the first category label value is less than the number of samples corresponding to the second category label value: determining a first sample from the samples corresponding to the first category label value, and determining a plurality of samples to be selected whose similarity to the first sample is within a preset range; determining a second sample from the plurality of samples to be selected; and generating a synthesized sample corresponding to the first category label value based on the second sample and the first sample.

[0060] In practical applications, samples can be classified according to different label values ​​(such as A, B, C, D), and the number of samples corresponding to each label value can be obtained to determine the first-class label value (minority class) and the second-class label value (majority class). Randomly select a sample (first sample) from the minority class and find the K nearest neighbors of the sample. These neighbors are other minority class samples that are closest to the selected sample or have the highest similarity to the selected sample. Randomly select a neighbor as the second sample, and then randomly generate a new sample on the line between the selected first sample and the second sample. This sample is the synthetic sample and belongs to the minority class. Repeat the above steps until a sufficient number of synthetic samples are generated, so that the number of samples in the minority class and the majority class is more balanced, avoiding the risk of overfitting caused by sample skew. The method of synthesizing samples can be expressed by the following formula: . The synthetic sample x new is the first sample Xi and the second sample x n Linear interpolation between n is x i λ is a random number between 0 and 1, which is used to determine the position of the synthetic sample between the two original samples (the first sample and the second sample).

[0061] Accordingly, step S404: encoding the feature values ​​of the preprocessed wide table and performing feature screening may include: Determine the distribution of eigenvalues ​​for each column of features. In the case of continuous eigenvalues, such as feature 4 "precipitation" in Table 1, the eigenvalues ​​can be normalized to ensure consistent dimensionality. In the case of large differences in the distribution of eigenvalues, such as "precipitation", the eigenvalues ​​can be logarithmically transformed to make them normally distributed, closer to the numerical distribution preset by the parameter model. For features with many values, such as "weather", the values ​​with a proportion lower than the preset threshold can be classified as other eigenvalues ​​others to reduce the feature dimension. Use Ordinal sequence encoding for low-cardinality discrete eigenvalues ​​and Count counting encoding for high-cardinality discrete eigenvalues ​​to avoid dimensionality explosion. You can also perform one-hot encoding on the eigenvalues ​​so that the model can better understand and utilize categorical features, thereby improving the performance of the machine learning model. After feature encoding, you can adjust and filter to reduce the dimension. For example, you can use the chi-square test and Pearson correlation coefficient to calculate the correlation between discrete and continuous features and exclude redundant features. The information entropy of the features is sorted using the random forest as the base model, and finally an agreed number (such as 10) of features whose importance exceeds the threshold are selected.

[0062] Accordingly, step S406: constructing the first prediction model based on the filtered features and the corresponding feature values ​​may include: constructing the first prediction model based on the filtered features and the corresponding feature values. The first prediction model may be a LightGBM model: using a gradient boosting algorithm, a series of decision trees are iteratively trained and combined to obtain a powerful prediction model. The construction process of the second prediction model and the third prediction model may refer to the construction process of the first prediction model above, and finally obtain the following Figure 3 The three LightGBM models shown.

[0063] In practical applications, in order to reduce the variance of the model and improve the predictive ability and stability of the model, the first prediction model, the second prediction model and the third prediction model can be directly superimposed or integrated through weighted mixing to obtain a prediction model with more powerful performance.

[0064] Direct superposition can be expressed by the following formula: .

[0065] Weighted mixing can be expressed by the following formula: .

[0066] Wherein, P is the prediction model, pi is used to represent the first prediction model, the second prediction model or the third prediction model, and a is the weight.

[0067] In order to prevent the model from overfitting, you can also use the pre-pruning method for the LightGBM model to limit the tree model layer height and the number of leaf nodes. LightGBM can also use L2 regularization to avoid overfitting. The TPE algorithm in the Optuna framework is used to perform automated hyperparameter search to determine the optimal hyperparameter. It should be noted that the Optuna framework and TPE algorithm are one of the implementation methods, and other algorithms can also be selected according to actual conditions, which is not limited here. The formula for TPE is as follows: , where EI is the expected improvement of optimization, and the iterative process of TPE is to find the hyperparameter configuration that maximizes EI.

