Photovoltaic module cleaning scheduling method, device, equipment and medium

Abstract

The embodiment of the invention provides a photovoltaic module cleaning scheduling method and device, equipment and a medium. The method comprises the following steps: acquiring environment data corresponding to a photovoltaic module to be cleaned and state data of the photovoltaic module, inputting the environment data and the state data into a pre-trained dust accumulation prediction model, and outputting a predicted dust thickness of a future set time length; and inputting the predicted dust thickness into a cleaning cost prediction algorithm, outputting a corresponding cleaning cost index, and when the cleaning cost index and the predicted dust thickness meet set conditions, generating a cleaning scheduling instruction corresponding to the photovoltaic module. The method is used for solving the problem that the cleaning scheduling effect of the photovoltaic module in a fixed period is poor in the related technology.

Description

Technical Field

[0001] This application relates to the field of photovoltaic equipment technology, and in particular to a method, apparatus, equipment and medium for cleaning and scheduling photovoltaic modules. Background Technology

[0002] With the widespread application of photovoltaic (PV) power generation technology, the cleanliness of PV modules directly affects their power generation efficiency. Dust accumulation on the module surface leads to a decrease in the light absorption rate of the PV cells, thus reducing power generation. Therefore, regular cleaning of PV modules is essential for maintaining their efficient operation. However, PV power plants are typically located in remote areas or regions with complex environmental conditions, making manual cleaning not only time-consuming and labor-intensive but also costly. Therefore, automating the cleaning of PV modules using automated equipment has become a major research direction in this field.

[0003] In related technologies, cleaning scheduling algorithms are usually scheduled according to a fixed period. This fails to take into account the dynamic changes in dust accumulation, which may lead to excessive cleaning and increased operating costs, or insufficient cleaning and reduced power generation efficiency. As a result, the effect of automated cleaning scheduling is poor, which affects the economic benefits of photovoltaic power generation. Summary of the Invention

[0004] This application provides a photovoltaic module cleaning scheduling method, apparatus, equipment, and medium to solve the problem of poor cleaning scheduling effect of fixed-cycle photovoltaic modules in related technologies.

[0005] In a first aspect, embodiments of this application provide a photovoltaic module cleaning scheduling method, including:

[0006] Obtain environmental data and status data of the photovoltaic modules to be cleaned. The status data is used to indicate the operating status of the photovoltaic modules.

[0007] Environmental and state data are input into a pre-trained dust accumulation prediction model, which outputs the predicted dust thickness over a set time period in the future.

[0008] The predicted dust thickness is input into the cleaning cost prediction algorithm, which outputs the corresponding cleaning cost index. The cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness. The cleaning cost index is an indicator used to represent the energy consumption cost.

[0009] When the cleaning cost index and the predicted dust thickness meet the set conditions, a cleaning scheduling instruction corresponding to the photovoltaic module is generated. The cleaning scheduling instruction is used to instruct the cleaning module corresponding to the photovoltaic module to clean the photovoltaic module after a set time.

[0010] In one possible implementation, environmental data includes: solar irradiance, precipitation, humidity, temperature, and wind speed in the area where the photovoltaic module is located; and status data includes the current decay rate, capacity, azimuth angle, tilt angle, and rollover angle of the photovoltaic module.

[0011] In one possible implementation, after obtaining the environmental data and state data of the photovoltaic module to be cleaned, the method further includes: filling in missing values ​​in the environmental data and state data; cleaning outliers in the environmental data and state data; and standardizing the environmental data and state data after filling and cleaning, respectively.

[0012] In one possible implementation, the cleaning cost prediction model is trained based on a long short-term memory network model. The cleaning cost prediction model includes an input layer, a hidden layer, and an output layer. The input layer includes dimensions corresponding to environmental data, dimensions corresponding to state data, and dimensions corresponding to historical data. The historical data includes historical dust thickness, historical state data, and / or historical environmental data. The hidden layer includes two bidirectional long short-term memory network layers. The output layer is used to output the predicted dust thickness for a set future time period. The cleaning cost prediction model is trained based on an adaptive moment estimation optimizer and a mean absolute error loss function.

[0013] In one possible implementation, the predicted dust thickness is input into the cleaning cost prediction model, and the corresponding cleaning cost index is output, including: determining the power generation loss caused by dust based on the predicted dust thickness; and calculating the cleaning cost index based on the power generation loss.

[0014] In one possible implementation, the formula for calculating power generation loss is expressed as:

[0015] ΔP=P max •(1 – e -k•d ),

[0016] Where ΔP represents the power generation loss, P max The maximum power generation of the photovoltaic module is represented by , k represents the degradation coefficient of the photovoltaic module material, and d is the predicted dust thickness; the formula for calculating the cleaning cost index is as follows:

[0017] I = (ΔP • C E ) / C W • (1 + 0.5 • R rain );

[0018] Where I represents the cleaning cost index, C E C represents the price cost per unit of electricity generated. W R represents the price cost per cleaning session. rain This indicates the probability of rainfall corresponding to a set future duration.

