Photovoltaic power station dust coverage degree determination method and device based on neural network

Abstract

This application relates to a method and apparatus for determining dust coverage of a photovoltaic power station based on a neural network. The specific scheme is as follows: Acquire photovoltaic module operating condition data at N times on the Mth day after a new round of dust cleaning of the photovoltaic power station; send the photovoltaic module operating condition data at N times to a second server; acquire the predicted real-time output power of the power station at the N times sent by the second server; acquire the measured real-time output power of the power station at the N times on the Mth day; integrate the predicted real-time output power of the power station at the N times on the Mth day to obtain a first value; integrate the measured real-time output power of the power station at the N times on the Mth day to obtain a second value; based on the first and second values, determine the dust coverage of the photovoltaic power station on the Mth day. This application improves the accuracy of dust coverage calculation.

Description

Technical Field

[0001] This application relates to the field of photovoltaic power plant technology, and in particular to a method and apparatus for determining dust coverage in photovoltaic power plants based on neural networks. Background Technology

[0002] In related technologies, photovoltaic power plants can provide a set of sample strings at any time. These sample strings are cleaned daily by cleaning robots or manually. A comparative analysis is performed between the power generation or real-time current integral value of the sample string within a certain period and the power generation or real-time current integral value of the string affected by dust cover within the same period. The difference between the two is calculated, and the power generation loss caused by dust cover in the string is calculated from this difference. Furthermore, the power generation loss of the entire photovoltaic power plant due to dust cover is then calculated. Since the sample strings need to be cleaned daily, the cost of installing automatic cleaning robots is relatively high, and each cleaning robot can only handle a very small number of strings, mainly limited by the arrangement and distribution of the photovoltaic support structures. Therefore, the number of samples that a photovoltaic power plant can provide is generally small. Due to the inconsistency in the working state of the power plant strings, a small number of sample photovoltaic strings cannot describe the reference state of all strings in the power plant when they are not covered by dust. This leads to inaccurate calculations of power generation loss caused by dust cover, thus affecting the timeliness of string cleaning. Summary of the Invention

[0003] Therefore, this application provides a method and apparatus for determining dust coverage in photovoltaic power plants based on neural networks. The technical solution of this application is as follows:

[0004] According to a first aspect of the embodiments of this application, a method for determining dust coverage in a photovoltaic power station based on a neural network is provided, applied to a first server, the method comprising:

[0005] Obtain photovoltaic module operating condition data at time N on day M after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition data includes the operating temperature data of the photovoltaic module and the irradiance data received by the photovoltaic module; M is an integer greater than 0; N is an integer greater than 0;

[0006] The photovoltaic module operating condition data at the N time points are sent to the second server; the photovoltaic module operating condition data at the N time points are used to trigger the second server to determine the predicted real-time output power of the power plant at the N time points through a pre-trained power prediction model; the power prediction model is a neural network model constructed based on the photovoltaic module operating condition sample data at the N time points within a preset time period.

[0007] Obtain the predicted real-time output power of the power plant at the N time points sent by the second server;

[0008] Obtain the measured real-time output power of the power station at N times on the Mth day;

[0009] The predicted real-time output power of the power station at N times on the Mth day is integrated to obtain a first value, and the measured real-time output power of the power station at N times on the Mth day is integrated to obtain a second value.

[0010] Based on the first and second values, the dust coverage of the photovoltaic power station on day M is determined.

[0011] According to one embodiment of this application, determining the dust coverage of the photovoltaic power station on day M based on the first and second values ​​includes:

[0012] Subtract the first value from the second value to obtain the first difference;

[0013] Divide the first difference by the first value to obtain the change in dust coverage of the photovoltaic power station on day M relative to the preset time period;

[0014] Obtain historical dust coverage; wherein, the historical dust coverage is the dust coverage during the preset time period;

[0015] Based on the historical dust coverage and the change in dust coverage, the dust coverage of the photovoltaic power station on day M is obtained.

[0016] According to one embodiment of this application, before obtaining the photovoltaic module operating temperature data and irradiance data at time N on the Mth day after the completion of a new round of photovoltaic power station dust cleaning, the method further includes:

[0017] In response to the completion of a new round of dust cleaning at the photovoltaic power station, photovoltaic module operating condition sample data are acquired at N times within a preset time period according to a preset frequency; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and measured real-time output power of the power station sample data.

[0018] Photovoltaic module operating condition sample data at N times within the preset time period are sent to the second server; the photovoltaic module operating condition sample data at N times within the preset time period is used to trigger the second server to train the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period; the photovoltaic module operating condition sample data at N times within the preset time period is collected according to the preset frequency.

[0019] According to one embodiment of this application, before obtaining the photovoltaic module operating temperature data and irradiance data at time N on the Mth day after the completion of a new round of photovoltaic power station dust cleaning, the method further includes:

[0020] Obtain weather data for photovoltaic power plants;

[0021] Based on the weather data of the photovoltaic power station, determine whether day M+1 will be a rainy day;

[0022] In response to determining that the M+1th day is a rainy day, the M+1th day is designated as the new Mth day, and the step of determining whether the M+1th day is a rainy day based on the weather data of the photovoltaic power station is re-executed;

[0023] In response to determining that the (M+1)th day is a sunny day, the (M+1)th day is designated as the new (M)th day, and the step of determining whether the (M+1)th day is a rainy day based on the weather data of the photovoltaic power station is stopped.

[0024] According to one embodiment of this application, the N times within the preset time period correspond one-to-one with the N times on the Mth day; the N times on the (M+1)th day correspond one-to-one with the N times on the Mth day.

[0025] According to a second aspect of the embodiments of this application, a method for determining dust coverage in a photovoltaic power station based on a neural network is provided, applied to a second server, the method comprising:

[0026] Receive photovoltaic module operating condition data at N times on the Mth day from the first server; the Mth day is the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; M is an integer greater than 0; N is an integer greater than 0;

[0027] The operating condition data at N times on the Mth day are input into the pre-trained power prediction model;

[0028] Obtain the predicted real-time output power of the power plant at N times on the Mth day, as output by the pre-trained power prediction model.

[0029] The predicted real-time output power of the power plant at N times on day M is sent to the first server; the predicted real-time output power of the power plant at N times on day M is used to trigger the first server to determine the dust coverage of the photovoltaic power plant on day M based on the predicted real-time output power of the power plant at N times on day M and the measured real-time output power of the power plant at N times on day M.

[0030] According to one embodiment of this application, before receiving the photovoltaic module operating condition data at N times on the Mth day sent by the first server, the method further includes:

[0031] In response to receiving photovoltaic module operating condition sample data acquired at a preset frequency from the first server, the system trains the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain a first power prediction model; wherein, the photovoltaic module operating condition sample data consists of photovoltaic module operating condition sample data at N times within a preset time period after a new round of dust cleaning of the photovoltaic power station is completed; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time measured output power of the power station sample data.

[0032] The first power prediction model is determined as the pre-trained power prediction model.

