An AGC precise power control algorithm and system considering communication failures of multiple photovoltaic devices

Abstract

The present invention relates to the technical field of power grid power control, and specifically to an AGC precise power control algorithm and system considering communication failures of multiple photovoltaic devices, comprising: obtaining monitoring image information of the photovoltaic device with the communication failure; obtaining the ash layer thickness of the photovoltaic device; obtaining the current performance data of the photovoltaic device; predicting the first current output data of the photovoltaic device; obtaining the second current output data and AGC adjustment instructions of the remaining photovoltaic devices, the AGC adjustment instructions being the target power value; and performing power control based on the first current output data, the second current output data and the target power value. The present invention obtains the ash layer thickness of the photovoltaic device with the communication failure through monitoring image information, and then accurately obtains the current performance data of the photovoltaic device with the communication failure, providing a reliable basis for predicting the output. This avoids the problem of being unable to grasp the device status due to communication interruption, and greatly improves the reliability of the entire photovoltaic system.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid power control, and in particular to an AGC precise power control algorithm and system considering communication failures of multiple photovoltaic devices. Background Art

[0002] With the increasing global emphasis on and utilization of renewable energy, photovoltaic power generation has gained widespread development as a clean and sustainable energy source. The Automatic Generation Control (AGC) system plays a crucial role in the power generation process of photovoltaic equipment. It adjusts the output power of photovoltaic equipment in real time according to the needs of the power grid to ensure stable operation of the grid.

[0003] However, in practical applications, photovoltaic systems may face various problems, among which communication failures are a common challenge. When communication failures occur among multiple photovoltaic devices, traditional AGC systems may be unable to accurately obtain output data from the faulty devices, resulting in inaccurate power control and affecting grid stability. Summary of the Invention

[0004] (1) Purpose of the invention

[0005] The purpose of the present invention is to provide an AGC precise power control algorithm and system considering communication failures of multiple photovoltaic devices, which can more accurately predict the output data of photovoltaic devices with communication failures.

[0006] (2) Technical solution

[0007] To solve the above problems, the present invention provides an AGC precise power control algorithm considering communication failures of multiple photovoltaic devices, including:

[0008] Obtain monitoring image information of photovoltaic equipment with communication failure;

[0009] Obtaining a thickness of a dust layer of the photovoltaic device according to the image information;

[0010] obtaining current performance data of the photovoltaic device according to the thickness of the ash layer;

[0011] predicting first current output data of the photovoltaic device based on the current performance data;

[0012] Obtaining second current output data and AGC adjustment instructions of other photovoltaic devices, wherein the AGC adjustment instructions are target power values;

[0013] Power control is performed according to the first current output data, the second current output data and the target power value.

[0014] In another aspect of the present invention, preferably, obtaining the thickness of the ash layer of the photovoltaic device according to the image information includes:

[0015] splicing and segmenting the image information to obtain a first image;

[0016] Obtaining a recognized first dust image according to the first image and a preset dust layer recognition model;

[0017] Calculating a similarity between the first dust image and a preset reference dust image to obtain a second dust image, where the second dust image is the reference dust image with the highest similarity;

[0018] The thickness of the dust layer of the photovoltaic device is obtained according to the second dust image.

[0019] In another aspect of the present invention, preferably, the gray layer recognition model is based on a fully convolutional model, and the gray layer recognition model is trained using the following loss function:

[0020]

[0021] Among them, L represents the loss function, I represents the original image, Y represents the labeled image, W represents the parameters of the gray layer recognition model, P j represents the predicted value of the j-th pixel in the original image, Pr represents the probability of the predicted pixel being 0 or 1, Y+ represents the number of predicted pixels being 1, Y- represents the number of predicted pixels being 0, c0 and c1 represent the total number of non-gray layer pixels 0 and gray layer pixels 1 in the training set, respectively.

