Transformer area output prediction device and prediction method suitable for transformer area micro-grid autonomy

Abstract

The invention discloses a transformer area output prediction device suitable for transformer area micro-grid autonomy, which comprises a master device and a plurality of slave devices, and is characterized in that the master device is connected in parallel with a sky acquisition imager and a meteorological element sensor to acquire weather data; and the master device is in data parallel connection with the plurality of slave devices so as to record and manage the high-time-resolution data from the plurality of slave devices. Aiming at higher requirements of a novel digital intelligent power distribution network on distributed photovoltaic output prediction, the output prediction device which is low in cost, higher in prediction time resolution and lower in energy consumption is provided. The problem that at present, distributed photovoltaic connected to a distribution network side lacks ultra-short-term high-time-resolution power generation power prediction is solved.

Description

Technical Field

[0001] The invention relates to the field of distributed energy storage technology, and in particular to a substation output prediction device and prediction method suitable for substation microgrid autonomy. Background Art

[0002] Building a new power system with new energy as the main body is an important method and approach to achieve the goals of carbon peak and carbon neutrality. Compared with the traditional power system, the morphology and structure of the new power system will undergo major changes. As an organic complement to large energy bases, distributed new energy will occupy an important position in the future energy system. Unlike centralized new energy, which is connected to the grid on the main grid side, distributed new energy will be connected to the grid on the distribution grid side closer to the user. The distribution network is the access carrier for distributed power sources and electric vehicle charging facilities. It is the key support for the comprehensive and efficient utilization of various distributed energy sources and energy conservation and emission reduction for power users. In this context, the digital transformation of the distribution network is imperative, and the construction of a modern smart distribution network is an important practice to support the transformation. The modern smart distribution network can be regarded as the distribution network form of the new power system. Through modern information and communication technologies such as "big cloud, things, mobile, and intelligence", artificial intelligence technology is deeply integrated with the active distribution network, and the new distribution system is empowered with digitalization, intelligence, and wisdom to achieve the development goals of a safe, reliable, economical, efficient, clean and low-carbon modern distribution network. Renewable energy power generation has the characteristics of intermittency and randomness. Large-scale grid connection has a great impact on the stability, continuity and adjustability of the power grid. In order to ensure the smooth operation of the power system, it is necessary to predict the power generation capacity of renewable energy.

[0003] At present, the power prediction system of photovoltaic stations widely used in China mainly adopts the power prediction technology based on numerical weather forecast. Numerical weather forecast is a meteorological forecasting method that uses large computers to solve the equations describing atmospheric motion with numerical analysis methods under given initial conditions and boundary conditions, and predicts the weather conditions for a period of time in the future based on the known initial state of the atmosphere. There is a close relationship between numerical weather forecast (NWP) and photovoltaic output prediction. Numerical weather forecast provides information on various weather elements including temperature, radiation, etc., which has an important impact on the power output of photovoltaic power stations. By inputting NWP data into the power prediction model, the accuracy of photovoltaic output prediction can be improved.

[0004] Specifically, the application of NWP in photovoltaic power generation prediction mainly includes the following steps:

[0005] Data assimilation: This step involves fusing real-time meteorological observations into numerical weather models to optimize the model’s initial conditions, thereby improving forecast accuracy.

[0006] Power downscaling: Since PV plants usually cover a small area, while the area simulated by NWP may cover hundreds of thousands of square kilometers, it is necessary to downscale conventional NWP data to the specific location and height of PV plants to meet the needs of refined power forecasting.

[0007] Adjustment of time resolution: PV power forecasts usually require higher time resolution (e.g. once every 15 minutes), while NWP may provide hourly data. Therefore, it is necessary to improve the time resolution through technical means to meet the needs of power forecasting.

[0008] Ensemble forecast: Due to the uncertainty of weather forecast, ensemble forecast provides the most likely weather forecast value by summarizing multiple forecast results, and gives the predicted confidence interval by calculating the discreteness of each forecast result, which helps the decision-making of power grid dispatch and power trading.