[0033] According to a third aspect of the embodiments of this application, a photovoltaic power station dust coverage calculation device based on a neural network is provided, applied to a first server, the device comprising:

[0034] The first acquisition module is used to acquire photovoltaic module operating condition data at time N on the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition data includes photovoltaic module operating temperature data and photovoltaic module irradiance data; M is an integer greater than 0; N is an integer greater than 0;

[0035] The first sending module is used to send the photovoltaic module operating condition data at the N time points to the second server; the photovoltaic module operating condition data at the N time points is used to trigger the second server to determine the predicted real-time output power of the power plant at the N time points through a pre-trained power prediction model; the power prediction model is a neural network model constructed based on the photovoltaic module operating condition sample data at the N time points within a preset time period.

[0036] The second acquisition module is used to acquire the predicted real-time output power of the power plant at the N times sent by the second server;

[0037] The third acquisition module is used to acquire the measured real-time output power of the power station at N times on the Mth day;

[0038] An integration module is used to integrate the predicted real-time output power of the power station at N times on the Mth day to obtain a first value, and to integrate the measured real-time output power of the power station at N times on the Mth day to obtain a second value.

[0039] The first determining module is used to determine the dust coverage of the photovoltaic power station on day M based on the first value and the second value.

[0040] According to one embodiment of this application, the determining module includes:

[0041] The subtraction submodule is used to subtract the first value and the second value to obtain a first difference;

[0042] The phase division submodule is used to divide the first difference by the first value to obtain the change in dust coverage of the photovoltaic power station on day M relative to the preset time period.

[0043] The acquisition submodule is used to acquire historical dust coverage; wherein, the historical dust coverage is the dust coverage during the preset time period;

[0044] The addition submodule is used to obtain the dust coverage of the photovoltaic power station on day M based on the historical dust coverage and the change in dust coverage.

[0045] According to one embodiment of this application, the apparatus further includes:

[0046] The fourth acquisition module is used to acquire photovoltaic module operating condition sample data at N times within a preset time period according to a preset frequency in response to the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time output power measured value sample of the power station.

[0047] The second sending module is used to send photovoltaic module operating condition sample data at N times within the preset time period to the second server; the photovoltaic module operating condition sample data at N times within the preset time period is used to trigger the second server to train the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period; the photovoltaic module operating condition sample data at N times within the preset time period is collected according to the preset frequency.

[0048] According to one embodiment of this application, the apparatus further includes:

[0049] The second determining module is used to determine whether day M+1 will be a rainy day based on the weather data of the photovoltaic power station.

[0050] The third determining module is used to determine the M+1th day as a new Mth day in response to determining that the M+1th day is a rainy day, and to re-execute the step of determining whether the M+1th day is a rainy day based on the weather data of the photovoltaic power station.

[0051] The fourth determining module is used to, in response to determining that the (M+1)th day is a sunny day, determine the (M+1)th day as the new Mth day, stop executing the step of determining whether the (M+1)th day is a rainy day based on the weather data of the photovoltaic power station.

[0052] According to one embodiment of this application, the N times within the preset time period correspond one-to-one with the N times on the Mth day; the N times on the (M+1)th day correspond one-to-one with the N times on the Mth day.

[0053] According to a fourth aspect of the embodiments of this application, a photovoltaic power station dust coverage calculation device based on a neural network is provided, applied to a second server, the device comprising:

[0054] The first receiving module is used to receive photovoltaic module operating condition data at N times on the Mth day sent by the first server; the Mth day is the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; M is an integer greater than 0; N is an integer greater than 0.

[0055] The input module inputs the operating condition data at N times on the Mth day into the pre-trained power prediction model;

[0056] The acquisition module is used to acquire the predicted real-time output power of the power plant at N times on the Mth day, as output by the pre-trained power prediction model.

[0057] The sending module is used to send the predicted real-time output power of the power plant at N times on the Mth day to the first server; the predicted real-time output power of the power plant at N times on the Mth day is used to trigger the first server to determine the dust coverage of the photovoltaic power plant on the Mth day based on the predicted real-time output power of the power plant at N times on the Mth day and the measured real-time output power of the power plant at N times on the Mth day.

[0058] According to one embodiment of this application, the apparatus further includes:

[0059] The first training module is used to respond to receiving photovoltaic module operating condition sample data acquired at a preset frequency from the first server, and to train the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain a first power prediction model; wherein, the photovoltaic module operating condition sample data is photovoltaic module operating condition sample data at N times within a preset time period after a new round of dust cleaning of the photovoltaic power station is completed; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time output power measured value sample of the power station.

[0060] The fifth determining module is used to determine the first power prediction model as the pre-trained power prediction model.

[0061] According to a fifth aspect of the embodiments of this application, a storage medium is provided, wherein when the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in any one of the first aspects, or to perform the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in any one of the second aspects.

[0062] According to a sixth aspect of the embodiments of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in any one of the first aspects, or executes the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in any one of the second aspects.

[0063] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects:

[0064] By acquiring photovoltaic module operating condition data at N times on day M after the completion of a new round of dust cleaning at the photovoltaic power station; sending the photovoltaic module operating condition data at N times to a second server; obtaining the predicted real-time output power of the power station at N times sent by the second server; obtaining the measured real-time output power of the power station at N times on day M; integrating the predicted real-time output power of the power station at N times on day M to obtain a first value; integrating the measured real-time output power of the power station at N times on day M to obtain a second value; and determining the dust coverage of the photovoltaic power station on day M based on the first and second values. By predicting the change in dust coverage using the predicted and measured real-time output power of the power station, the accuracy of dust coverage calculation is improved, allowing for a more reasonable scheduling of photovoltaic module cleaning.

[0065] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0066] 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, and do not constitute an undue limitation of this application.

[0067] Figure 1 This is a flowchart illustrating a method for determining dust coverage in a photovoltaic power station based on a neural network, as described in this application.

[0068] Figure 2 This is a flowchart of another method for determining dust coverage in a photovoltaic power station based on a neural network, as described in this application.

[0069] Figure 3 This is a flowchart illustrating yet another method for determining dust coverage in a photovoltaic power station based on a neural network, as described in this application.

[0070] Figure 4 This is a structural block diagram of a photovoltaic power station dust coverage calculation device based on a neural network, as described in an embodiment of this application.

[0071] Figure 5 This is a structural block diagram of another photovoltaic power station dust coverage calculation device based on a neural network in an embodiment of this application;

[0072] Figure 6 This is a block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0073] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0074] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. 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.

[0075] It should be noted that dust accumulation on the surface of photovoltaic (PV) modules is one of the factors affecting their power generation efficiency. Severe dust accumulation can significantly reduce the output of grid-connected PV power plants and the conversion efficiency of PV modules. It is generally believed that the accumulation of dust on the surface of PV modules creates shading, hindering the absorption of solar radiation, which is the main reason why dust affects the output power of PV modules.

[0076] In related technologies, photovoltaic (PV) power plants can provide a set of sample strings at any time. These sample strings refer to strings that are cleaned daily by cleaning robots or manually; therefore, they are strings unaffected by dust cover. The number of sample strings in a PV power plant should be as large as possible, and these sample strings cannot be faulty. The power generation (or real-time current integral value) of the sample strings in a certain period is compared with that of strings affected by dust cover in the same period. The difference between the two is calculated, and the power generation loss caused by dust cover in the string is calculated from this difference. This difference is then used to calculate the power generation loss of the entire PV power plant caused by dust cover. Since the sample strings need to be cleaned daily, the cost of installing automatic cleaning robots is relatively high, and each cleaning robot can only handle a very small number of strings, mainly limited by the arrangement and distribution of the PV support structures. Therefore, the number of samples that a PV power plant can generally provide is very small. Furthermore, due to factors such as aging, the operating status of all strings cannot be synchronized (i.e., maintain the same value, which is evaluated by integrating the real-time current value of the string or the daily real-time current value of the string over time). This results in inaccurate calculation of power generation loss caused by dust covering the strings, which in turn affects the timeliness of string cleaning.