[0019] In one possible implementation, when the cleaning cost index and the predicted dust thickness meet the set conditions, a cleaning scheduling instruction corresponding to the photovoltaic module is generated, including: if the cleaning cost index is greater than a first set value and the predicted dust thickness is greater than a second set value, a cleaning scheduling instruction is generated; if the probability of rainfall within a future set time period is greater than a third set value and the cleaning cost index is less than a fourth set value, the generated cleaning scheduling instruction corresponding to the future set time period is delayed or canceled.

[0020] Secondly, embodiments of this application provide a photovoltaic module cleaning and scheduling device, comprising:

[0021] The acquisition module is used to acquire environmental data and status data of the photovoltaic modules to be cleaned. The status data is used to indicate the operating status of the photovoltaic modules.

[0022] The prediction module is used to input environmental and state data into a pre-trained dust accumulation prediction model and output the predicted dust thickness over a set time period in the future.

[0023] The calculation module is used to input the predicted dust thickness into the cleaning cost prediction algorithm and output the corresponding cleaning cost index. The cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness. The cleaning cost index is an indicator used to represent the energy consumption cost.

[0024] The determination module is used to generate a cleaning scheduling instruction for the photovoltaic module when the cleaning cost index and the predicted dust thickness meet the set conditions. The cleaning scheduling instruction is used to instruct the cleaning module corresponding to the photovoltaic module to clean the photovoltaic module after a set time.

[0025] In one possible implementation, the acquisition module specifically includes environmental data such as solar irradiance, precipitation, humidity, temperature, and wind speed in the area where the photovoltaic module is located; and status data such as the current decay rate, capacity, azimuth angle, tilt angle, and rollover angle of the photovoltaic module.

[0026] In one possible implementation, the acquisition module is further configured to: after acquiring the environmental data and state data of the photovoltaic module to be cleaned, fill in the missing values ​​in the environmental data and state data; clean the outliers in the environmental data and state data; and standardize the environmental data and state data after the filling and cleaning processes, respectively.

[0027] In one possible implementation, the prediction module specifically includes a cleaning cost prediction model trained on a long short-term memory network model. The cleaning cost prediction model includes an input layer, a hidden layer, and an output layer. The input layer includes dimensions corresponding to environmental data, dimensions corresponding to state data, and dimensions corresponding to historical data. The historical data includes historical dust thickness, historical state data, and / or historical environmental data. The hidden layer includes two bidirectional long short-term memory network layers. The output layer is used to output the predicted dust thickness for a set future time period. The cleaning cost prediction model is trained based on an adaptive moment estimation optimizer and a mean absolute error loss function.

[0028] In one possible implementation, the calculation module is specifically used to determine the power generation loss caused by dust based on the predicted dust thickness; and to calculate the cleaning cost index based on the power generation loss.

[0029] In one possible implementation, the calculation module specifically includes a formula for calculating power generation loss, expressed as:

[0030] ΔP=P max •(1 – e -k•d ),

[0031] Where ΔP represents the power generation loss, P max The maximum power generation of the photovoltaic module is represented by , k represents the degradation coefficient of the photovoltaic module material, and d is the predicted dust thickness; the formula for calculating the cleaning cost index is as follows:

[0032] I = (ΔP • C E ) / C W • (1 + 0.5 • R rain );

[0033] Where I represents the cleaning cost index, C E C represents the price cost per unit of electricity generated. W R represents the price cost per cleaning session. rain This indicates the probability of rainfall corresponding to a set future duration.

[0034] In one possible implementation, the determining module is specifically used to generate a cleaning scheduling instruction if the cleaning cost index is greater than a first set value and the predicted dust thickness is greater than a second set value; and to delay or cancel the cleaning scheduling instruction corresponding to the future set time if the probability of rainfall within the time range to the future set time is greater than a third set value and the cleaning cost index is less than a fourth set value.