[0022] In another aspect of the present invention, preferably, the calculating the similarity between the first dust image and a preset reference dust image includes:

[0023] Performing discrete cosine transform on the dust image to obtain a DCT coefficient matrix, where the dust image includes a first dust image and a preset reference dust image;

[0024] Extracting a submatrix of a preset size from the DCT coefficient matrix;

[0025] Calculating the mean of the submatrix;

[0026] Comparing each DCT coefficient in the submatrix with the mean of the submatrix to generate a binary hash value;

[0027] Calculating a Hamming distance between a binary hash value of the first dust image and a binary hash value of a preset reference dust image;

[0028] The similarity between the first dust image and a preset reference dust image is calculated according to the Hamming distance.

[0029] In another aspect of the present invention, preferably, the current performance data of the photovoltaic device is obtained using the following formula:

[0030] η'=η·exp(-r·D)

[0031] Where η' represents the current photovoltaic device conversion efficiency, η represents the initial photovoltaic device conversion efficiency, D represents the dust thickness, and r represents the attenuation coefficient.

[0032] In another aspect of the present invention, preferably,

[0033] The attenuation coefficient is calculated using the following formula:

[0034]

[0035] Where r represents the attenuation coefficient, d i It represents the difference between the level of the i-th ash layer thickness and the level of the output data reduction rate. The level means converting the data values ​​of the ash layer thickness and the output data reduction rate into levels. If the data values ​​are the same, the average level is assigned. n represents the number of data pairs of ash layer thickness and output data reduction rate.

[0036] In another aspect of the present invention, preferably,

[0037] The first current output data is predicted using the following formula:

[0038]

[0039] Among them, E represents the first current output data, H represents the irradiation, η' represents the current photovoltaic equipment conversion efficiency, P az Indicates installed capacity, E sc represents the irradiance under STC conditions, Losses represents the system loss, t sun Indicates the sunshine time in a cycle.

[0040] In another aspect of the present invention, preferably,

[0041] The irradiation amount is calculated using the following formula:

[0042]

[0043] Where H represents the radiation dose, H BR Indicates the horizontal direct radiation, H DR Represents the horizontal scattered radiation, R BR The ratio of direct solar radiation on the photovoltaic module surface to the horizontal surface, H RT represents the ground reflection, β represents the angle between the photovoltaic device and the horizontal plane, H w Indicates the solar radiation on the horizontal surface outside the atmosphere, H RTIndicates the amount of ground reflection.

[0044] In another aspect of the present invention, preferably, controlling according to the first current output data, the second current output data and the target power value includes:

[0045] If the second current output data is greater than or equal to the target power value, the remaining photovoltaic devices output according to the target power value;

[0046] If the second current output data is less than the target power value, and the sum of the first current output data and the second current output data is greater than or equal to the target power value, calculating the ratio of the first current output data to the second current output data, and calculating the output of the photovoltaic device and the remaining photovoltaic devices according to the ratio and the target power value;

[0047] If the sum of the first current output data and the second current output data is less than the target power value, the photovoltaic device and the remaining photovoltaic devices output full power.

[0048] In another aspect of the present invention, preferably, an AGC precise power control system considering communication failures of multiple photovoltaic devices includes:

[0049] The first acquisition module is used to acquire monitoring image information of photovoltaic equipment with communication failure;

[0050] A second acquisition module: acquiring the thickness of the ash layer of the photovoltaic device according to the image information;

[0051] A third acquisition module: acquiring current performance data of the photovoltaic device according to the thickness of the ash layer;

[0052] Prediction module: predicting first current output data of the photovoltaic device according to the current performance data;

[0053] A fourth acquisition module is configured to acquire second current output data and AGC adjustment instructions of the remaining photovoltaic devices, wherein the AGC adjustment instructions are target power values;

[0054] Control module: performs power control according to the first current output data, the second current output data and the target power value.