[0068] The above embodiments exemplarily introduce the construction method and construction process of the prediction model. The following will introduce how to pre-regulate the stent adjustment strategy and perform stent tracking adjustment based on pre-regulation.

[0069] By inputting the predicted environmental data into the pre-trained prediction model, the initial bracket adjustment strategy can be obtained. The initial bracket adjustment strategy is used to indicate: the optimal bracket angle of the photovoltaic component in each time unit to maximize the solar radiation received by the photovoltaic component in each time unit. The initial bracket adjustment strategy can be stored in the database for the photovoltaic power station to call. In practical applications, the power generation of a photovoltaic power station will be affected by many factors, such as: solar radiation, efficiency of photovoltaic components, performance ratio, etc. When photovoltaic components are exposed to the outside for a long time, the dust and pollution on the photovoltaic components will reduce light penetration and reduce the solar radiation received by the photovoltaic components. In addition, severe weather can also cause damage to photovoltaic components, resulting in a decrease in power generation, photoelectric conversion efficiency and performance ratio. It can be seen that there is still room for further optimization of the initial bracket adjustment strategy. The optimization method will be exemplified by multiple embodiments below.

[0070] Step S102 , based on the predicted environmental data and the initial support adjustment strategy, the pre-constructed objective function is optimized to obtain a target support adjustment strategy.

[0071] Among them, the objective function is used to calculate the power generation of the photovoltaic power station within the preset time; the target bracket adjustment strategy is used to indicate: dividing the preset time into target time units and non-target time units, executing the cleaning strategy in the target time units, and executing the initial bracket adjustment strategy in the non-target time units.

[0072] In order to maximize the power generation of the photovoltaic power station within the preset time, the different environmental conditions that the photovoltaic modules may face can be considered, and the initial bracket adjustment strategy can be selected. For example, within the rainfall time unit, it is also possible not to maintain the optimal bracket angle, but to adjust the photovoltaic bracket to an angle suitable for cleaning, and reduce the dust accumulation on the photovoltaic modules through rainwater washing. Although the photovoltaic modules are not at the optimal bracket angle when the bracket is cleaned, resulting in a decrease in power generation efficiency at that time. However, the cleaned photovoltaic modules can receive more solar radiation and improve the power generation efficiency at other times. In order to maximize the long-term power generation efficiency (power generation), this embodiment proposes to use mixed integer linear programming to further optimize the initial bracket adjustment strategy.

[0073] Exemplarily, the decision variable may be set to determine whether to execute the cleaning strategy, and the value of the decision variable may be 0 or 1. 1 indicates that the cleaning strategy is executed, and 0 indicates that the cleaning strategy is not executed.

[0074] The objective function is to maximize the power generation of the photovoltaic power station within the preset time, which can be expressed as follows: ; Wherein, T is the preset time; Pt is the actual power generation of the t-th time unit; Dt is the dust accumulation ratio of the t-th time unit; St is the bracket angle of the t-th time unit (the value range is [Smin, Smax]); At is the cleaning decision variable of the t-th time unit (1 means cleaning, 0 means no cleaning); Wt is the rainfall forecast of the t-th time unit (1 means rain, 0 means no rain).

[0075] The value of St can be determined according to the initial support adjustment strategy, and the value of Wt can be determined according to the predicted meteorological data in the predicted environmental data.

[0076] The actual power generation is affected by dust accumulation and cleaning conditions. The calculation formula can be expressed as follows: ; Among them, Pdesign is the design power of the power station, and k is the reduction ratio of power generation under full power operation (same sunshine conditions) before cleaning.

[0077] The update of dust accumulation ratio can be expressed by the following formula: ; in, It is the efficiency of photovoltaic modules, which can be determined according to the models and specifications of photovoltaic modules.