[0035] Thirdly, embodiments of this application provide a control device, including: a memory and a processor;

[0036] The memory stores computer-executed instructions;

[0037] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0038] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0039] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0040] The photovoltaic module cleaning scheduling method, apparatus, equipment, and medium provided in this application, by acquiring environmental data and photovoltaic module status data, combined with a pre-trained dust accumulation prediction model and cleaning cost prediction algorithm, achieve the goal of dynamically determining the cleaning timing and optimizing cleaning costs. By inputting multi-dimensional data into the model, it can accurately predict future dust thickness and perform reasonable cleaning scheduling based on the cleaning cost index, avoiding the problems of overly frequent or insufficient cleaning in traditional methods. Therefore, it not only improves the power generation efficiency of photovoltaic modules but also significantly reduces cleaning costs, enhancing the overall economic benefits and operational efficiency of photovoltaic power plants. Attached Figure Description

[0041] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0042] Figure 1 This is an application scenario diagram of the photovoltaic module cleaning and scheduling method provided in the embodiments of this disclosure;

[0043] Figure 2 A flowchart of a photovoltaic module cleaning scheduling method provided in one embodiment of this disclosure;

[0044] Figure 3 A flowchart of a photovoltaic module cleaning scheduling method provided in yet another embodiment of this disclosure;

[0045] Figure 4 A schematic diagram of the structure of a photovoltaic module cleaning and scheduling device provided in yet another embodiment of this disclosure;

[0046] Figure 5 This is a schematic diagram of the structure of a control device provided in one embodiment of the present disclosure.

[0047] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0048] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0049] In photovoltaic (PV) power generation systems, the surface cleanliness of PV modules directly affects their photoelectric conversion efficiency. Dust accumulation obstructs the surface of PV cells, reducing light absorption and thus power generation. Therefore, regular cleaning of PV modules is crucial for maintaining their efficient operation. However, PV power plants are typically located in remote areas or regions with complex environmental conditions, making traditional manual cleaning methods not only time-consuming and labor-intensive but also costly. To address this, automated cleaning scheduling systems have emerged, using intelligent algorithms to optimize cleaning frequency and timing to reduce cleaning costs while ensuring power generation efficiency.

[0050] In related technologies, dust cleaning scheduling algorithms typically rely on fixed periods or simple linear regression models to predict dust accumulation. This approach fails to adequately consider the dynamic changes of multi-dimensional factors, making it difficult to adjust the cleaning frequency according to actual conditions. Fixed-period scheduling may lead to overly frequent cleaning, increasing operating costs, or insufficient cleaning, affecting power generation efficiency. Simple prediction models, lacking comprehensive analysis of complex environmental factors, have low prediction accuracy and struggle to accurately determine the optimal cleaning time. Therefore, existing dust cleaning scheduling algorithms suffer from poor performance.

[0051] The photovoltaic module cleaning scheduling method provided in this application acquires environmental and status data of the photovoltaic modules and inputs this data into a pre-trained dust accumulation prediction model to predict future dust thickness. Subsequently, the prediction results are input into a cleaning cost prediction algorithm to calculate a cleaning cost index. When the cleaning cost index and the predicted dust thickness meet set conditions, a cleaning scheduling command is generated, instructing the automatic cleaning equipment to perform cleaning at the optimal time. This effectively optimizes the cleaning frequency, reduces cleaning costs, and improves the power generation efficiency of the photovoltaic modules.

[0052] Figure 1 This is a schematic diagram illustrating an application scenario of the photovoltaic module cleaning and scheduling method provided in this application, such as... Figure 1As shown, the specific application scenario of this application is as follows: In the photovoltaic module cleaning scheduling, the server 100 obtains the environmental data of the photovoltaic power station 110 and the status data of the photovoltaic module 111. After determining the timing of cleaning, it sends an instruction to the cleaning component 112 to clean the photovoltaic module 111.

[0053] It should be noted that, Figure 1 The scenario shown includes a server, a photovoltaic power station, photovoltaic modules, and a cleaning component, which are only used as examples or in a specific number. However, this disclosure is not limited to this. That is to say, the number of servers, photovoltaic power stations, photovoltaic modules, and cleaning components can be arbitrary.

[0054] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0055] Figure 2 Flowchart of the photovoltaic module cleaning and scheduling method provided in this application Figure 1 ,like Figure 2 As shown, the method includes:

[0056] S201. Obtain the environmental data and status data of the photovoltaic module to be cleaned.

[0057] Among them, the status data is used to represent the operating status of photovoltaic modules.

[0058] Specifically, this embodiment is used to provide a general description of the main process of photovoltaic module cleaning and scheduling.

[0059] In this embodiment of the disclosure, the executing entity is a control system, controller, or server that can acquire relevant status data and environmental data of photovoltaic modules and control the cleaning of the modules. For ease of description, it will be referred to as a server below.

[0060] Environmental data typically includes weather-related factors such as precipitation, humidity, wind speed, and temperature. These factors directly affect the rate at which dust accumulates on the surface of photovoltaic modules and the necessity of cleaning.

[0061] Servers can monitor these environmental parameters in real time through weather stations or sensor networks installed near photovoltaic power plants.

[0062] Furthermore, with the development of IoT technology, more and more photovoltaic power plants are adopting smart sensors, which can provide more detailed and real-time environmental data, thereby improving the accuracy and timeliness of the data.