[0009] However, the traditional centralized output prediction method is not suitable for distributed photovoltaics connected to the grid on the distribution network side. This is mainly due to the scattered and disordered geographical locations of distributed photovoltaics, the low capacity of single stations, and the lack of meteorological element collection devices. Distributed photovoltaic systems are scattered in different geographical locations, and the environmental conditions (such as sunshine, temperature and shadow) of each system may be different. These differences require more fine-grained environmental data and more sophisticated prediction models for the prediction of distributed photovoltaic systems. Collecting more fine-grained environmental data requires the deployment of a large number of meteorological element collection devices, which will greatly increase the cost of the prediction system. It is also very difficult to improve the prediction precision of the model. First, the atmospheric system itself can be regarded as a chaotic system, which is naturally extremely difficult to predict. Then, it is very difficult and energy-consuming to establish and solve more sophisticated prediction models. On the one hand, the parameters in the atmospheric motion and energy transfer equations solved by numerical weather forecasting are statistical parameters on a larger time and space scale. Many factors on a fine scale, such as the movement of clouds, are still difficult to accurately predict through numerical weather forecasting. On the other hand, solving more sophisticated models will greatly increase energy consumption. Currently, numerical weather forecasts rely on large and super-large computers. More sophisticated models will greatly increase the amount of calculations and increase electricity consumption. However, the capacity of a single distributed photovoltaic station is small, and the electricity generated by a single station is relatively limited. It will be very uneconomical to equip each distributed photovoltaic power station with a power prediction service based on numerical weather forecasting technology.

[0010] In addition, the time resolution of photovoltaic output prediction technology based on numerical weather forecast is usually once every fifteen minutes, which is difficult to meet the needs of smart distribution network under the new situation. In the context of the new power system, the distribution network will be connected to a large number of new power sources and new loads. For example, people's demand for fast charging of electric vehicles is more urgent. High-power short-time charging scenarios put forward higher requirements on the time resolution of distributed photovoltaic output prediction.

[0011] In summary, traditional photovoltaic prediction methods are mainly designed for centralized photovoltaic stations in the context of traditional power systems where new energy accounts for a small proportion. The methods based on numerical weather forecasts have high calculation costs, rough time resolution, and weak perception of the impact of clouds on solar irradiance. In view of the above shortcomings, this invention provides a substation output prediction device suitable for autonomous substation microgrids, which performs more precise, higher time resolution, and low energy consumption prediction of distributed photovoltaic output connected to the new smart distribution network. Summary of the invention

[0012] The purpose of the present invention is to solve the shortcomings of the prior art, propose a power output prediction device and prediction method suitable for autonomous substation microgrids, and provide a power output prediction device with low cost, higher prediction time resolution, and lower energy consumption in response to the higher requirements for distributed photovoltaic output prediction for new digital intelligent distribution networks. This solves the problem that the distributed photovoltaics currently connected to the distribution network lack ultra-short-term high-time resolution power generation prediction.

[0013] The present invention discloses a substation output prediction device suitable for substation microgrid autonomy, comprising a master device and multiple slave devices, wherein a sky imager and a meteorological element sensor are connected in parallel on the master device to collect weather data, and the data on the master device is connected in parallel to multiple slave devices to record and manage high time resolution data from multiple slave devices.

[0014] Preferably, the main device includes a housing module, a main control board module, an AI computing power module, an antenna module, a serial port interface module, an Ethernet interface module, an HPLC interface module, a power module, an encryption module and a communication module at the hardware level;

[0015] The protection level of the housing module shall not be lower than the IP51 requirement specified in GB / T 4208;

[0016] The mainboard module includes the main control CPU, memory, flash memory module, and security chip. The CPU is a safe and controllable industrial-grade chip. To provide sufficient AI computing power, the memory supports 4GB or more. To provide a large amount of data storage capacity, the flash memory module is greater than or equal to 64GB and supports expansion. The security chip supports the terminal's own protection and the security protection when the terminal interacts with the main station, distributed new energy power stations, and operation and maintenance tools.

[0017] The AI ​​computing power module provides dedicated hardware support for AI algorithms to meet the computing power requirements of AI algorithms.

[0018] Antenna module, including transmission antenna, Beidou interface and pulse-second interface, used for remote communication, geographic location information collection and time synchronization;

[0019] Serial port interface module, used for communication between master and slave devices, with multiple RS-485 and multiple RS-232 / RS-485 channels and supports serial port switching;

[0020] Ethernet interface module, used for master station communication or remote, HPLC module, used for master device and slave device communication, to adapt to a variety of different scenarios of communication with the slave device;

[0021] Power module, used to power terminal equipment;

[0022] Encryption module, used for longitudinal data encryption, local communication data encryption and local storage data encryption during longitudinal data transmission between the terminal and the master station. The encryption module meets the safety requirements of power dispatching;

[0023] The communication module is used to process the communication protocol used when the master device communicates with the slave device or other devices.

[0024] Preferably, the main device includes an operating system and functional applications at the software level. The operating system includes an autonomous and controllable embedded Linux operating system and a hardware adapter driver. The security functions involve secure access, data encryption, trusted authentication and security baselines. Business applications include predictive algorithms, data management and data collection.