[0077] To address the aforementioned issues, this application proposes a method and apparatus for determining dust coverage in photovoltaic (PV) power plants based on neural networks. This method involves: acquiring PV module operating condition data at N times on day M after a new round of dust cleaning; sending this data to a second server; obtaining the predicted real-time output power of the power plant at N times from the second server; obtaining the measured real-time output power of the power plant at N times on day M; integrating the predicted real-time output power of the power plant at N times on day M to obtain a first value; integrating the measured real-time output power of the power plant at N times on day M to obtain a second value; and determining the dust coverage of the PV power plant on day M based on the first and second values. By predicting the change in dust coverage using the predicted and measured real-time output power of the power plant, the accuracy of dust coverage calculation is improved, allowing for a more reasonable scheduling of PV module cleaning.

[0078] Figure 1 This is a flowchart illustrating a method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application. It should be noted that the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application, is applied to a first server. Furthermore, the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application, can be used in the dust coverage calculation device for a photovoltaic power plant based on a neural network, which can be configured in an electronic device.

[0079] like Figure 1 As shown, the method for determining dust coverage in a photovoltaic power station based on a neural network includes:

[0080] Step 110: Obtain the photovoltaic module operating condition data at time N on day M after the completion of the new round of dust cleaning of the photovoltaic power station.

[0081] In this embodiment of the application, the photovoltaic module operating condition data includes the photovoltaic module's operating temperature data and the irradiance data received by the photovoltaic module.

[0082] It should be noted that the irradiance (i.e., real-time solar intensity) of photovoltaic (PV) modules is positively correlated with their real-time power output, but the relationship is not non-linear. The total daily power generation of a PV power plant is determined by the integral result of the real-time irradiance of the PV modules. The operating temperature of PV modules and the power output capacity of a PV power plant are inversely related; that is, when the operating temperature of PV modules increases, the power output capacity of the PV power plant will be suppressed and decrease.

[0083] Optionally, the operating temperature data of the photovoltaic module can be real-time measured data, or a method of fitting a prediction function with short-term historical data can be used to predict the operating temperature data of the photovoltaic module using measured air temperature and wind speed data.

[0084] In this embodiment of the application, M is an integer greater than 0.

[0085] In this embodiment of the application, N is an integer greater than 0.

[0086] As a possible example, the photovoltaic power station can be cleaned periodically according to actual needs. The first server obtains the photovoltaic module operating condition data at time N on the Mth day after the completion of the new round of dust cleaning of the photovoltaic power station.

[0087] Step 120: Send the photovoltaic module operating condition data at N time points to the second server.

[0088] In this embodiment of the application, the photovoltaic module operating condition data at N time points are used to trigger the second server to determine the predicted real-time output power of the power plant at N time points through a pre-trained power prediction model.

[0089] In this embodiment of the application, the power prediction model is a neural network model constructed based on photovoltaic module operating condition sample data at N times within a preset time period.

[0090] It should be noted that due to natural aging, the output power of photovoltaic modules will slowly decrease year by year, so the output power of photovoltaic power plants will also show a slow downward trend year by year. The output capacity of photovoltaic power plants decreases slowly year by year, but within a very short period of time, such as three to five days, the output capacity of photovoltaic power plants remains unchanged. Therefore, the output capacity of photovoltaic power plants is a constant value in the short term.

[0091] It is understandable that the power prediction model can be a neural network model. The power prediction model can be trained based on the photovoltaic module operating data of a certain day. The trained power prediction model can predict the real-time output power of the power station on the Mth day after that day.

[0092] As a possible example, the first server sends photovoltaic module operating condition data at N time points to the second server. After receiving the photovoltaic module operating condition data at N time points, the second server inputs the photovoltaic module operating condition data at N time points into a pre-trained power prediction model, thereby obtaining the real-time output power prediction values ​​of the power plant at N time points predicted by the power prediction model.

[0093] Step 130: Obtain the predicted real-time output power of the power plant at N time points sent by the second server.

[0094] As a possible example, the second server sends the predicted real-time output power of the power plant at N time points, as predicted by the power prediction model, to the first server.

[0095] Step 140: Obtain the measured real-time output power of the power station at the Nth time on day M.

[0096] As a possible example, the first server obtains the measured real-time output power of the power plant at time N on day M.

[0097] Step 150: Integrate the predicted real-time output power of the power station at N times on day M to obtain the first value; integrate the measured real-time output power of the power station at N times on day M to obtain the second value.

[0098] Integrating the predicted real-time output power of the photovoltaic power station at N times on day M, we obtain the first value Q. yc The specific calculation method is as follows:

[0099]

[0100] Among them, P i1 T represents each predicted value of the output power of a photovoltaic power plant. i1 This represents the time interval between two predicted values.

[0101] Integrating the measured real-time output power of the photovoltaic power station at N times on day M, we obtain the second value Q. sf The specific calculation method is as follows:

[0102]

[0103] Among them, P i2 T represents each measured value of the output power of a photovoltaic power station. i2 T represents the time interval between two measured values. i1 equal to T i2 .

[0104] Step 160: Based on the first and second values, determine the dust coverage of the photovoltaic power station on day M.

[0105] It's important to note that under natural conditions, dust accumulates on photovoltaic (PV) modules over time. This dust reduces the intensity of sunlight reaching the PV panels, resulting in power generation loss. Severe dust accumulation can lead to significant power loss, which is unacceptable for PV power plants. Therefore, dust coverage in PV power plants continuously increases over time, except during rainy days (due to the rinsing effect of rainwater). When dust accumulates to a certain level, it reaches an unacceptable threshold, requiring a thorough cleaning of the PV modules. After this cleaning, the dust coverage will return to zero, and the PV power plant will enter the next dust accumulation cycle. Within a short period (e.g., three to five days), the power output of a PV power plant remains constant. During this period, only newly added dust accumulation will cause a decrease in power output, which can be described by the real-time output power of the PV power plant.

[0106] As a possible implementation example, after determining the dust coverage of the photovoltaic power station on day M, the cleaning time of the photovoltaic modules can be more accurately determined based on the dust coverage on day M. This allows for timely and reasonable scheduling of the cleaning of the photovoltaic modules, ensuring both timeliness and economy in the cleaning process.

[0107] In some embodiments of this application, step 160 includes:

[0108] Step 161: Subtract the first value from the second value to obtain the first difference.

[0109] Step 162: Divide the first difference by the first value to obtain the change in dust coverage of the photovoltaic power station on day M relative to the preset time period.

[0110] Understandably, after a new round of dust cleaning of the photovoltaic power station is completed, if all days from day one to day M are sunny, the dust coverage will continue to increase, and the change in dust coverage will be positive. However, if there is at least one rainy day from day one to day M, the dust coverage will decrease due to the photovoltaic modules being washed away by rainwater, and the change in dust coverage may be negative.