[0063] The server can periodically retrieve this data (such as every 6 hours, every hour, every 15 minutes, or any other time interval) and determine whether to perform cleaning scheduling.

[0064] On the other hand, the status data of photovoltaic modules is used to indicate their operating status, including the degree of aging of the photovoltaic modules (such as current decay rate) and their orientation (including azimuth angle, tilt angle, and rollover angle). These data reflect the actual working performance and installation conditions of the photovoltaic modules.

[0065] The server can obtain this status data through the control unit of the photovoltaic module.

[0066] S202. Input environmental data and state data into a pre-trained dust accumulation prediction model and output the predicted dust thickness for a set future time period.

[0067] Specifically, after obtaining environmental and status data, the next step is to input this data into the dust accumulation prediction model.

[0068] This model can learn the patterns of dust accumulation under different environmental conditions and component states by training on historical data, thereby predicting the dust thickness within a specific time period in the future (such as any time period such as 72 hours, 48 ​​hours, 24 hours, etc.).

[0069] By determining in advance when cleaning is needed, excessive dust accumulation can be effectively avoided, thus preventing it from affecting power generation efficiency.

[0070] In some embodiments, more input features, such as historical dust thickness, historical state data, and historical environmental data, can be introduced to enrich the model's input information.

[0071] The specific model can be selected from various machine learning algorithms, such as convolutional neural networks (CNN) or Transformer models, to ensure the model's feature extraction capabilities and prediction performance.

[0072] With the increase in data volume and the improvement of computing power, future dust accumulation prediction models will be able to provide more accurate and timely prediction results, providing strong support for cleaning scheduling.

[0073] S203. Input the predicted dust thickness into the cleaning cost prediction algorithm and output the corresponding cleaning cost index.

[0074] Among them, the cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness, and the cleaning cost index is an indicator used to represent the corresponding energy consumption cost.

[0075] Specifically, after obtaining the predicted dust thickness, the next step is to assess its potential impact on the power generation efficiency of photovoltaic modules and calculate the corresponding cleaning costs.

[0076] The cleaning cost prediction algorithm correlates predicted dust thickness with energy consumption costs, outputting a cleaning cost index. This index is used to quantify the impact of dust accumulation on photovoltaic module performance and the economics of cleaning operations.

[0077] Typically, the calculation of the cleaning cost index needs to consider multiple factors, including the loss of power generation due to dust thickness, the direct costs of cleaning operations (such as water consumption, power consumption, etc.), and potential weather effects (such as the natural cleaning effect that rainfall may bring).

[0078] To improve the accuracy of cleaning cost prediction, more complex economic models can be used. For example, a dynamic electricity price model can be introduced to reflect the impact of electricity price fluctuations on the cost of power generation losses.

[0079] In addition, weather forecast information can be used to assess the impact of future rainfall on cleaning needs, thereby optimizing the timing of cleaning.

[0080] S204. When the cleaning cost index and the predicted dust thickness meet the set conditions, generate the cleaning scheduling instruction corresponding to the photovoltaic module.

[0081] Among them, the cleaning scheduling instruction is used to instruct the cleaning component corresponding to the photovoltaic module to clean the photovoltaic module after a set time.

[0082] Specifically, when the cleaning cost index and the predicted dust thickness reach the set threshold conditions, the system will generate a cleaning scheduling instruction so that the automatic cleaning equipment can perform cleaning operations at the optimal time, thereby ensuring the efficient operation of photovoltaic modules.

[0083] The setting conditions typically include the cleaning cost index exceeding a certain threshold and the dust thickness reaching a level that affects power generation efficiency.

[0084] These conditions need to be adjusted based on actual operational experience and economic analysis to achieve the best balance between cleaning frequency and cost.

[0085] The photovoltaic module cleaning scheduling method provided in this application achieves the goal of dynamically determining the cleaning timing and optimizing cleaning costs by acquiring environmental data and photovoltaic module status data, combined with a pre-trained dust accumulation prediction model and cleaning cost prediction algorithm. By inputting multi-dimensional data into the model, it can accurately predict future dust thickness and perform reasonable cleaning scheduling based on the cleaning cost index, avoiding the problems of overly frequent or insufficient cleaning in traditional methods. Therefore, it not only improves the power generation efficiency of photovoltaic modules but also significantly reduces cleaning costs, enhancing the overall economic benefits and operational efficiency of photovoltaic power plants.

[0086] Figure 3 A schematic diagram of the photovoltaic module cleaning and scheduling process provided in this application Figure 2 ,like Figure 3 As shown, in this embodiment... Figure 2 Based on the embodiments, the specific implementation process of the photovoltaic module cleaning scheduling method is described in detail. The method includes:

[0087] S301, used to obtain environmental data and status data of the photovoltaic module to be cleaned.