[0055] (3) Beneficial effects

[0056] The above technical solution of the present invention has the following beneficial technical effects:

[0057] The present invention obtains the thickness of the dust layer of the photovoltaic equipment with communication failure by monitoring image information, and then accurately obtains the current performance data of the photovoltaic equipment with communication failure, providing a reliable basis for predicting output. It avoids the problem of being unable to grasp the status of the equipment due to communication interruption, and greatly improves the reliability of the entire photovoltaic system. It does not rely on traditional communication channels to obtain equipment information, reduces the risk of the system's dependence on a single communication method, and enhances the stability of the system in complex environments. Combining the second current output data of the remaining photovoltaic equipment and the AGC adjustment instruction, that is, the target power value for power control, can achieve precise power regulation of the entire photovoltaic system, better meet the needs of the power grid, and improve the stability of the power grid and the quality of power. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is an overall flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present invention.

[0060] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] In the description of the present invention, it should be noted that the terms "first", "second" and "third" are only used for descriptive purposes and should not be understood as indicating or implying relative importance.

[0062] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0063] The present invention will be described in more detail below with reference to the accompanying drawings. In each of the accompanying drawings, identical elements are represented by similar reference numerals. For the sake of clarity, the various parts in the accompanying drawings are not drawn to scale.

[0064] Example 1

[0065] An AGC precise power control algorithm considering communication failures of multiple photovoltaic devices. Figure 1 FIG. 1 shows an overall flow chart of an embodiment of the present invention, as shown in FIG. Figure 1 Shown, including:

[0066] Acquire monitoring images of photovoltaic equipment experiencing communication failures. Photovoltaic equipment is installed in a photovoltaic power station equipped with monitoring systems at various angles. If communication failures occur across multiple photovoltaic equipment, the first step is to obtain monitoring images of the equipment experiencing the communication failure using these monitoring systems. These monitoring images can capture the appearance of the photovoltaic equipment experiencing the communication failure from various angles.

[0067] The thickness of the dust layer of the photovoltaic device is obtained based on the image information; the acquired monitoring image information is processed using image analysis technology, and the thickness of the dust layer on the surface of the photovoltaic device can be inferred by identifying features in the image, such as color changes, texture differences, etc.

[0068] In this embodiment, obtaining the thickness of the ash layer of the photovoltaic device according to the image information includes:

[0069] The image information is stitched and segmented to obtain a first image. Image information acquired from various monitoring angles within a photovoltaic power station may be fragmented and incomplete. Image stitching technology can integrate these images from different angles into a coherent, complete image, providing a more comprehensive view of the photovoltaic equipment with a communication failure. The stitched image is then segmented to separate the photovoltaic equipment with a communication failure from the background, enabling more accurate analysis of the dust layer on the equipment.

[0070] A first dust image is obtained based on the first image and a preset gray layer recognition model. The preset gray layer recognition model is trained with a large amount of data and is capable of accurately identifying gray layers in an image. The first image is input into the preset gray layer recognition model, which analyzes the first image, identifies the gray layer portion therein, and generates a first dust image.

[0071] Furthermore, in this embodiment, the preset gray layer recognition model is trained using the following loss function:

[0072]

[0073] Among them, L represents the loss function, I represents the original image, Y represents the labeled image, W represents the parameters of the gray layer recognition model, P j represents the predicted value of the j-th pixel in the original image, Pr represents the probability of the predicted pixel being 0 or 1, Y+ represents the number of predicted pixels being 1, Y- represents the number of predicted pixels being 0, c0 and c1 represent the total number of non-gray layer pixels 0 and gray layer pixels 1 in the training set, respectively.

[0074] The similarity between the first dust image and a preset baseline dust image is calculated to obtain a second dust image, which is the baseline dust image with the highest similarity. A set of baseline dust images with different dust layer thicknesses is preset, representing different degrees of dust layer conditions. The similarity between the first dust image and each baseline dust image can be calculated using an image similarity algorithm, such as the structural similarity index (SSIM) or peak signal-to-noise ratio (PSNR). By comparing the similarities, the baseline dust image with the highest similarity to the first dust image is found and used as the second dust image. By comparing with a known baseline image, the thickness of the dust layer on the photovoltaic device can be more accurately determined.