[0111] It's important to note that the operating temperature sample data and the irradiance sample data received by the photovoltaic modules serve as input variables for the neural network model, determining the calculated real-time power output prediction value of the power plant. For example, training the neural network model using today's operating temperature sample data, irradiance sample data, and the measured real-time power output value yields a power prediction model that accurately reflects the precise relationship between today's operating temperature sample data, irradiance sample data, and the predicted real-time power output value. Because the output capacity of a photovoltaic power plant is constant over a very short period (e.g., three to five days), today's neural network model training results can be used for tomorrow or the day after. However, today's calculation model is only affected by today's dust cover, not by tomorrow's or the day after's dust cover; tomorrow's or the day after's dust cover will definitely be greater than today's. For this reason, today's power prediction model can be used to predict the real-time power output of the photovoltaic power plant tomorrow or the day after. If the power prediction model trained today is used to predict the real-time output power of the photovoltaic power station tomorrow, the predicted real-time output power of the power station tomorrow should be slightly larger than the measured real-time output power of the photovoltaic power station tomorrow. The difference between the predicted real-time output power of the power station tomorrow and the measured real-time output power of the photovoltaic power station tomorrow is the power generation loss caused by the difference in dust coverage between today and tomorrow.

[0112] As a possible example, the change in dust cover C of a photovoltaic power plant on day M. z The calculation method is as follows:

[0113]

[0114] Step 163: Obtain historical dust coverage.

[0115] In this embodiment of the application, the historical dust coverage is the dust coverage over a preset time period.

[0116] Understandably, since the output capacity of a photovoltaic power station is a constant value only in the short term, it is necessary to acquire new photovoltaic module operating condition sample data periodically and retrain the power prediction model using this new data. The aforementioned preset time period is determined based on a preset frequency. For example, if the preset frequency is set to 2 days, the power prediction model will be retrained every two days. Photovoltaic module operating condition sample data will be acquired at 24 time points (i.e., 24 moments) within 24 hours on January 1st. After training the power prediction model based on this data, photovoltaic module operating condition sample data will be acquired again at 24 time points within 24 hours on January 4th, and the model will be trained again based on this data. The preset time period is 24 hours on January 4th, and the historical dust coverage is the dust coverage determined based on the newly acquired photovoltaic module operating condition sample data within 24 hours on January 4th.

[0117] Step 164: Based on historical dust coverage and changes in dust coverage, obtain the dust coverage of the photovoltaic power station on day M.

[0118] As a possible example, the historical dust coverage is added to the change in dust coverage to obtain the dust coverage of the photovoltaic power plant on day M.

[0119] According to the neural network-based method for determining dust coverage in a photovoltaic power station according to embodiments of this application, the following steps are taken: First, obtain the photovoltaic module operating condition data at N times on the Mth day after a new round of dust cleaning of the photovoltaic power station is completed. Then, send the photovoltaic module operating condition data at N times to a second server. Next, obtain the predicted real-time output power of the power station at N times sent by the second server. Then, obtain the measured real-time output power of the power station at N times on the Mth day. Integrate the predicted real-time output power of the power station at N times on the Mth day to obtain a first value, and integrate the measured real-time output power of the power station at N times on the Mth day to obtain a second value. Based on the first and second values, determine the dust coverage of the photovoltaic power station on the Mth day. By predicting the change in dust coverage using the predicted and measured real-time output power of the power station, the accuracy of dust coverage calculation is improved, thereby enabling timely cleaning of the photovoltaic modules.

[0120] Figure 2 This is a flowchart illustrating another method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment. It should be noted that the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment, is applied to a first server. Furthermore, the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment, can be used in the neural network-based photovoltaic power plant dust coverage calculation device described in this application embodiment, which can be configured in an electronic device.

[0121] like Figure 2 As shown, the method for determining dust coverage in a photovoltaic power station based on a neural network includes:

[0122] Step 210: In response to the completion of a new round of dust cleaning of the photovoltaic power station, obtain photovoltaic module operating condition sample data at N times within a preset time period according to a preset frequency.

[0123] In some embodiments of this application, the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and power plant real-time output power measured value sample.

[0124] As a possible implementation example, in response to the completion of a new round of dust cleaning of the photovoltaic power station, the first server acquires photovoltaic module operating condition sample data at N times within a preset time period according to a preset frequency.

[0125] Step 220: Send the photovoltaic module operating condition sample data at N times within the preset time period to the second server.

[0126] In some embodiments of this application, photovoltaic module operating condition sample data at N times within a preset time period is used to trigger a second server to train a power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period.

[0127] In some embodiments of this application, the photovoltaic module operating condition sample data at N times within a preset time period are collected at a preset frequency.

[0128] As one possible implementation example, the first server sends photovoltaic module operating condition sample data at N times within a preset time period to the second server. After receiving the photovoltaic module operating condition sample data at N times within the preset time period, the second server trains the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period.

[0129] In some embodiments of this application, prior to step 230, the following steps are also included:

[0130] Step a1: Obtain weather data for the photovoltaic power station and determine whether day M is a rainy day based on the weather data.

[0131] It is understandable that rainwater washes away the dust on the surface of photovoltaic modules, gradually reducing the dust coverage of the photovoltaic power station rather than increasing it, which leads to inaccurate predictions from the pre-trained power prediction model.

[0132] As one possible implementation example, the first server obtains weather data from the photovoltaic power station and determines whether day M is a rainy day based on the weather data.

[0133] Step a2: In response to determining that day M is a rainy day, based on the weather data of the photovoltaic power station, determine whether day M+1 will be a rainy day.

[0134] As a possible implementation example, in response to determining that day M+1 is a rainy day, day M+1 is designated as the new day M, and step a1 is repeated. For example, if day M is day 5, and day 5 is determined to be a rainy day, then day 6 is designated as the new day M, and the process of determining whether day 6 is a rainy day continues. In step a3, in response to determining that day M+1 is a sunny day, day M+1 is designated as the new day M, and the process of determining whether day M+1 is a rainy day based on the weather data from the photovoltaic power station is stopped.

[0135] As an example of possible implementation, in response to determining that day M+1 is a sunny day, day M+1 is determined as the new day M, and step 210 is performed.

[0136] In some embodiments of this application, N times within a preset time period correspond one-to-one with N times on day M; and N times on day M+1 correspond one-to-one with N times on day M.

[0137] It should be noted that since the power prediction model needs to predict the real-time output power of the power plant at N times on day M based on the photovoltaic module operating condition data at N times on day M, the sample data used to train the power prediction model and the data used for prediction need to correspond one-to-one with the data acquisition time nodes to ensure the accuracy of the predicted real-time output power of the power plant.

[0138] Step 230: Obtain the photovoltaic module operating condition data at time N on day M after the completion of the new round of dust cleaning of the photovoltaic power station.

[0139] In this embodiment of the application, the photovoltaic module operating condition data includes the photovoltaic module's operating temperature data and the irradiance data received by the photovoltaic module.

[0140] In this embodiment of the application, M is an integer greater than 0; N is an integer greater than 0.

[0141] In the embodiments of this application, step 230 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they be described in detail.