[0078] In this embodiment, the CPLEX solver can be called to obtain the result, which is in the form of whether to perform cleaning. By deciding whether to perform cleaning in each time unit, the target bracket adjustment strategy can be obtained: the preset time is divided into target time units and non-target time units, cleaning is performed in the target time units, and the optimal bracket angle in the initial bracket adjustment strategy is executed in the non-target time units. While maximizing power generation within the preset time, this embodiment can avoid the invasion of severe weather and extend the service life of the bracket. The cleaning records can also be saved to the database to provide data support for the next optimization decision on whether to clean.

[0079] In some embodiments, the decision variable can also be set to whether to adjust the bracket to a wind-resistant angle when facing strong winds. Accordingly, the target bracket adjustment strategy can indicate which time units in the preset time need to adjust the bracket to a wind-resistant angle, and which time units execute the bracket angle provided by the initial bracket adjustment strategy. By implementing the wind-resistant strategy, the service life of the bracket can be extended and it can be used in complex and harsh environments. It should be noted that wind resistance and cleaning are only examples, and the bracket adjustment strategy can also be optimized according to actual needs, such as dustproof, snow-resistant, etc., which are not limited here.

[0080] In an optional embodiment, if Figure 5 As shown, executing a cleaning strategy may include: Step S500: Based on the predicted environmental data, the predicted meteorological data corresponding to the target time unit is acquired to determine the predicted wind data and the predicted precipitation data corresponding to the target time unit.

[0081] Step S502: determining the wind direction and precipitation amount corresponding to the target time unit based on the predicted wind force data and predicted precipitation data corresponding to the target time unit.

[0082] Step S504: determining a cleaning angle of the photovoltaic components in the photovoltaic power station according to the wind direction and precipitation corresponding to the target time unit.

[0083] Step S506, controlling the support angle of the photovoltaic components in the photovoltaic power station to be adjusted to the cleaning angle, so as to clean the photovoltaic components in the photovoltaic power station.

[0084] In this embodiment, the wind direction and precipitation (reflecting precipitation intensity) within the target time unit are first determined to determine the cleaning angle based on the wind direction and precipitation. The photovoltaic module is controlled to adjust the bracket angle to the cleaning angle when it reaches the target time unit, so as to remove the dust on the surface of the photovoltaic module through wind and rain, improve the light penetration rate of the photovoltaic module, and optimize the performance of the photovoltaic module. In an optional embodiment, if Figure 6 As shown in the figure, the estimated power generation of the photovoltaic power station within the preset period can be obtained by the following operations: Step S600: acquiring the shielding factor of the photovoltaic power station in each of the time units according to the predicted environmental data.

[0085] Step S602: acquiring the solar radiation of the photovoltaic power station in each of the time units according to the predicted environmental data and the target support adjustment strategy.

[0086] Step S604: determining the power generation of the photovoltaic power station in each of the time units according to the shielding factor and the solar radiation of the photovoltaic power station in each of the time units.

[0087] Step S606: determining the power generation of the photovoltaic power station within the preset time according to the power generation of the photovoltaic power station within each of the time units.

[0088] In practical applications, shadow shading will affect the power output of photovoltaic modules, resulting in a reduction in power generation. In order to accurately estimate the power generation, a machine learning model of the shading factor can be established. The shading factor is the reduction ratio of the power generation under specific environmental conditions. The shading factor can be calculated based on the total power designed for the photovoltaic power station and the current power generation. For example, machine learning modeling can be performed based on historical cloud cover data to establish a decision tree regression model. The training set can use the cloud cover and terrain of other existing power stations, and the output is the shading factor. Put the shading factor into the power generation formula for calculation to obtain the power generation of the photovoltaic power station per time unit. The power generation formula can be expressed as follows: E_h=A×Geff×PR×η; Geff = Gh × (1 − SL); Among them, A is the total installation area of ​​PV modules; η is the efficiency of PV modules; PR is the performance ratio of PV modules; Gh is the solar radiation per time unit; Geff is the effective solar radiation per time unit; SL is the shielding factor, which represents the radiation loss ratio caused by shadows. The cloud cover data of the National Satellite Meteorological Center is used to predict the shielding factor of each future time unit, and the power generation formula is used to calculate the power generation of each future time unit, so as to accumulate and calculate the estimated power generation of the PV power station at the preset time.