[0075] The thickness of the photovoltaic device's dust layer is obtained based on the second dust image. A second dust image with the highest similarity to the first dust image is identified, and the photovoltaic device's dust layer thickness is determined based on the dust layer thickness corresponding to the second dust image. Because the preset baseline dust image is associated with a specific dust layer thickness, the thickness value corresponding to the second dust image can be directly used as the photovoltaic device's dust layer thickness, enabling accurate determination of the photovoltaic device's dust layer thickness through image analysis.

[0076] In this embodiment, calculating the similarity between the first dust image and a preset reference dust image includes:

[0077] A discrete cosine transform (DCT) coefficient matrix is ​​obtained by performing a discrete cosine transform on the dust image. The dust image includes a first dust image and a preset baseline dust image. The discrete cosine transform (DCT) is a mathematical transformation method that converts an image from the spatial domain to the frequency domain. After performing a discrete cosine transform on the first dust image and the preset baseline dust image, the corresponding DCT coefficient matrix is ​​obtained. The DCT coefficient matrix reflects the energy distribution of the image across different frequency components. In the frequency domain, the low-frequency portion of an image typically contains the image's primary information, while the high-frequency portion corresponds to image detail and noise. The DCT transform concentrates the image's energy on a small number of low-frequency coefficients, thereby achieving data compression and feature extraction.

[0078] A submatrix of a preset size is extracted from the DCT coefficient matrix. To further extract image features, a submatrix of the preset size is selected from the DCT coefficient matrix. The size of this submatrix can be adjusted based on actual conditions; a smaller size can better capture local features of the image. The position of the submatrix can be randomly selected from the DCT coefficient matrix or selected according to a specific rule, such as from the center region of the matrix or within a specific frequency range.

[0079] Calculate the mean of the submatrix; for the extracted submatrix, calculate the mean of all its elements. The mean can reflect the average size of the DCT coefficients in the submatrix. The mean is calculated by adding the values ​​of all elements in the submatrix and then dividing by the number of elements in the submatrix.

[0080] Each DCT coefficient in the submatrix is ​​compared to the mean of the submatrix to generate a binary hash value. Each DCT coefficient in the submatrix is ​​then compared to the calculated mean. If the DCT coefficient is greater than or equal to the mean, the corresponding bit is set to 1; otherwise, it is set to 0. This generates a binary hash value that reflects the relative magnitude of the DCT coefficients in the submatrix to the mean. Binary hash values ​​are simple to calculate and easy to store, making them suitable for quickly comparing image similarities.

[0081] Calculate the Hamming distance between the binary hash value of the first dust image and the binary hash value of a preset reference dust image. The Hamming distance is the number of different characters at corresponding positions in two strings of equal length. Here, calculating the Hamming distance between the binary hash value of the first dust image and the binary hash value of the preset reference dust image involves counting the number of different binary bits in the two hash values.

[0082] Based on the Hamming distance, the similarity between the first dust image and a preset baseline dust image is calculated. The similarity between the first dust image and the preset baseline dust image is determined based on the calculated Hamming distance. Similarity can be defined as 1 minus the ratio of the Hamming distance and the length of the hash value. For example, if the length of the hash value is n and the Hamming distance is d, the similarity can be calculated as 1-d / n. A similarity value between 0 and 1 is obtained, which is used to measure the degree of similarity between the two images.

[0083] Current performance data of the photovoltaic system is obtained based on the ash layer thickness. Ash layer thickness significantly impacts photovoltaic system performance. The thicker the ash layer, the lower the photovoltaic system's power generation efficiency. Based on the known relationship between ash layer thickness and photovoltaic system performance, current photovoltaic system performance data can be determined.

[0084] In this embodiment, the current performance data of the photovoltaic device is obtained using the following formula:

[0085] η'=η·exp(-r·D)

[0086] Where η' represents the current photovoltaic device conversion efficiency, η represents the initial photovoltaic device conversion efficiency, D represents the dust thickness, and r represents the attenuation coefficient. The attenuation coefficient can be obtained through laboratory experiments. During laboratory experiments, output data corresponding to different dust layer thicknesses are collected to form several data pairs. Based on these data pairs, the attenuation coefficient is calculated.