[0025] Preferably, the slave device is connected to and communicates with the inverter of the distributed photovoltaic power generation component, collects real-time information of the photovoltaic component including voltage, current, power and other information with high-frequency time resolution, and sends it to the master device; wherein, there are multiple distributed photovoltaic power generation users in a substation area, and there are multiple distributed photovoltaic power generation component inverters, each inverter is connected to a slave device, and multiple slave devices are connected to a master device together.

[0026] A prediction method for a power output prediction device for a power station area that is applicable to a power station area microgrid autonomy. The prediction device collects ground cloud images through a sky imager, integrates information such as voltage, current, power, etc. of photovoltaic power generation components and information such as temperature and humidity collected by other related sensors, and predicts the output of distributed photovoltaic power generation in the power station area. The steps include:

[0027] C. The fluctuation of solar irradiance is the direct cause of the fluctuation of photovoltaic output. This method uses ground-based cloud images to analyze the impact of cloud shading and scattering on solar irradiance. The cloud image data is captured by a sky imager, and the original cloud image data is pre-processed by image enhancement, image denoising, overexposure suppression, etc.

[0028] D. The ground cloud image after image preprocessing is passed through a multi-layer artificial neural network model consisting of convolutional layers, pooling layers, and normalization layers to extract high-level abstract feature information related to cloud images and photovoltaic output;

[0029] E. The high-level abstract feature information extracted by the artificial neural network model and the photovoltaic power generation component information collected from the equipment, such as voltage, current, power, etc., are passed through the feature fusion neural network layer composed of a fully connected neural network to generate an initial photovoltaic output prediction value;

[0030] F. Local meteorological factors, such as water vapor in the air, can also cause small fluctuations in solar irradiance. Temperature can affect the photoelectric conversion efficiency of photovoltaic power generation components and can also affect the output of photovoltaic components. Therefore, it is also necessary to correct the initial photovoltaic output forecast value based on the local meteorological factors near the photovoltaic components to generate the final photovoltaic output forecast value.

[0031] The advantages and technical effects of the present invention are:

[0032] 1. Based on the fact that cloud shading and scattering are the main factors causing fluctuations in distributed photovoltaic output, the present invention uses a sky imager to capture cloud images in the sky, extracts and analyzes information in the cloud images using a deep neural network model, combines important operating parameters of photovoltaic modules with weather factors, and uses artificial intelligence methods to predict the distributed photovoltaic output of the substation area. Therefore, the capture of cloud conditions is more timely and accurate, and the analysis is more in-depth, which is the main advantage of this method.

[0033] 2. Due to the use of advanced artificial intelligence methods, the present invention provides a substation output prediction device suitable for substation microgrid autonomy, which has the advantages of fast calculation speed and high reliability.

[0034] 3. The main device of the present invention has AI computing power and is deployed on the station side. Since the processing power is closer to the edge, the Internet bandwidth required by business applications becomes less, which greatly reduces network costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of system connection of the present invention.

[0036] Figure 2 It is a structural diagram of the hardware layer of the main device of the present invention.

[0037] Figure 3It is a schematic diagram of the framework of the main device software layer in the present invention;

[0038] Figure 4 It is a structural diagram of the hardware layer of the slave device in the present invention;

[0039] Figure 5 Flow chart of the grayscale publishing method in an embodiment of the present invention;

[0040] In the figure: A1-housing module; A2-main control board module; A3-AI computing module; A4-antenna module; A5-serial port interface module; A6-Ethernet interface module; A7-HPLC interface module; A8-power module; A9-encryption module; A10-communication module;

[0041] B1-housing; B2-slave main control board module; B3-slave serial port interface module; B4-slave HPLC interface module; B5-slave power supply module. DETAILED DESCRIPTION

[0042] The embodiments of the invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the invention, and cannot be understood as limiting the invention.

[0043] In the description of the invention, it should be understood that the terms "center", "longitudinal", "lateral", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0044] The present invention discloses a substation output prediction device suitable for substation microgrid autonomy, comprising a master device and multiple slave devices, wherein a sky imager and a meteorological element sensor are connected in parallel on the master device to collect weather data, and the data on the master device is connected in parallel to multiple slave devices to record and manage high time resolution data from multiple slave devices.