[0088] Among them, the status data is used to represent the operating status of photovoltaic modules.

[0089] Specifically, in the cleaning and scheduling of photovoltaic modules, it is first necessary to obtain comprehensive environmental and status data.

[0090] In some embodiments, environmental data include: solar irradiance, precipitation, humidity, temperature, and wind speed in the area where the photovoltaic module is located.

[0091] These factors directly affect the rate at which dust accumulates on the component surface and the need for cleaning. Servers can periodically acquire and monitor these environmental parameters through weather stations or sensor networks installed near the photovoltaic power plant.

[0092] Status data includes the photovoltaic module's current decay rate, capacity, azimuth angle, tilt angle, and flip angle.

[0093] These data reflect the actual operating performance and installation conditions of photovoltaic modules. Among them, the current decay rate can help determine whether the module's power generation efficiency has decreased due to dust accumulation, while the module's physical parameters affect the dust deposition pattern and the difficulty of cleaning.

[0094] Servers can obtain this status data by installing dedicated monitoring equipment on photovoltaic modules or by collecting data through the module's control system.

[0095] S302. Fill in the missing values ​​in the environmental and status data.

[0096] Specifically, during data acquisition, data gaps may occur due to sensor malfunctions, network problems, or other unforeseen factors. The presence of missing values ​​can affect the accuracy of data analysis and the predictive power of the model. Therefore, it is necessary to fill in missing values ​​in environmental and state data.

[0097] Common imputation methods include mean imputation, median imputation, and interpolation. Those skilled in the art can select the appropriate imputation method based on the characteristics and missing data patterns.

[0098] In some embodiments, missing values ​​based on time series can be filled using time series prediction models, while missing data with spatial correlation can be filled using spatial statistical methods such as geographic weighted regression.

[0099] In some embodiments, to ensure the effectiveness of subsequent model analysis, machine learning algorithms such as K-nearest neighbor imputation (KNN) and multiple interpolation (MICE) can be selected to fill in missing values.

[0100] For example, taking the K-nearest neighbor imputation algorithm, we can choose to have K=5 neighbors and use Euclidean distance as the distance metric. We then use inverse distance weighting to interpolate consecutive missing values ​​(e.g., within 3 hours); for missing values ​​from more than one day, we mark them as invalid data.

[0101] S303. Perform outlier cleaning on environmental and status data.

[0102] Specifically, outliers refer to values ​​in the dataset that deviate from the normal range, which may be caused by sensor malfunctions, data entry errors, or other reasons.

[0103] Outliers in the environmental and condition data of photovoltaic modules can cause biases in the prediction model, so they need to be cleaned.

[0104] Common outlier detection methods include interquartile range (ICR) analysis, box plot analysis, Z-score analysis, and statistical distribution-based outlier detection. These methods can identify and process outliers in data, ensuring data quality.

[0105] In some embodiments, multiple methods can be combined for comprehensive analysis to improve the accuracy of outlier detection. For example, machine learning algorithms such as Isolation Forest and Support Vector Machine (SVM) can be used for anomaly detection to process high-dimensional data and capture complex anomaly patterns.

[0106] In addition, reasonable thresholds and rules can be set by combining domain knowledge to identify outliers in specific situations. For example, when wind speed shows an outlier (e.g., >60m / s), it can be replaced with the valid value from the previous 15 minutes; when PM2.5 shows an outlier (e.g., <0 or >1000μg / m³), it can be marked as a missing value, triggering the aforementioned missing value supplementation action.

[0107] S304. Standardize the environmental and status data after filling and cleaning processes, respectively.

[0108] Specifically, standardization is an important step in data preprocessing, which aims to eliminate dimensional differences between different features and enable data to be compared on the same scale.

[0109] For environmental and state data, standardization can improve the convergence speed and prediction accuracy of the model.

[0110] Commonly used standardization methods include min-maximum standardization and Z-score standardization.

[0111] These methods can transform data into a standard normal distribution with a mean of 0 and a variance of 1, thereby improving the stability of the model.

[0112] During the standardization process, an appropriate method can be selected based on the characteristics of the data. For example, for features with clear boundaries, min-max standardization can be used, while for features that follow a normal distribution, Z-score standardization can be used.

[0113] In addition, nonlinear transformation methods such as logarithmic transformation and Box-Cox transformation can be selected based on the distribution characteristics of the data and the requirements of the model to improve the linear separability of the data.

[0114] This provides higher-quality data input for subsequent model training, ensuring the accuracy and reliability of model predictions.

[0115] S305. Input environmental data and state data into a pre-trained dust accumulation prediction model and output the predicted dust thickness for a set future time.