[0142] Step 240: Send the photovoltaic module operating condition data at N time points to the second server.

[0143] In this embodiment of the application, the photovoltaic module operating condition data at N time points are used to trigger the second server to determine the predicted real-time output power of the power plant at N time points through a pre-trained power prediction model.

[0144] In the embodiments of this application, step 240 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they elaborate further.

[0145] Step 250: Obtain the predicted real-time output power of the power plant at N time points sent by the second server.

[0146] In the embodiments of this application, step 250 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they be described in detail.

[0147] Step 260: Obtain the measured real-time output power of the power station at the Nth time on day M.

[0148] In the embodiments of this application, step 260 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they be described in detail.

[0149] Step 270: Integrate the predicted real-time output power of the power station at N times on day M to obtain the first value, and integrate the measured real-time output power of the power station at N times on day M to obtain the second value.

[0150] In the embodiments of this application, step 270 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they be described in detail.

[0151] Step 280: Based on the first and second values, determine the dust coverage of the photovoltaic power station on day M.

[0152] In the embodiments of this application, step 280 can be implemented in any of the ways described in the various embodiments of this application. The embodiments of this application do not limit this, nor will they elaborate further.

[0153] According to the neural network-based method for determining dust coverage in photovoltaic power plants, in response to the completion of a new round of dust cleaning at the photovoltaic power plant, sample data of photovoltaic module operating conditions at N times within a preset time period are acquired at a preset frequency; this sample data is then sent to a second server. This process trains the power prediction model, ensuring the accuracy of the power prediction model's predictions.

[0154] Figure 3This is a flowchart illustrating another method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment. It should be noted that the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment, is applied to a second server. Furthermore, the method for determining dust coverage in a photovoltaic power plant based on a neural network, as described in this application embodiment, can be used in the neural network-based photovoltaic power plant dust coverage calculation device described in this application embodiment, which can be configured in an electronic device.

[0155] like Figure 3 As shown, the method for determining dust coverage in a photovoltaic power station based on a neural network includes:

[0156] Step 310: Receive photovoltaic module operating condition data at N times on the Mth day sent by the first server.

[0157] In this embodiment of the application, day M is the day after the completion of a new round of dust cleaning of the photovoltaic power station.

[0158] In this embodiment of the application, M is an integer greater than 0; N is an integer greater than 0.

[0159] As a possible example, the photovoltaic power station can be cleaned periodically according to actual needs. After the first server obtains the photovoltaic module operating condition data at time N on the Mth day after the completion of the new round of dust cleaning of the photovoltaic power station, it sends the photovoltaic module operating condition data to the second server.

[0160] Step 320: Input the operating condition data of N times on day M into the pre-trained power prediction model.

[0161] Understandably, the power prediction model needs to predict the real-time output power of the power plant at N times on day M.

[0162] As a possible example, the second server inputs the operating condition data at N times on day M into the pre-trained power prediction model.

[0163] Step 330: Obtain the predicted real-time power output of the power plant at N times on the Mth day from the pre-trained power prediction model.

[0164] As a possible example, the power prediction model predicts the real-time output power of the power plant at N times on day M based on the photovoltaic module operating condition data at N times on day M, and outputs the predicted real-time output power of the power plant at N times on day M.

[0165] Step 340: Send the predicted real-time output power of the power plant at N times on day M to the first server.

[0166] In this embodiment of the application, the predicted real-time output power of the power station at N times on day M is used to trigger the first server to determine the change in dust coverage of the photovoltaic power station on day M based on the predicted real-time output power of the power station at N times on day M and the measured real-time output power of the power station at N times on day M.

[0167] The second server sends the predicted real-time output power of the power plant at N times on day M to the first server. Based on the predicted real-time output power of the power plant at N times on day M and the measured real-time output power of the power plant at N times on day M, the first server determines the change in dust coverage of the photovoltaic power plant on day M.

[0168] In some embodiments of this application, prior to step 310, the following steps are also included:

[0169] Step b1: In response to receiving photovoltaic module operating condition sample data acquired at a preset frequency from the first server, train the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain the first power prediction model.

[0170] In this embodiment of the application, the photovoltaic module operating condition sample data is the photovoltaic module operating condition sample data at N times within a preset time period after the completion of a new round of dust cleaning of the photovoltaic power station.

[0171] In this embodiment of the application, the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and power station real-time output power measured value sample.

[0172] As a possible example, sample data of the photovoltaic module's operating temperature, the irradiance received by the photovoltaic module, and the real-time output power of the power plant are input into the neural network model to be trained. The trained neural network model determines the predicted value of the power plant's real-time output power based on the sample data of the photovoltaic module's operating temperature and the sample data of the irradiance received by the photovoltaic module. Based on the predicted value of the power plant's real-time output power and then the sample data of the measured value of the power plant's real-time output power, the training result of the neural network model is evaluated. If the evaluation result does not meet the requirements, the neural network model is retrained until the evaluation result meets the requirements. The neural network model is then determined as the first power prediction model.

[0173] Step b2: The first power prediction model is determined as the pre-trained power prediction model.

[0174] According to the neural network-based method for determining dust coverage in photovoltaic power plants according to embodiments of this application, the method involves receiving photovoltaic module operating condition data at N times on day M from a first server; inputting the operating condition data at N times on day M into a pre-trained power prediction model; obtaining the predicted real-time output power of the power plant at N times on day M from the pre-trained power prediction model; and sending the predicted real-time output power of the power plant at N times on day M back to the first server. This method predicts the real-time output power of the power plant at N times on day M, improving the accuracy of dust coverage calculation in photovoltaic power plants.

[0175] To implement the above embodiments, this application proposes a photovoltaic power plant dust coverage calculation system based on neural networks. This system includes a photovoltaic power plant dust coverage calculation and analysis module and a server equipped with Matlab (matrix & laboratory).

[0176] In some embodiments of this application, the photovoltaic power station dust coverage calculation and analysis module cannot independently complete all calculations. Based on existing technology, programs developed using function curve fitting and the basic principles of neural networks can perform two calculation tasks: function curve fitting and neural network model training. However, the calculation accuracy is significantly lower than that of Matlab. This patent's calculation process requires high accuracy, thus necessitating the use of Matlab. Matlab is expensive and requires specialized personnel to use, which is beyond the capabilities of on-duty personnel at new energy power stations. Furthermore, due to system security requirements, the new energy power generation company's business systems can only operate on an internal local area network, preventing direct information exchange between the photovoltaic power station dust coverage calculation and analysis module and the outside world. Based on these reasons, the specific execution steps are as follows:

[0177] (1) The photovoltaic operation and maintenance software platform runs on the internal local area network of the new energy power generation regional company.

[0178] (2) The dust coverage calculation and analysis module of photovoltaic power station is integrated on the photovoltaic operation and maintenance software platform.

[0179] (3) The Matlab software system is deployed on the research institute’s server.

[0180] (4) The photovoltaic power station dust coverage calculation and analysis module program regularly summarizes and stores the monthly collected photovoltaic power station wind speed data, air temperature data, and photovoltaic module operating temperature data into an Excel spreadsheet template file. Then, it automatically sends the template file data to the designated folder of the workstation designated by the new energy regional company through the internal local area network.