[0089] In an optional embodiment, the pre-planned photovoltaic bracket tracking method may also include: obtaining real-time environmental data of the current time unit; generating a bracket tracking strategy based on the real-time environmental data of the current time unit and the target bracket adjustment strategy; wherein the bracket tracking strategy is used to adjust the bracket angle of the photovoltaic components in the photovoltaic power station in the current time unit.

[0090] As mentioned above, the target bracket adjustment strategy is a bracket adjustment strategy that is calculated, predicted, and planned in advance based on the predicted environmental data. In practical applications, every time a new time unit is reached, the bracket adjustment can be directly performed according to the target bracket adjustment strategy to achieve efficient response, reduce the impact on power generation, and improve the reliability and automation of the power generation system. In fact, the predicted environmental data is based on the deduction of historical environmental data, which may differ from the real-time environmental data, so the implementation effect of the target bracket adjustment strategy may not be as expected. In order to further optimize the bracket adjustment strategy and improve the adaptability and detection accuracy in different time, climate, astronomical, and terrain scenarios, the real-time environmental data of the current time unit can also be collected in real time. According to the difference between the real-time environmental data and the predicted environmental data of the current time unit, the target bracket adjustment strategy is fine-tuned to generate a bracket tracking strategy. The bracket tracking strategy can include the optimal bracket angle for the current time unit.

[0091] In this embodiment, based on the real-time environmental data of the current time unit and the target bracket adjustment strategy, the bracket tracking strategy of the current time unit is determined. The bracket angle can be adjusted more intelligently according to the real-time environment, thereby improving the flexibility and accuracy of the bracket adjustment. The automation level and operation efficiency of the photovoltaic power station can also be improved, thereby further improving the power generation of the photovoltaic power station.

[0092] After determining the target support adjustment strategy or support tracking strategy, support angle adjustment can be performed. The following is an exemplary introduction to the process of controlling and adjusting the support angle.

[0093] Exemplarily, the target bracket adjustment strategy (for initial deployment) or the bracket tracking strategy (for real-time adjustment) can be obtained from the preset database to obtain a JSON string carrying the bracket angle. The pulse information is transmitted to the motor control chip through the MQTT protocol, and the pulse information includes the bracket angle to be executed. The motor control chip controls the angular displacement by outputting a corresponding number of PWM pulses, thereby driving the motor (bracket shaft) to achieve bracket adjustment. In practical applications, mechanical transmission is prone to generate and accumulate errors, resulting in reduced bracket adjustment accuracy. There are many ways to reduce errors, such as replacing high-precision transmission components, maintenance, replacement of worn parts, error compensation, etc. An exemplary scheme is provided below.

[0094] In an optional embodiment, the pre-regularized photovoltaic bracket tracking method may also include: correcting the bracket tracking strategy through a pre-trained transmission error model to obtain a target bracket angle; controlling the bracket angle of the photovoltaic components in the photovoltaic power station to adjust to the target bracket angle.

[0095] For example, a transmission error model can be established by training through a machine learning algorithm based on a large amount of historical data and real-time data. The transmission error model can be used to predict and compensate for errors in mechanical transmission and calculate the motor transmission ratio after incorporating the mechanical transmission error. The target bracket angle can be obtained by correcting the bracket angle provided by the bracket tracking strategy or the target bracket adjustment strategy through the transmission error model. The bracket angle of the photovoltaic module is controlled to be adjusted to the target bracket angle.

[0096] In this embodiment, the bracket angle is corrected taking into account the mechanical transmission error. The corrected target bracket angle can minimize the transmission deviation caused by factors such as mechanical wear, assembly errors and temperature changes, thereby ensuring that the photovoltaic modules always remain at the optimal bracket angle, thereby improving the power generation efficiency and system stability of the photovoltaic power station.

[0097] In order to make this application easier to understand, the following Figure 7 An exemplary application is provided.