[0045] Preferably, the main device includes a housing module A1, a main control board module A2, an AI computing module A3, an antenna module A4, a serial port interface module A5, an Ethernet interface module A6, an HPLC interface module A7, a power module A8, an encryption module A9 and a communication module A10 at the hardware level;

[0046] The protection level of the housing module shall not be lower than the IP51 requirement specified in GB / T 4208;

[0047] The mainboard module includes the main control CPU, memory, flash memory module, and security chip. The CPU is a safe and controllable industrial-grade chip. To provide sufficient AI computing power, the memory supports 4GB or more. To provide a large amount of data storage capacity, the flash memory module is greater than or equal to 64GB and supports expansion. The security chip supports the terminal's own protection and the security protection when the terminal interacts with the main station, distributed new energy power stations, and operation and maintenance tools.

[0048] The AI ​​computing power module provides dedicated hardware support for AI algorithms to meet the computing power requirements of AI algorithms.

[0049] Antenna module, including transmission antenna, Beidou interface and pulse-second interface, used for remote communication, geographic location information collection and time synchronization;

[0050] Serial port interface module, used for communication between master and slave devices, with multiple RS-485 and multiple RS-232 / RS-485 channels and supports serial port switching;

[0051] Ethernet interface module, used for master station communication or remote, HPLC module, used for master device and slave device communication, to adapt to a variety of different scenarios of communication with the slave device;

[0052] Power module, used to power terminal equipment;

[0053] Encryption module, used for longitudinal data encryption, local communication data encryption and local storage data encryption during longitudinal data transmission between the terminal and the master station. The encryption module meets the safety requirements of power dispatching;

[0054] The communication module is used to process the communication protocol used when the master device communicates with the slave device or other devices.

[0055] Preferably, the main device includes an operating system and functional applications at the software level. The operating system includes an autonomous and controllable embedded Linux operating system and a hardware adapter driver. The security functions involve secure access, data encryption, trusted authentication and security baselines. Business applications include predictive algorithms, data management and data collection.

[0056] Preferably, the slave device includes a housing B1, a slave main control board module B2, a slave serial port interface module B3, a slave HPLC interface module B4, and a slave power supply module B5.

[0057] Preferably, the slave device is connected to and communicates with the inverter of the distributed photovoltaic power generation component, collects real-time information of the photovoltaic component including voltage, current, power and other information with high-frequency time resolution, and sends it to the master device; wherein, there are multiple distributed photovoltaic power generation users in a substation area, and there are multiple distributed photovoltaic power generation component inverters, each inverter is connected to a slave device, and multiple slave devices are connected to a master device together.

[0058] A prediction method for a power output prediction device for a power station area that is applicable to a power station area microgrid autonomy. The prediction device collects ground cloud images through a sky imager, integrates information such as voltage, current, power, etc. of photovoltaic power generation components and information such as temperature and humidity collected by other related sensors, and predicts the output of distributed photovoltaic power generation in the power station area. The steps include:

[0059] C. The fluctuation of solar irradiance is the direct cause of the fluctuation of photovoltaic output. This method uses ground-based cloud images to analyze the impact of cloud shading and scattering on solar irradiance. The cloud image data is captured by a sky imager, and the original cloud image data is pre-processed by image enhancement, image denoising, overexposure suppression, etc.

[0060] D. The ground cloud image after image preprocessing is passed through a multi-layer artificial neural network model consisting of convolutional layers, pooling layers, and normalization layers to extract high-level abstract feature information related to cloud images and photovoltaic output;

[0061] E. The high-level abstract feature information extracted by the artificial neural network model and the photovoltaic power generation component information collected from the equipment, such as voltage, current, power, etc., are passed through the feature fusion neural network layer composed of a fully connected neural network to generate an initial photovoltaic output prediction value;

[0062] F. Local meteorological factors, such as water vapor in the air, can also cause small fluctuations in solar irradiance. Temperature can affect the photoelectric conversion efficiency of photovoltaic power generation components and can also affect the output of photovoltaic components. Therefore, it is also necessary to correct the initial photovoltaic output forecast value based on the local meteorological factors near the photovoltaic components to generate the final photovoltaic output forecast value.

[0063] In order to more clearly describe the specific implementation mode of the present invention, an embodiment is provided below:

[0064] The present invention provides a power output prediction device and prediction method for a substation microgrid autonomy, which realizes edge computing deployment and high-time resolution data acquisition based on a "master-slave" architecture. The prediction device consists of a master device and several slave devices (such as Figure 1 shown) where:

[0065] A. The master device has relatively rich edge AI computing power, which can provide sufficient hardware support for the output prediction algorithm program. In addition, the master device records and manages high-time resolution data collected from multiple slave devices and various sensors.