[0181] (5) The duty officer of the New Energy Regional Company shall check the specified folder under the workstation specified in step (4) every day to see if there are any newly added template files. If so, copy the newly added template files out through a secure USB flash drive and send them to the email address of the project leader of the research institute through the workstation connected to the public network.

[0182] (6) The project leader of the research institute logs into their email daily. If a new Excel template file is found, they download it, then import the data from the template file into the Matlab server via a secure USB drive. They then run the Matlab software to complete the curve fitting of the photovoltaic module's operating temperature prediction function. The curve fitting results are also saved in an Excel template file. The template file is copied using a secure USB drive and then sent via email to the duty officer's email address at the New Energy Regional Company via a workstation connected to the public network.

[0183] (7) The duty officer of the New Energy Regional Company regularly downloads the Excel template file from the email address, copies the template file to a secure USB flash drive, and then imports it into the designated folder of the workstation designated by the New Energy Regional Company.

[0184] (8) The photovoltaic power station dust coverage calculation and analysis module program regularly retrieves the latest Excel template file from the designated folder of the workstation designated by the new energy regional company through the internal local area network, imports the data in the template file into the system, and iterates the existing prediction model.

[0185] (9) The photovoltaic power station dust coverage calculation and analysis module program regularly (once every 3 days) summarizes and stores the photovoltaic power station irradiance data, photovoltaic module operating temperature data (prediction function calculation results), and photovoltaic power station real-time output power data into an Excel template file, and then automatically sends the template file to the designated folder of the workstation designated by the new energy regional company through the internal LAN.

[0186] (10) The Excel template file is sent to the email address of the project leader of the research institute. The process is exactly the same as step (5).

[0187] (11) The research institute project leader regularly downloads the Excel template file, then imports the data from the template file into a server equipped with Matlab, runs the Matlab software, builds a neural network training program, and obtains a neural network training model through training. The training model reads the input data to predict yesterday's real-time output power data of the photovoltaic power station, and summarizes and stores the prediction results in the Excel template file. The template file is copied out using a secure USB flash drive, and then the prediction results data are sent to the email address of the duty officer of the new energy regional company via email through a workstation connected to the public network. Then the duty officer imports the template file into the system. This part of the operation is completely consistent with (7).

[0188] It should be noted that the photovoltaic power station operates during the day and stops operating at night due to the absence of sunlight; therefore, all complete data is generated during the daytime. The neural network training model cannot run independently of the Matlab software environment. The neural network training model requires input data for prediction; otherwise, it cannot calculate predicted data. For these reasons, the system can only calculate the previous day's predicted data and cannot calculate today's predicted data in real time. The new energy regional company has staff on duty 24 hours a day. Data generated during the day is automatically packaged and pushed to the workstation by the program, and the duty officer will send it to the email address of the research institute's project leader during the night shift. This data will be processed by Matlab the following day, and the predicted values ​​will be sent back to the photovoltaic power station dust coverage calculation and analysis module program system the next day.

[0189] The operating principle of the photovoltaic power station dust coverage calculation and analysis module is as follows:

[0190] (1) The photovoltaic power station shall collect real-time data on the operating temperature of the photovoltaic modules, air temperature, and wind speed for three consecutive days each month. The data collection frequency shall be no less than once every minute. The data collection results shall be imported into the system by the power station staff.

[0191] (2) The photovoltaic power station dust coverage calculation and analysis module summarizes and stores the data from step (1) into an Excel template file, and then automatically sends the template file to the designated folder of the workstation designated by the new energy regional company through the internal local area network.

[0192] (3) The duty officer of the new energy regional company shall send the template file of step 2) to the email address of the project leader of the research institute in accordance with the prescribed procedures (detailed description in the system logic structure section).

[0193] (4) The research institute project leader downloads the template file, imports the data into a server configured with Matlab, runs the Matlab software, and completes the curve fitting of the prediction function. The curve fitting result of the prediction function is sent back to the email address of the duty officer of the New Energy Regional Company. The duty officer imports the data into the system; the data transmission process is described in detail in the system logic structure section.

[0194] (5) The dust coverage calculation and analysis module of the photovoltaic power station captures the curve fitting results of the prediction function and iterates the existing prediction function.

[0195] (6) The photovoltaic power station dust coverage calculation and analysis module captures the wind speed data and temperature data of the photovoltaic power station collected in the real-time database, and calculates the real-time operating temperature of the photovoltaic module through the prediction function obtained in step (5).

[0196] Description: Wind speed and temperature data for photovoltaic power plants are available from all power plants. The requirement here is to obtain minute-level data, i.e., data collection frequency of once per minute.

[0197] (7) The photovoltaic power station dust coverage calculation and analysis module retrieves photovoltaic power station irradiance data and photovoltaic power station real-time output power data from the real-time database, retrieves photovoltaic module operating temperature data (prediction function calculation results) from the database, summarizes and stores the three data into an Excel template file, and then automatically sends the template file to the designated folder of the workstation designated by the new energy regional company through the internal LAN.

[0198] (8) The duty officer of the new energy regional company shall send the template file of step 7) to the email address of the project leader of the research institute in accordance with the prescribed procedures (detailed description in the system logic structure section).

[0199] (9) The research institute project leader downloads the template file, imports the data into a server equipped with Matlab, runs the Matlab software, builds a neural network training program, and obtains a neural network training model through training. The training model reads the input data to predict yesterday's real-time output power data of the photovoltaic power station, and the prediction results are summarized and stored in an Excel template file. The template file is copied using a secure USB flash drive and then sent via email to the duty officer's email address at the New Energy Regional Company via a workstation connected to the public network. The duty officer then imports the template file into the system according to the prescribed procedures.

[0200] (10) The photovoltaic power station dust coverage calculation and analysis module retrieves yesterday's real-time output power measurement data from the real-time database, and combines it with the predicted value returned in step (9). The module calculates the change in dust coverage yesterday relative to the previous time by calculating the difference between the integrals of the two values.

[0201] Figure 4 This is a structural block diagram of a photovoltaic power station dust coverage calculation device based on a neural network, as described in an embodiment of this application. The device is applied to a first server. Figure 4 As shown, the device includes:

[0202] The first acquisition module 401 is used to acquire photovoltaic module operating condition data at time N on the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition data includes the operating temperature data of the photovoltaic module and the irradiance data received by the photovoltaic module; M is an integer greater than 0; N is an integer greater than 0;

[0203] The first sending module 402 is used to send photovoltaic module operating condition data at N time points to the second server; the photovoltaic module operating condition data at N time points is used to trigger the second server to determine the predicted real-time output power of the power plant at N time points through a pre-trained power prediction model; the power prediction model is a neural network model constructed based on photovoltaic module operating condition sample data at N time points within a preset time period;

[0204] The second acquisition module 403 is used to acquire the predicted real-time output power of the power plant at N time points sent by the second server.

[0205] The third acquisition module 404 is used to acquire the measured real-time output power of the power station at N times on day M;

[0206] The integration module 405 is used to integrate the predicted real-time output power of the power station at N times on day M to obtain a first value, and to integrate the measured real-time output power of the power station at N times on day M to obtain a second value.

[0207] Module 406 is used to determine the dust coverage of the photovoltaic power station on day M based on the first and second values.