[0098] S1: Establish a big data platform to store collected data into the lake, extract, integrate, cleanse, and calculate indicators to support upper-level machine learning modeling.

[0099] S2: Terrain data collection: Use drones to survey and collect terrain, elevation, occlusion, and slope information.

[0100] S3: Meteorological data collection: Deploy meteorological sensor arrays and sky cameras to collect cloud cover and real-time weather information.

[0101] S4: Astronomical data collection, collecting astronomical information from the National Comprehensive Earth Observation Data Sharing Platform, including solar angle, power station longitude and latitude, and rotation angle.

[0102] S5: Three machine learning models are established based on the three types of data on the big data platform, and the three models are integrated. The integrated models are used to predict the bracket angle planning, and the prediction results of the models (initial bracket adjustment strategy) are stored in the preset database.

[0103] S6: Pre-planning: Based on the predicted environmental data and the initial bracket adjustment strategy, the objective function is optimized to maximize the future long-term power generation of the photovoltaic power station, and the target bracket adjustment strategy is obtained and stored in the preset database.

[0104] S7: Real-time dynamic adjustment: Based on real-time environmental data and target support adjustment strategy, the support tracking strategy is calculated and determined in real time.

[0105] S8: Benefit calculation: Based on the target bracket adjustment strategy, comprehensively calculate the power generation of the photovoltaic power station in one year or throughout its entire life cycle.

[0106] S9: Mechanical transmission error correction: Establish a transmission error model, calculate according to the output angle value of the bracket tracking strategy or the target bracket adjustment strategy, and output the corrected angle adjustment value, that is, the target bracket angle, in real time.

[0107] S10: Transmit the pulse information to be executed to the motor control chip through the MQTT protocol.

[0108] S12: The motor control chip outputs the number of PWM pulses to control the angular displacement to drive the motor to adjust the angle of the bracket.

[0109] In this exemplary application: (1) the future long-term bracket adjustment strategy can be pre-planned before the bracket is deployed, and the bracket adjustment strategy can be selected based on the predicted environmental data to obtain the global optimal target bracket adjustment strategy. This can maximize the power generation of the photovoltaic power station within the preset time, reduce the impact of the initial bracket adjustment stage on the power generation, and adapt to various complex and harsh environments; (2) the bracket angle can be adjusted more intelligently according to the real-time environment, improving the flexibility and accuracy of the bracket adjustment, and also improving the automation level and operation efficiency of the photovoltaic power station, thereby further increasing the power generation of the photovoltaic power station.

[0110] Embodiment 2 Figure 8 The block diagram of the photovoltaic support tracking device based on pre-regulation according to the second embodiment of the present application is schematically shown. The device can be divided into one or more program modules, one or more program modules are stored in a storage medium, and are executed by one or more processors to complete the embodiment of the present application. The program module referred to in the embodiment of the present application refers to a series of computer program instruction segments that can complete specific functions. The following description will specifically introduce the functions of each program module in this embodiment. Figure 8 As shown, the apparatus 1000 may include: a first pre-regularization module 1100 and a second pre-regularization module 1200, wherein: The first pre-regularization module 1100 is used to obtain predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes a plurality of time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit; A second pre-regularization module 1200 is used to optimize a pre-constructed objective function based on the predicted environmental data and the initial bracket adjustment strategy to obtain a target bracket adjustment strategy, wherein the target bracket adjustment strategy is used to indicate: dividing the preset time into a target time unit and a non-target time unit, executing a cleaning strategy in the target time unit, and executing the initial bracket adjustment strategy in the non-target time unit; The objective function is used to calculate the power generation of the photovoltaic power station within the preset time.

[0111] As an optional embodiment, the predicted environmental data includes predicted meteorological data, and the predicted meteorological data includes predicted wind data and predicted precipitation data; Execute cleaning strategies, including: Based on the predicted environmental data, obtaining predicted meteorological data corresponding to the target time unit to determine predicted wind data and predicted precipitation data corresponding to the target time unit; Determining the wind direction and precipitation corresponding to the target time unit based on the predicted wind force data and the predicted precipitation data corresponding to the target time unit; determining a cleaning angle of photovoltaic components in the photovoltaic power station according to the wind direction and precipitation corresponding to the target time unit; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the cleaning angle to clean the photovoltaic components in the photovoltaic power station.