[0116] Specifically, in this embodiment, the cleaning cost prediction model is trained based on a long short-term memory network model, which can process time series data and capture complex nonlinear relationships.

[0117] The cleaning cost prediction model includes an input layer, a hidden layer, and an output layer. The input layer includes dimensions corresponding to environmental data, dimensions corresponding to state data, and dimensions corresponding to historical data. Historical data includes historical dust thickness, historical state data, and / or historical environmental data. By combining historical data, the input information of the model is enriched, enabling the model to better identify the patterns of dust accumulation under different environmental conditions and component states, thereby predicting the dust thickness within a specific time period in the future.

[0118] The hidden layer consists of two bidirectional long short-term memory network layers, each with 128 nodes, to effectively capture the temporal dependencies of the data.

[0119] The output layer is used to output the predicted dust thickness over a set future time period, typically in g / m². 2 .

[0120] The cleaning cost prediction model is trained using an adaptive moment estimator optimizer (learning rate configurable to 0.001) and a mean absolute error loss function. The training period can be 50.

[0121] S306. Based on the predicted dust thickness, determine the power generation loss caused by dust.

[0122] Specifically, after obtaining the predicted dust thickness, the next step is to assess its potential impact on the power generation efficiency of photovoltaic modules.

[0123] The formula for calculating power generation loss is as follows:

[0124] ΔP=P max •(1 – e -k•d ),

[0125] Where ΔP represents the power generation loss, P max denoted by , k represents the maximum power generation of the photovoltaic module, k represents the degradation coefficient of the photovoltaic module material (e.g., the degradation coefficient of glass is ), and d represents the predicted dust thickness.

[0126] This allows us to quantify the impact of dust thickness on power generation, providing a basis for subsequent cleaning cost assessments.

[0127] In some embodiments, model correction can be performed by combining actual operating data to improve the accuracy of power generation loss calculation.

[0128] For example, the change in power generation under different dust thicknesses can be experimentally measured to determine a more accurate attenuation coefficient k.

[0129] In addition, the model parameters can be adjusted to suit different application scenarios by taking into account the characteristics of photovoltaic modules with different materials and installation conditions.

[0130] S307. Calculate the cleaning cost index based on power generation loss.

[0131] Specifically, the formula for calculating the cleaning cost index is as follows:

[0132] I = (ΔP • C E ) / C W • (1 + 0.5 • R rain );

[0133] Where I represents the cleaning cost index, C E C represents the price cost per unit of electricity generated. W R represents the price cost per cleaning session. rain This indicates the probability of rainfall corresponding to a set future duration.

[0134] By comprehensively considering power generation loss, cleaning costs, and weather factors, the economics of cleaning operations can be quantified, providing a basis for cleaning scheduling decisions.

[0135] Furthermore, to improve the accuracy of the cleaning cost index calculation, a dynamic electricity price model can be introduced to reflect the impact of electricity price fluctuations on power generation loss costs, such as C. E This refers to the real-time electricity price.

[0136] In addition, weather forecast information can be used to determine R. rain Assess the impact of future rainfall on cleaning needs to optimize cleaning timing.

[0137] S308. If the cleaning cost index is greater than the first set value and the predicted dust thickness is greater than the second set value, a cleaning scheduling instruction is generated.

[0138] Specifically, when the cleaning cost index and the predicted dust thickness reach the set threshold conditions, the server will generate a cleaning scheduling instruction.

[0139] The setting conditions typically include the cleaning cost index exceeding a certain threshold and the dust thickness reaching a level that affects power generation efficiency.

[0140] These conditions need to be adjusted based on actual operational experience and economic analysis to achieve the best balance between cleaning frequency and cost.

[0141] For example, a cleaning scheduling command can be generated when I ≥ 1.2 (i.e., the first set value) and the predicted dust thickness ≥ 0.8 g / m² (i.e., the second set value).

[0142] S309. If the probability of rainfall within the future set time period is greater than the third set value, and the cleaning cost index is less than the fourth set value, delay or cancel the cleaning scheduling instruction corresponding to the generated future set time period.

[0143] Specifically, during the cleaning and scheduling process, weather factors, especially the probability of rainfall, have a crucial impact on the cleaning plan.

[0144] Under certain conditions, the server can choose to delay or cancel generated cleaning schedule instructions. This avoids unnecessary cleaning operations, saves operating costs, and utilizes natural rainfall to clean components.

[0145] For example, R can be used in the next 72 hours or 24 hours. rain When the rate is ≥50%, delay low-priority tasks (i.e., pre-determined cleaning schedule instructions with I<1.5).

[0146] In some embodiments, to improve the accuracy of decision-making, more refined weather forecast data and dynamic adjustment mechanisms can be combined. For example, short-term weather forecast models can be used to provide more accurate rainfall probability predictions, and historical weather data can be combined for trend analysis.