[0208] According to one embodiment of this application, the determining module 406 includes:

[0209] The subtraction submodule is used to subtract the first value from the second value to obtain the first difference.

[0210] The phase division submodule is used to divide the first difference by the first value to obtain the change in dust coverage of the photovoltaic power station on day M relative to the preset time period.

[0211] The acquisition submodule is used to acquire historical dust coverage; where historical dust coverage is the dust coverage over a preset time period.

[0212] The summation submodule is used to obtain the dust coverage of the photovoltaic power station on day M based on the historical dust coverage and the change in dust coverage.

[0213] According to one embodiment of this application, the device further includes:

[0214] The fourth acquisition module is used to acquire photovoltaic module operating condition sample data at N times within a preset time period in response to the completion of a new round of dust cleaning of the photovoltaic power station. The photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time output power measured value sample of the power station.

[0215] The second sending module is used to send photovoltaic module operating condition sample data at N times within a preset time period to the second server; the photovoltaic module operating condition sample data at N times within the preset time period is used to trigger the second server to train the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period; the photovoltaic module operating condition sample data at N times within the preset time period is collected at a preset frequency.

[0216] According to one embodiment of this application, the device further includes:

[0217] The second determination module is used to determine whether day M+1 will be a rainy day based on weather data from the photovoltaic power station.

[0218] The third determination module is used to determine whether day M+1 is a rainy day in response to determining that day M+1 is a new day M, and to re-execute the step of determining whether day M+1 is a rainy day based on the weather data of the photovoltaic power station.

[0219] The fourth determination module is used to determine whether the M+1th day is a rainy day in response to determining that the M+1th day is a sunny day, setting the M+1th day as the new Mth day, stopping the execution of the steps based on the weather data of the photovoltaic power station, and determining whether the M+1th day is a rainy day.

[0220] According to one embodiment of this application, N times within a preset time period correspond one-to-one with N times on day M; and N times on day M+1 correspond one-to-one with N times on day M.

[0221] The photovoltaic power station dust coverage calculation device based on a neural network according to an embodiment of this application obtains photovoltaic module operating condition data at N times on the Mth day after a new round of dust cleaning of the photovoltaic power station; sends the photovoltaic module operating condition data at N times to a second server; obtains the predicted real-time output power of the power station at N times sent by the second server; obtains the measured real-time output power of the power station at N times on the Mth day; integrates the predicted real-time output power of the power station at N times on the Mth day to obtain a first value; integrates the measured real-time output power of the power station at N times on the Mth day to obtain a second value; and determines the dust coverage of the photovoltaic power station on the Mth day based on the first value and the second value. By predicting the change in dust coverage using the predicted real-time output power of the power station and the measured real-time output power of the power station, the accuracy of dust coverage calculation is improved, thereby enabling timely cleaning of photovoltaic modules.

[0222] To achieve the above embodiments, this application proposes a photovoltaic power plant dust coverage calculation device based on neural networks.

[0223] Figure 5 This is a structural block diagram of another photovoltaic power plant dust coverage calculation device based on a neural network, as described in this application embodiment. This device is applied to a second server. Figure 5 As shown, the device includes:

[0224] The first receiving module 501 is used to receive photovoltaic module operating condition data at N times on the Mth day sent by the first server; the Mth day is the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; M is an integer greater than 0; N is an integer greater than 0;

[0225] Input module 502 is used to input the operating condition data of N times on day M into the pre-trained power prediction model;

[0226] The acquisition module 503 is used to acquire the real-time output power prediction values ​​of the power plant at N times on the Mth day, output by the pre-trained power prediction model.

[0227] The sending module 504 is used to send the predicted real-time output power of the power plant at N times on day M to the first server; the predicted real-time output power of the power plant at N times on day M is used to trigger the first server to determine the dust coverage of the photovoltaic power plant on day M based on the predicted real-time output power of the power plant at N times on day M and the measured real-time output power of the power plant at N times on day M.

[0228] According to one embodiment of this application, the device further includes:

[0229] The first training module is used to respond to receiving photovoltaic module operating condition sample data sent by the first server at a preset frequency, and to train the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain the first power prediction model; wherein, the photovoltaic module operating condition sample data is the photovoltaic module operating condition sample data at N times within a preset time period after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time output power measured value sample of the power station.

[0230] The fifth determination module is used to determine the first power prediction model as the pre-trained power prediction model.

[0231] The photovoltaic power plant dust coverage calculation device based on a neural network according to an embodiment of this application receives photovoltaic module operating condition data at N times on day M from a first server; inputs the operating condition data at N times on day M into a pre-trained power prediction model; obtains the predicted real-time output power of the power plant at N times on day M from the pre-trained power prediction model; and sends the predicted real-time output power of the power plant at N times on day M to the first server. This improves the accuracy of calculating the dust coverage of the photovoltaic power plant.

[0232] Figure 6 This is a block diagram of an electronic device according to an embodiment of this application. For example... Figure 6 As shown, the electronic device may include: a transceiver 61, a processor 62, and a memory 63.

[0233] Processor 62 executes computer execution instructions stored in memory, causing processor 62 to perform the scheme in the above embodiments. Processor 62 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0234] The memory 63 is connected to the processor 62 via the system bus and completes communication between them. The memory 63 is used to store computer program instructions.

[0235] Transceiver 61 can be used to obtain the task to be run and its configuration information.

[0236] The system bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus. Transceivers are used to enable communication between database access devices and other computers (e.g., clients, read-write libraries, and read-only libraries). Memory may include random access memory (RAM) and may also include non-volatile memory.

[0237] The electronic device provided in this application embodiment can be the terminal device described in the above embodiments.

[0238] This application also provides a chip for executing instructions, which is used to execute the message processing method described in the above embodiments.

[0239] This application also provides a computer-readable storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the message processing method described in the above embodiments.

[0240] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium, and when the at least one processor executes the computer program, it can implement the technical solution of the message processing method in the above embodiments.

[0241] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0242] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for determining dust coverage in photovoltaic power plants based on neural networks, characterized in that, Applied to a first server, the method includes: Obtain photovoltaic module operating condition data at time N on day M after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition data includes the operating temperature data of the photovoltaic module and the irradiance data received by the photovoltaic module; M is an integer greater than 0; N is an integer greater than 0; The photovoltaic module operating condition data at the N time points are sent to the second server; the photovoltaic module operating condition data at the N time points are used to trigger the second server to determine the predicted real-time output power of the power plant at the N time points through a pre-trained power prediction model; the power prediction model is a neural network model constructed based on the photovoltaic module operating condition sample data at the N time points within a preset time period. Obtain the predicted real-time output power of the power plant at the N time points sent by the second server; Obtain the measured real-time output power of the power station at N times on the Mth day; The predicted real-time output power of the power station at N times on the Mth day is integrated to obtain a first value, and the measured real-time output power of the power station at N times on the Mth day is integrated to obtain a second value. Based on the first and second values, determine the dust coverage of the photovoltaic power station on day M; Before acquiring the photovoltaic module operating condition data at time N on day M after the completion of a new round of dust cleaning of the photovoltaic power station, the method further includes: in response to the completion of a new round of dust cleaning of the photovoltaic power station, acquiring photovoltaic module operating condition sample data at N times within a preset time period according to a preset frequency; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and measured real-time output power of the power station sample data; sending the photovoltaic module operating condition sample data at N times within the preset time period to the second server; the photovoltaic module operating condition sample data at N times within the preset time period is used to trigger the second server to train the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period; the photovoltaic module operating condition sample data at N times within the preset time period is collected according to the preset frequency. Before acquiring the photovoltaic module operating condition data at N times on the Mth day after the completion of a new round of dust cleaning at the photovoltaic power station, the method further includes: acquiring weather data of the photovoltaic power station, and determining whether the Mth day is a rainy day based on the weather data; in response to determining that the Mth day is a rainy day, determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station; in response to determining that the M+1th day is a rainy day, determining the M+1th day as the new Mth day, and re-executing the step of determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station; in response to determining that the M+1th day is a sunny day, determining the M+1th day as the new Mth day, and stopping the step of determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station.