[0112] As an optional embodiment, the power generation of the photovoltaic power station within the preset time is calculated by the following operations: According to the predicted environmental data, obtaining a shielding factor of the photovoltaic power station in each of the time units; According to the predicted environmental data and the target bracket adjustment strategy, obtaining the solar radiation amount of the photovoltaic power station in each of the time units; Determining the power generation of the photovoltaic power station in each of the time units according to the shielding factor and the solar radiation of the photovoltaic power station in each of the time units; The power generation of the photovoltaic power station within the preset time is determined according to the power generation of the photovoltaic power station within each of the time units.

[0113] As an optional embodiment, the device 1000 is further used for: Get the real-time environment data of the current time unit; Generate a support tracking strategy based on the real-time environmental data of the current time unit and the target support adjustment strategy; The support tracking strategy is used to adjust the support angle of the photovoltaic components in the photovoltaic power station in the current time unit.

[0114] As an optional embodiment, the device 1000 is further used for: The support tracking strategy is corrected by using a pre-trained transmission error model to obtain a target support angle; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the target support angle.

[0115] As an optional embodiment, the prediction model is obtained by the following operations: Acquiring historical environmental data, wherein the historical environmental data includes historical meteorological data, historical astronomical data and historical topographic data; Based on the historical meteorological data, the historical astronomical data and the historical terrain data, constructing a first prediction model, a second prediction model and a third prediction model in a one-to-one correspondence; The first prediction model, the second prediction model and the third prediction model are integrated to obtain the prediction model.

[0116] As an optional embodiment, based on the historical meteorological data, the astronomical meteorological data and the historical terrain data, a first prediction model, a second prediction model and a third prediction model are constructed in a one-to-one correspondence, including: Acquire multiple pieces of the historical meteorological data, and merge the multiple pieces of the historical meteorological data into a wide table, wherein each column of the wide table represents a feature, each row of the wide table represents a sample, the sample includes feature values ​​corresponding to multiple features, and each sample corresponds to one piece of the historical meteorological data; Preprocessing the wide table, the preprocessing comprising: deleting outlier eigenvalues, filling missing eigenvalues, synthesizing samples and adding the synthesized samples to the wide table; Encode the feature values ​​of the preprocessed wide table and perform feature screening; Based on the screened features and corresponding feature values, the first prediction model is constructed.

[0117] As an optional embodiment, the wide table further includes a label column, and the label column has multiple label values; The synthesizing sample and adding the synthesized sample to the wide table comprises: Obtaining the number of samples corresponding to different label values, determining a first type of label value and a second type of label value, wherein the number of samples corresponding to the first type of label value is less than the number of samples corresponding to the second type of label value; When the number of samples corresponding to the first category label value is less than the number of samples corresponding to the second category label value: determine a first sample from the samples corresponding to the first category label value, and determine a plurality of candidate samples whose similarity to the first sample is within a preset range; determine a second sample from the plurality of candidate samples; and generate a synthetic sample corresponding to the first category label value based on the second sample and the first sample.

[0118] Embodiment 3 Fig. 9The schematic diagram of the hardware architecture of the computer device 10000 suitable for implementing the photovoltaic support tracking method based on pre-regulation according to the third embodiment of the present application is schematically shown. In some embodiments, the computer device 10000 can be a terminal device such as a smart phone, a wearable device, a tablet computer, a personal computer, a vehicle terminal, a game console, a virtual device, a workbench, a digital assistant, a set-top box, a robot, etc. In other embodiments, the computer device 10000 can be a rack server, a blade server, a tower server or a cabinet server (including an independent server, or a server cluster composed of multiple servers), etc. Fig. 9 As shown, the computer device 10000 includes but is not limited to: a memory 10010, a processor 10020, and a network interface 10030 that can communicate with each other through a system bus. Among them: The memory 10010 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 10010 can be an internal storage module of the computer device 10000, such as a hard disk or memory of the computer device 10000. In other embodiments, the memory 10010 can also be an external storage device of the computer device 10000, such as a plug-in hard disk equipped on the computer device 10000, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 10010 can also include both the internal storage module of the computer device 10000 and its external storage device. In this embodiment, the memory 10010 is generally used to store the operating system and various application software installed in the computer device 10000, such as program codes based on the pre-defined photovoltaic support tracking method, etc. In addition, the memory 10010 can also be used to temporarily store various data that have been output or will be output.