[0147] The photovoltaic module cleaning scheduling method provided in this application improves data integrity and quality and enhances the accuracy of the dust accumulation prediction model by introducing preprocessing steps for environmental and status data, including missing value imputation, outlier cleaning, and standardization. Furthermore, by using a long short-term memory network model to predict dust thickness and combining this with the calculation of the cleaning cost index, the economics and necessity of cleaning can be assessed more accurately. By incorporating weather forecast information, the cleaning plan can be intelligently adjusted to avoid unnecessary cleaning operations, further reducing operating costs and improving the power generation efficiency of photovoltaic modules and the overall economic benefits of the photovoltaic power plant.

[0148] Figure 4 This is a schematic diagram of the photovoltaic module cleaning and scheduling device provided in this application, as shown below. Figure 4 As shown, the photovoltaic module cleaning and scheduling device 400 provided in this embodiment includes:

[0149] The acquisition module 410 is used to acquire environmental data and status data of the photovoltaic module to be cleaned. The status data is used to represent the operating status of the photovoltaic module.

[0150] The prediction module 420 is used to input environmental data and state data into a pre-trained dust accumulation prediction model and output the predicted dust thickness over a set time period in the future.

[0151] The calculation module 430 is used to input the predicted dust thickness into the cleaning cost prediction algorithm and output the corresponding cleaning cost index. The cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness. The cleaning cost index is an indicator used to represent the energy consumption cost.

[0152] The determination module 440 is used to generate a cleaning scheduling instruction corresponding to the photovoltaic module when the cleaning cost index and the predicted dust thickness meet the set conditions. The cleaning scheduling instruction is used to instruct the cleaning module corresponding to the photovoltaic module to clean the photovoltaic module after a set time.

[0153] In one possible implementation, the acquisition module 410 specifically includes environmental data such as solar irradiance, precipitation, humidity, temperature, and wind speed in the area where the photovoltaic module is located; and status data such as the current decay rate, capacity, azimuth angle, tilt angle, and rollover angle of the photovoltaic module.

[0154] In one possible implementation, the acquisition module 410 is further configured to: after acquiring the environmental data and the state data of the photovoltaic module to be cleaned, fill in the missing values ​​in the environmental data and the state data; clean the outliers in the environmental data and the state data; and standardize the environmental data and the state data after the filling process and the cleaning process, respectively.

[0155] In one possible implementation, the prediction module 420 specifically includes a cleaning cost prediction model trained on a long short-term memory network model. The cleaning cost prediction model includes an input layer, a hidden layer, and an output layer. The input layer includes dimensions corresponding to environmental data, dimensions corresponding to state data, and dimensions corresponding to historical data. The historical data includes historical dust thickness, historical state data, and / or historical environmental data. The hidden layer includes two bidirectional long short-term memory network layers. The output layer is used to output the predicted dust thickness for a set future time period. The cleaning cost prediction model is trained on an adaptive moment estimation optimizer and a mean absolute error loss function.

[0156] In one possible implementation, the calculation module 430 is specifically used to determine the power generation loss caused by dust based on the predicted dust thickness; and to calculate the cleaning cost index based on the power generation loss.

[0157] In one possible implementation, the calculation module 430 specifically includes a formula for calculating power generation loss, expressed as:

[0158] ΔP=P max •(1 – e -k•d ),

[0159] Where ΔP represents the power generation loss, P max The maximum power generation of the photovoltaic module is represented by , k represents the degradation coefficient of the photovoltaic module material, and d is the predicted dust thickness; the formula for calculating the cleaning cost index is as follows:

[0160] I = (ΔP • C E ) / C W • (1 + 0.5 • R rain );

[0161] Where I represents the cleaning cost index, C E C represents the price cost per unit of electricity generated. W R represents the price cost per cleaning session. rain This indicates the probability of rainfall corresponding to a set future duration.

[0162] In one possible implementation, the determining module 440 is specifically used to generate a cleaning scheduling instruction if the cleaning cost index is greater than a first set value and the predicted dust thickness is greater than a second set value; and to delay or cancel the cleaning scheduling instruction corresponding to the future set time if the probability of rainfall within the time range to the future set time is greater than a third set value and the cleaning cost index is less than a fourth set value.

[0163] The photovoltaic module cleaning and scheduling device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0164] Figure 5 A schematic diagram of the control device provided in this application. Figure 5 As shown, the control device 500 provided in this embodiment includes at least one processor 520 and a memory 510. Optionally, the device 50 also includes a communication component. The processor 520, memory 510, and communication component are connected via a bus 530.

[0165] In a specific implementation, at least one processor 520 executes computer execution instructions stored in memory 510, causing at least one processor 520 to perform the above-described method.