2. The method according to claim 1, characterized in that, The determination of the dust coverage of the photovoltaic power station on day M based on the first and second values ​​includes: Subtract the first value from the second value to obtain the first difference; Divide the first difference by the first value to obtain the change in dust coverage of the photovoltaic power station on day M relative to the preset time period; Obtain historical dust coverage; wherein, the historical dust coverage is the dust coverage during the preset time period; Based on the historical dust coverage and the change in dust coverage, the dust coverage of the photovoltaic power station on day M is obtained.

3. The method according to claim 1, characterized in that, The N times within the preset time period correspond one-to-one with the N times on the Mth day; the N times on the (M+1)th day correspond one-to-one with the N times on the Mth day.

4. A method for determining dust coverage in photovoltaic power plants based on neural networks, characterized in that, Applied to a second server, the method includes: Receive photovoltaic module operating condition data at N times on the Mth day from the first server; the Mth day is the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; M is an integer greater than 0; N is an integer greater than 0; Input the operating condition data at N times on the Mth day into the pre-trained power prediction model; obtain the real-time output power prediction values ​​of the power plant at N times on the Mth day output by the pre-trained power prediction model; The predicted real-time output power of the power plant at N times on day M is sent to the first server; the predicted real-time output power of the power plant at N times on day M is used to trigger the first server to determine the dust coverage of the photovoltaic power plant on day M based on the predicted real-time output power of the power plant at N times on day M and the measured real-time output power of the power plant at N times on day M. Before receiving the photovoltaic module operating condition data at N times on the Mth day sent by the first server, the method further includes: In response to receiving photovoltaic module operating condition sample data acquired at a preset frequency from the first server, the system trains the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain a first power prediction model; wherein, the photovoltaic module operating condition sample data consists of photovoltaic module operating condition sample data at N times within a preset time period after a new round of dust cleaning of the photovoltaic power station is completed; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time measured output power of the power station sample data. The first power prediction model is determined as the pre-trained power prediction model.

5. A photovoltaic power station dust coverage calculation device based on neural networks, characterized in that, Applied to a first server, the device includes: The first acquisition module is used to acquire photovoltaic module operating condition data at time N on the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; the photovoltaic module operating condition data includes photovoltaic module operating temperature data and photovoltaic module irradiance data; M is an integer greater than 0; N is an integer greater than 0; The first sending module is used to send the photovoltaic module operating condition data at the N time points to the second server; the photovoltaic module operating condition data at the N time points is used to trigger the second server to determine the predicted real-time output power of the power plant at the N time points through a pre-trained power prediction model; the power prediction model is a neural network model constructed based on the photovoltaic module operating condition sample data at the N time points within a preset time period. The second acquisition module is used to acquire the predicted real-time output power of the power plant at the N times sent by the second server; The third acquisition module is used to acquire the measured real-time output power of the power station at N times on the Mth day; An integration module is used to integrate the predicted real-time output power of the power station at N times on the Mth day to obtain a first value, and to integrate the measured real-time output power of the power station at N times on the Mth day to obtain a second value. The first determining module is used to determine the dust coverage of the photovoltaic power station on day M based on the first value and the second value. Before acquiring the photovoltaic module operating condition data at time N on day M after the completion of a new round of dust cleaning of the photovoltaic power station, the method further includes: in response to the completion of a new round of dust cleaning of the photovoltaic power station, acquiring photovoltaic module operating condition sample data at N times within a preset time period according to a preset frequency; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and measured real-time output power of the power station sample data; sending the photovoltaic module operating condition sample data at N times within the preset time period to the second server; the photovoltaic module operating condition sample data at N times within the preset time period is used to trigger the second server to train the power prediction model based on the photovoltaic module operating condition sample data at N times within the preset time period; the photovoltaic module operating condition sample data at N times within the preset time period is collected according to the preset frequency. Before acquiring the photovoltaic module operating condition data at N times on the Mth day after the completion of a new round of dust cleaning at the photovoltaic power station, the method further includes: acquiring weather data of the photovoltaic power station, and determining whether the Mth day is a rainy day based on the weather data; in response to determining that the Mth day is a rainy day, determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station; in response to determining that the M+1th day is a rainy day, determining the M+1th day as the new Mth day, and re-executing the step of determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station; in response to determining that the M+1th day is a sunny day, determining the M+1th day as the new Mth day, and stopping the step of determining whether the M+1th day will be a rainy day based on the weather data of the photovoltaic power station.

6. A photovoltaic power plant dust coverage calculation device based on neural networks, characterized in that, Applied to a second server, the device includes: The first receiving module is used to receive photovoltaic module operating condition data at N times on the Mth day sent by the first server; the Mth day is the Mth day after the completion of a new round of dust cleaning of the photovoltaic power station; M is an integer greater than 0; N is an integer greater than 0. The input module inputs the operating condition data at N times on the Mth day into the pre-trained power prediction model; The acquisition module is used to acquire the predicted real-time output power of the power plant at N times on the Mth day, as output by the pre-trained power prediction model. The sending module is used to send the predicted real-time output power of the power plant at N times on the Mth day to the first server; the predicted real-time output power of the power plant at N times on the Mth day is used to trigger the first server to determine the dust coverage of the photovoltaic power plant on the Mth day based on the predicted real-time output power of the power plant at N times on the Mth day and the measured real-time output power of the power plant at N times on the Mth day. Before receiving the photovoltaic module operating condition data at N times on the Mth day sent by the first server, the method further includes: In response to receiving photovoltaic module operating condition sample data acquired at a preset frequency from the first server, the system trains the neural network model to be trained based on the photovoltaic module operating condition sample data to obtain a first power prediction model; wherein, the photovoltaic module operating condition sample data consists of photovoltaic module operating condition sample data at N times within a preset time period after a new round of dust cleaning of the photovoltaic power station is completed; the photovoltaic module operating condition sample data includes photovoltaic module operating temperature sample data, photovoltaic module irradiance sample data, and real-time measured output power of the power station sample data. The first power prediction model is determined as the pre-trained power prediction model.

7. A storage medium, wherein when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is able to perform the method for determining dust coverage of a photovoltaic power plant based on a neural network as described in any one of claims 1 to 3, or to perform the method for determining dust coverage of a photovoltaic power plant based on a neural network as described in claim 4.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in any one of claims 1 to 3, or it implements the method for determining the dust coverage of a photovoltaic power station based on a neural network as described in claim 4.

Images