[0119] In some embodiments, the processor 10020 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other chips. The processor 10020 is generally used to control the overall operation of the computer device 10000, such as performing control and processing related to data interaction or communication with the computer device 10000. In this embodiment, the processor 10020 is used to run the program code stored in the memory 10010 or process data.

[0120] The network interface 10030 may include a wireless network interface or a wired network interface, and the network interface 10030 is generally used to establish a communication link between the computer device 10000 and other computer devices. For example, the network interface 10030 is used to connect the computer device 10000 to an external terminal through a network, and to establish a data transmission channel and a communication link between the computer device 10000 and the external terminal. The network may be a wireless or wired network such as an intranet, the Internet, the Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network, Bluetooth, Wi-Fi, etc.

[0121] It should be pointed out that Fig. 9 Only a computer device having components 10010 - 10030 is shown, but it should be understood that implementation of all of the components shown is not a requirement, and more or fewer components may alternatively be implemented.

[0122] In this embodiment, the pre-regularized photovoltaic support tracking method stored in the memory 10010 can also be divided into one or more program modules and executed by one or more processors (such as processor 10020) to complete the embodiment of the present application.

[0123] Embodiment 4 An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the photovoltaic support tracking method based on pre-regulation in the embodiment are implemented.

[0124] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Of course, the computer-readable storage medium may also include both an internal storage unit of a computer device and an external storage device thereof. In this embodiment, the computer-readable storage medium is generally used to store an operating system and various application software installed on a computer device, such as a program code based on a pre-regulated photovoltaic bracket tracking method in an embodiment, etc. In addition, the computer-readable storage medium may also be used to temporarily store various types of data that have been output or are to be output.

[0125] Embodiment 5 An embodiment of the present application also provides a computer program product, including a computer program, which implements the method in the above embodiment when executed by a processor.

[0126] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present application can be implemented by general-purpose computer devices, they can be concentrated on a single computer device, or distributed on a network composed of multiple computer devices, optionally, they can be implemented by executable program codes of computer devices, so that they can be stored in a storage device and executed by the computer device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0127] It should be noted that the above are only preferred embodiments of the present application, and the patent protection scope of the present application is not limited thereto. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A photovoltaic support tracking method based on pre-regulation, characterized in that: The method comprises: Acquire predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes a plurality of time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit; Based on the predicted environmental data and the initial support adjustment strategy, optimizing the pre-constructed objective function to obtain a target support adjustment strategy; Among them, the objective function is used to calculate the power generation of the photovoltaic power station within the preset time; the target bracket adjustment strategy is used to indicate: dividing the preset time into target time units and non-target time units, executing the cleaning strategy in the target time units, and executing the initial bracket adjustment strategy in the non-target time units.

2. The method according to claim 1, characterized in that The predicted environmental data includes predicted meteorological data, and the predicted meteorological data includes predicted wind data and predicted precipitation data; The execution of the cleaning strategy includes: Based on the predicted environmental data, obtaining predicted meteorological data corresponding to the target time unit to determine predicted wind data and predicted precipitation data corresponding to the target time unit; Determining the wind direction and precipitation corresponding to the target time unit based on the predicted wind force data and the predicted precipitation data corresponding to the target time unit; determining a cleaning angle of photovoltaic components in the photovoltaic power station according to the wind direction and precipitation corresponding to the target time unit; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the cleaning angle to clean the photovoltaic components in the photovoltaic power station.