[0087] The attenuation coefficient is calculated using the following formula:

[0088]

[0089] Where r represents the attenuation coefficient, d i It represents the difference between the level of the i-th ash layer thickness and the level of the output data reduction rate. The level means converting the data values ​​of the ash layer thickness and the output data reduction rate into levels. If the data values ​​are the same, the average level is assigned. n represents the number of data pairs of ash layer thickness and output data reduction rate.

[0090] According to the current performance data, the first current output data of the photovoltaic device is predicted; based on the acquired current performance data and combined with meteorological conditions, the first current output data of the photovoltaic device can be predicted using a statistical model based on historical data or a mathematical model based on physical principles.

[0091] In this embodiment, the first current output data is predicted using the following formula:

[0092]

[0093] Among them, E represents the first current output data, H represents the irradiation, η' represents the current photovoltaic equipment conversion efficiency, P az Indicates installed capacity, E sc represents the irradiance under STC conditions, Losses represents the system loss, t sun Indicates the sunshine duration within a cycle. Using the current photovoltaic equipment conversion efficiency to calculate the first current output data, the prediction accuracy is higher.

[0094] The irradiation amount is calculated using the following formula:

[0095]

[0096] Where H represents the radiation dose, H BR Indicates the horizontal direct radiation, H DR Represents the horizontal scattered radiation, R BR The ratio of direct solar radiation on the photovoltaic module surface to the horizontal surface, H RT represents the ground reflection, β represents the angle between the photovoltaic device and the horizontal plane, H w Indicates the solar radiation on the horizontal surface outside the atmosphere, H RT Represents the amount of ground reflection. By comprehensively considering multiple factors that affect irradiance, the actual irradiance received by PV systems can be more accurately calculated. Based on real-time meteorological data and PV system installation parameters, irradiance is dynamically calculated, supporting intelligent control and optimized operation of PV systems.

[0097] Obtain the second current output data and AGC adjustment instructions for the remaining photovoltaic devices. The AGC adjustment instructions are target power values. Obtain the second current output data for the remaining photovoltaic devices in the photovoltaic power station that are communicating normally. This can be obtained through traditional communication channels. In addition, obtain the AGC adjustment instructions. The AGC adjustment instructions are given in the form of target power values, representing the power demand of the photovoltaic power station by the power grid.

[0098] Power control is performed based on the first current output data, the second current output data, and the target power value. A reasonable power control strategy is formulated based on the first current output data (predicted output of the faulty device), the second current output data (output of the normal device), and the target power value. For example, an optimization algorithm can be used to adjust the output power of each photovoltaic device while meeting the target power value to achieve optimal operation of the entire photovoltaic system.

[0099] In this embodiment, controlling according to the first current output data, the second current output data, and the target power value includes:

[0100] If the second current output data is greater than or equal to the target power value, the remaining PV systems will output at the target power value. When the second current output data of the remaining PV systems is greater than or equal to the target power value, the PV systems with normal communication alone can meet the grid's target power demand. In this case, if the target power value is 1000 kilowatts and the second current output data of the remaining PV systems is 1200 kilowatts, the PV systems with normal communication can adjust their output power to stabilize at 1000 kilowatts to meet the grid's requirements.

[0101] If the second current output data is less than the target power value, and the sum of the first and second current output data is greater than or equal to the target power value, the ratio of the first and second current output data is calculated. Based on this ratio and the target power value, the output of the PV system and the remaining PV systems is calculated. This ratio reflects the relative contribution of the faulty PV system and the remaining PV systems to meeting the target power value. For example, if the first current output data is 400 kilowatts, the second current output data is 600 kilowatts, and the target power value is 900 kilowatts, then the ratio is 400 / 600 = 2 / 3. If the target power value is 900 kilowatts, the ratio is 2 / 3. Therefore, the output of the remaining PV systems is 900 × 3 / 5 = 540 kilowatts, and the output of the faulty PV system is 900 × 2 / 5 = 360 kilowatts. While ensuring that the target power value is met, the output of the faulty PV system and the remaining PV systems is rationally allocated to achieve precise power control.