[0166] The specific implementation process of processor 520 can be found in the above method embodiments, and its implementation principle and technical effect are similar, so it will not be repeated here.

[0167] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0168] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0169] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0170] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0171] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0172] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0173] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0174] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0175] The units described as separate components may or may not be physically separate. 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0176] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0177] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0178] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0179] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A photovoltaic module cleaning scheduling method, characterized in that, include: Obtain environmental data and status data of the photovoltaic module to be cleaned, wherein the status data is used to indicate the operating status of the photovoltaic module; The environmental data and the state data are input into a pre-trained dust accumulation prediction model, which outputs the predicted dust thickness over a set time period in the future. The predicted dust thickness is input into the cleaning cost prediction algorithm, and the corresponding cleaning cost index is output. The cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness. The cleaning cost index is an indicator used to represent the energy consumption cost. When the cleaning cost index and the predicted dust thickness meet the set conditions, a cleaning scheduling instruction corresponding to the photovoltaic module is generated. The cleaning scheduling instruction is used to instruct the cleaning component corresponding to the photovoltaic module to clean the photovoltaic module after the set time period.

2. The method according to claim 1, characterized in that, The environmental data includes: solar irradiance, precipitation, humidity, temperature, and wind speed in the area where the photovoltaic module is located; The status data includes the current decay rate, capacity, azimuth angle, tilt angle, and rollover angle of the photovoltaic module.

3. The method according to claim 2, characterized in that, After obtaining the environmental data corresponding to the photovoltaic module to be cleaned and the status data of the photovoltaic module, the process further includes: Fill in the missing values ​​in the environmental and status data; Outlier cleaning is performed on the environmental and status data; The environmental and status data after filling and cleaning processes were standardized respectively.

4. The method according to claim 1, characterized in that, The cleaning cost prediction model is trained based on a long short-term memory network model, and the cleaning cost prediction model includes an input layer, a hidden layer, and an output layer. The input layer includes the environmental data corresponding dimension, the state data corresponding dimension, and the historical data corresponding dimension. The historical data includes historical dust thickness, historical state data, and / or historical environmental data. The hidden layer comprises two bidirectional long short-term memory network layers; The output layer is used to output the predicted dust thickness over a set future time period; The cleaning cost prediction model is trained based on an adaptive moment estimator optimizer and a mean absolute error loss function.

5. The method according to any one of claims 1 to 4, characterized in that, The step of inputting the predicted dust thickness into the cleaning cost prediction model and outputting the corresponding cleaning cost index includes: Based on the predicted dust thickness, the power generation loss caused by dust is determined; The cleaning cost index is calculated based on the power generation loss.

6. The method according to claim 5, characterized in that, The formula for calculating the power generation loss is as follows: ΔP=P max •(1 – e -k•d ), Where ΔP represents the power generation loss, P max denoted by , where k represents the photovoltaic module's maximum power generation, k represents the photovoltaic module's material degradation coefficient, and d represents the predicted dust thickness. The formula for calculating the cleaning cost index is as follows: I = (ΔP • C E ) / C W • (1 + 0.5 • R rain ); Where I represents the cleaning cost index, C E C represents the price cost per unit of electricity generated. W R represents the price cost per cleaning session. rain This indicates the probability of rainfall corresponding to a set future duration.

7. The method according to any one of claims 1 to 4, characterized in that, When the cleaning cost index and the predicted dust thickness meet the set conditions, a cleaning scheduling instruction corresponding to the photovoltaic module is generated, including: If the cleaning cost index is greater than the first set value and the predicted dust thickness is greater than the second set value, the cleaning scheduling instruction is generated. If the probability of rainfall within the specified future time frame is greater than the third set value, and the cleaning cost index is less than the fourth set value, the cleaning scheduling instruction corresponding to the specified future time frame will be delayed or canceled.

8. A photovoltaic module cleaning and scheduling device, characterized in that, include: The acquisition module is used to acquire environmental data and status data of the photovoltaic module to be cleaned, wherein the status data is used to indicate the operating status of the photovoltaic module. The prediction module is used to input the environmental data and the state data into a pre-trained dust accumulation prediction model and output the predicted dust thickness over a set time period in the future. The calculation module is used to input the predicted dust thickness into the cleaning cost prediction algorithm and output the corresponding cleaning cost index. The cleaning cost prediction algorithm is used to determine the energy consumption cost required for the dust corresponding to the predicted dust thickness. The cleaning cost index is an indicator used to represent the energy consumption cost. The determination module is used to generate a cleaning scheduling instruction corresponding to the photovoltaic module when the cleaning cost index and the predicted dust thickness meet the set conditions. The cleaning scheduling instruction is used to instruct the cleaning module corresponding to the photovoltaic module to clean the photovoltaic module after the set time.

9. A control device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 7.

Images