3. The method according to claim 1, characterized in that The power generation of the photovoltaic power station within the preset time is calculated by the following operations: According to the predicted environmental data, obtaining a shielding factor of the photovoltaic power station in each of the time units; According to the predicted environmental data and the target bracket adjustment strategy, obtaining the solar radiation amount of the photovoltaic power station in each of the time units; Determining the power generation of the photovoltaic power station in each of the time units according to the shielding factor and the solar radiation of the photovoltaic power station in each of the time units; The power generation of the photovoltaic power station within the preset time is determined according to the power generation of the photovoltaic power station within each of the time units.

4. The method according to claim 1, characterized in that Also includes: Get the real-time environment data of the current time unit; Generate a support tracking strategy based on the real-time environmental data of the current time unit and the target support adjustment strategy; The support tracking strategy is used to adjust the support angle of the photovoltaic components in the photovoltaic power station in the current time unit.

5. The method according to claim 4, characterized in that Also includes: The support tracking strategy is corrected by using a pre-trained transmission error model to obtain a target support angle; The support angle of the photovoltaic components in the photovoltaic power station is controlled to be adjusted to the target support angle.

6. The method according to claim 1, characterized in that The prediction model is obtained by the following operations: Acquiring historical environmental data, wherein the historical environmental data includes historical meteorological data, historical astronomical data and historical topographic data; Based on the historical meteorological data, the historical astronomical data and the historical terrain data, constructing a first prediction model, a second prediction model and a third prediction model in a one-to-one correspondence; The first prediction model, the second prediction model and the third prediction model are integrated to obtain the prediction model.

7. The method according to claim 6, characterized in that Based on the historical meteorological data, the astronomical meteorological data and the historical terrain data, a first prediction model, a second prediction model and a third prediction model are constructed in a one-to-one correspondence, including: Acquire multiple pieces of the historical meteorological data, and merge the multiple pieces of the historical meteorological data into a wide table, wherein each column of the wide table represents a feature, each row of the wide table represents a sample, the sample includes feature values ​​corresponding to multiple features, and each sample corresponds to one piece of the historical meteorological data; Preprocessing the wide table, the preprocessing comprising: deleting outlier eigenvalues, filling missing eigenvalues, synthesizing samples and adding the synthesized samples to the wide table; Encode the feature values ​​of the preprocessed wide table and perform feature screening; Based on the screened features and corresponding feature values, the first prediction model is constructed.

8. The method according to claim 7, characterized in that The wide table also includes a label column, and the label column has multiple label values; The synthesizing sample and adding the synthesized sample to the wide table comprises: Obtaining the number of samples corresponding to different label values, determining a first type of label value and a second type of label value, wherein the number of samples corresponding to the first type of label value is less than the number of samples corresponding to the second type of label value; When the number of samples corresponding to the first category label value is less than the number of samples corresponding to the second category label value: determine a first sample from the samples corresponding to the first category label value, and determine a plurality of candidate samples whose similarity to the first sample is within a preset range; determine a second sample from the plurality of candidate samples; and generate a synthetic sample corresponding to the first category label value based on the second sample and the first sample.

9. A photovoltaic support tracking device based on pre-regulation, characterized in that: The device comprises: A first pre-regularization module is used to obtain predicted environmental data of the photovoltaic power station within a preset time, input the predicted environmental data into a pre-trained prediction model, and obtain an initial bracket adjustment strategy; wherein the preset time includes a plurality of time units, and the initial bracket adjustment strategy is used to indicate the bracket angle of the photovoltaic components in the photovoltaic power station within each time unit; A second pre-regularization module is used to optimize the pre-constructed objective function based on the predicted environmental data and the initial support adjustment strategy to obtain a target support adjustment strategy; Among them, the objective function is used to calculate the power generation of the photovoltaic power station within the preset time; the target bracket adjustment strategy is used to indicate: dividing the preset time into target time units and non-target time units, executing the cleaning strategy in the target time units, and executing the initial bracket adjustment strategy in the non-target time units.

10. A computer device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein: The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images