[0102] If the sum of the first and second current output data is less than the target power value, the PV system and the remaining PV systems will operate at full output. To maximize grid demand, the faulty PV system and the remaining PV systems must operate at full output. Full output means these systems will operate at their maximum possible output power to approach the target power value as closely as possible.

[0103] The present invention obtains the thickness of the dust layer of the photovoltaic equipment with communication failure by monitoring image information, and then accurately obtains the current performance data of the photovoltaic equipment with communication failure, providing a reliable basis for predicting output. It avoids the problem of being unable to grasp the status of the equipment due to communication interruption, and greatly improves the reliability of the entire photovoltaic system. It does not rely on traditional communication channels to obtain equipment information, reduces the risk of the system's dependence on a single communication method, and enhances the stability of the system in complex environments. Combining the second current output data of the remaining photovoltaic equipment and the AGC adjustment instruction, that is, the target power value for power control, can achieve precise power regulation of the entire photovoltaic system, better meet the needs of the power grid, and improve the stability of the power grid and the quality of power.

[0104] It should be understood that the above-described specific embodiments of the present invention are merely illustrative or illustrative of the principles of the present invention and do not constitute limitations of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included within the scope of protection of the present invention. In addition, the appended claims are intended to cover all variations and modifications that fall within the scope and metes and bounds of the appended claims, or equivalents thereof.

[0105] The present invention has been described above with reference to the embodiments thereof. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. The scope of the present invention is defined by the appended claims and their equivalents. Those skilled in the art may make various substitutions and modifications without departing from the scope of the present invention, and such substitutions and modifications are intended to fall within the scope of the present invention.

[0106] Although the embodiments of the present invention have been described in detail, it should be understood that the various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention.

[0107] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. An AGC precise power control algorithm considering communication failures of multiple photovoltaic devices, characterized by: include: Obtain monitoring image information of photovoltaic equipment with communication failure; Obtaining the thickness of the ash layer of the photovoltaic device according to the image information, including: splicing and segmenting the image information to obtain a first image; Obtaining a recognized first dust image according to the first image and a preset dust layer recognition model; Calculating a similarity between the first dust image and a preset reference dust image to obtain a second dust image, where the second dust image is the reference dust image with the highest similarity; acquiring a dust layer thickness of the photovoltaic device according to the second dust image; obtaining current performance data of the photovoltaic device according to the thickness of the ash layer; predicting first current output data of the photovoltaic device based on the current performance data; Obtaining second current output data and AGC adjustment instructions of other photovoltaic devices, wherein the AGC adjustment instructions are target power values; performing power control according to the first current output data, the second current output data, and the target power value; The calculating of the similarity between the first dust image and a preset reference dust image includes: Performing discrete cosine transform on the dust image to obtain a DCT coefficient matrix, where the dust image includes a first dust image and a preset reference dust image; Extracting a submatrix of a preset size from the DCT coefficient matrix; Calculating the mean of the submatrix; Comparing each DCT coefficient in the submatrix with the mean of the submatrix to generate a binary hash value; Calculating a Hamming distance between a binary hash value of the first dust image and a binary hash value of a preset reference dust image; Calculating a similarity between the first dust image and a preset reference dust image according to the Hamming distance; The current performance data of the photovoltaic device is obtained using the following formula: in, Indicates the current photovoltaic equipment conversion efficiency, represents the initial photovoltaic device conversion efficiency, D represents the dust thickness, and r represents the attenuation coefficient; The first current output data is predicted using the following formula: Among them, E represents the first current output data, H represents the radiation amount, Indicates the current photovoltaic equipment conversion efficiency, P az Indicates installed capacity, E sc represents the irradiance under STC conditions, Losses represents the system loss, t sun Indicates the sunshine time in a cycle.

2. The control algorithm according to claim 1, characterized in that: The gray layer recognition model is based on a fully convolutional model and is trained using the following loss function: Among them, L represents the loss function, I represents the original image, Y represents the labeled image, W represents the parameters of the gray layer recognition model, P j represents the predicted value of the j-th pixel in the original image, Pr represents the probability of the predicted pixel being 0 or 1, Y+ represents the number of predicted pixels being 1, Y- represents the number of predicted pixels being 0, c0 and c1 represent the total number of non-gray layer pixels 0 and gray layer pixels 1 in the training set, respectively.

3. The control algorithm according to claim 1, characterized in that: The attenuation coefficient is calculated using the following formula: Where r represents the attenuation coefficient, d i It represents the difference between the level of the i-th ash layer thickness and the level of the output data reduction rate. The level means converting the data values ​​of the ash layer thickness and the output data reduction rate into levels. If the data values ​​are the same, the average level is assigned. n represents the number of data pairs of ash layer thickness and output data reduction rate.

4. The control algorithm according to claim 1, characterized in that: The irradiation amount is calculated using the following formula: Where H represents the radiation dose, H BR Indicates the horizontal direct radiation, H DR Represents the horizontal scattered radiation, R BR The ratio of direct solar radiation on the photovoltaic module surface to the horizontal surface, H RT represents the ground reflection, β represents the angle between the photovoltaic device and the horizontal plane, H w Indicates the solar radiation on the horizontal surface outside the atmosphere, H RT Indicates the amount of ground reflection.

5. The control algorithm according to claim 2, characterized in that: The controlling according to the first current output data, the second current output data and the target power value includes: If the second current output data is greater than or equal to the target power value, the remaining photovoltaic devices output according to the target power value; If the second current output data is less than the target power value, and the sum of the first current output data and the second current output data is greater than or equal to the target power value, calculating the ratio of the first current output data to the second current output data, and calculating the output of the photovoltaic device and the remaining photovoltaic devices according to the ratio and the target power value; If the sum of the first current output data and the second current output data is less than the target power value, the photovoltaic device and the remaining photovoltaic devices output full power.

6. An AGC precise power control system considering communication failures of multiple photovoltaic devices, characterized by: include: The first acquisition module is used to acquire monitoring image information of photovoltaic equipment with communication failure; The second acquisition module is configured to acquire the thickness of the ash layer of the photovoltaic device according to the image information, including: splicing and segmenting the image information to obtain a first image; Obtaining a recognized first dust image according to the first image and a preset dust layer recognition model; Calculating a similarity between the first dust image and a preset reference dust image to obtain a second dust image, where the second dust image is the reference dust image with the highest similarity; acquiring a dust layer thickness of the photovoltaic device according to the second dust image; A third acquisition module: acquiring current performance data of the photovoltaic device according to the thickness of the ash layer; Prediction module: predicting first current output data of the photovoltaic device according to the current performance data; A fourth acquisition module is configured to acquire second current output data and AGC adjustment instructions of other photovoltaic devices, wherein the AGC adjustment instructions are target power values; A control module: performing power control according to the first current output data, the second current output data and the target power value; The calculating of the similarity between the first dust image and a preset reference dust image includes: Performing discrete cosine transform on the dust image to obtain a DCT coefficient matrix, where the dust image includes a first dust image and a preset reference dust image; Extracting a submatrix of a preset size from the DCT coefficient matrix; Calculating the mean of the submatrix; Comparing each DCT coefficient in the submatrix with the mean of the submatrix to generate a binary hash value; Calculating a Hamming distance between a binary hash value of the first dust image and a binary hash value of a preset reference dust image; Calculating a similarity between the first dust image and a preset reference dust image according to the Hamming distance; The current performance data of the photovoltaic device is obtained using the following formula: in, Indicates the current photovoltaic equipment conversion efficiency, represents the initial photovoltaic device conversion efficiency, D represents the dust thickness, and r represents the attenuation coefficient; The first current output data is predicted using the following formula: Among them, E represents the first current output data, H represents the radiation amount, Indicates the current photovoltaic equipment conversion efficiency, P az Indicates installed capacity, E sc represents the irradiance under STC conditions, Losses represents the system loss, t sun Indicates the sunshine time in a cycle.

Images