[0066] B. The slave device is connected and communicated with the inverter of the distributed photovoltaic power generation component, and collects the real-time information of the photovoltaic component with a higher time resolution, including but not limited to voltage, current, power and other information, and sends it to the master device. There are multiple distributed photovoltaic power generation users in a substation area, and there are multiple distributed photovoltaic power generation component inverters. Each inverter is connected to a slave device, and multiple slave devices are connected to a master device.

[0067] In another embodiment of a grayscale publishing method of the present invention:

[0068] 1. Data collection:

[0069] Use the Sky Imager to capture ground-based cloud images.

[0070] Collect information from PV panels, including voltage, current, power, etc., as well as other related sensors (such as temperature and humidity).

[0071] 2. Data preprocessing:

[0072] Fluctuations in solar irradiance directly affect photovoltaic output.

[0073] Cloud images are studied by analyzing the effects of cloud occlusion and scattering on solar irradiance.

[0074] The original cloud images are preprocessed, including enhancement, denoising, and overexposure suppression.

[0075] 3. Neural network feature extraction:

[0076] The preprocessed cloud images are processed using a neural network model.

[0077] The neural network consists of convolutional, pooling, and normalization layers to extract high-level abstract features related to cloud images and PV output.

[0078] 4. Neural network feature extraction and photovoltaic module data integration:

[0079] Combine high-level features extracted by the neural network with PV module data (voltage, current, power, etc.).

[0080] A fully connected neural network layer is used for feature fusion to obtain the initial photovoltaic output prediction value.

[0081] 5. Local weather correction:

[0082] Local meteorological factors, such as water vapor in the air, can cause small fluctuations in solar irradiance.

[0083] Temperature affects the photoelectric conversion efficiency of photovoltaic modules, thereby affecting photovoltaic output.

[0084] Consider local meteorological conditions near PV modules to optimize initial PV output forecasts.

[0085] 6. Final photovoltaic output prediction:

[0086] The cloud image features and PV module data are combined to adjust the initial prediction and generate the final PV output prediction.

[0087] Finally, all the parts of the invention not fully described adopt mature products and mature technical means in the existing technology.

[0088] In the description of this specification, reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in the embodiment or example of the invention.

[0089] Although embodiments of the invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A power output prediction device for a substation area suitable for substation area microgrid autonomy, characterized by: The invention comprises a master device and a plurality of slave devices, wherein the master device is connected in parallel with a sky imager and a meteorological element sensor to collect weather data, and the data on the master device is connected in parallel with a plurality of slave devices to record and manage high time resolution data from the plurality of slave devices.

2. The output prediction device for a substation area suitable for substation area microgrid autonomy according to claim 1, characterized in that: The main device includes a housing module, a main control board module, an AI computing module, an antenna module, a serial port interface module, an Ethernet interface module, an HPLC interface module, a power module, an encryption module and a communication module at the hardware level; The protection level of the housing module is not less than the IP51 requirement specified in GB / T 4208; The mainboard module includes a main control CPU, memory, flash memory module, and security chip, wherein the CPU is a safe and controllable industrial-grade chip, supports a memory greater than or equal to 4GB to provide sufficient AI computing power, and a flash memory module greater than or equal to 64GB to provide a large amount of data storage capacity, and supports expansion. The security chip supports terminal self-protection and security protection when the terminal interacts with the main station, distributed new energy power station, and operation and maintenance tools; The AI ​​computing power module provides dedicated hardware support for the AI ​​algorithm to meet the computing power requirements of the AI ​​algorithm; The antenna module includes a transmission antenna, a Beidou interface and a pulse-per-second interface, which are used for remote communication, geographic location information collection and time synchronization; The serial port interface module is used for communication between the master device and the slave device, has multiple RS-485 and multiple RS-232 / RS-485 channels and supports serial port switching; The Ethernet interface module is used for master station communication or remote, and the HPLC module is used for communication between the master device and the slave device to adapt to communication with the slave device in a variety of different scenarios; The power module is used to power the terminal device; The encryption module is used for longitudinal data encryption, local communication data encryption and local storage data encryption during longitudinal data transmission between the terminal and the master station. The encryption module meets the safety requirements of power dispatching; The communication module is used to process the communication protocol used when the master device communicates with the slave device or with other devices.

3. The output prediction device for a substation area suitable for substation area microgrid autonomy according to claim 1, characterized in that: The main device includes an operating system and functional applications at the software level. The operating system includes an autonomous and controllable embedded Linux operating system and a hardware adapter driver. The security functions involve secure access, data encryption, trusted authentication and security baselines. Business applications include prediction algorithms, data management and data collection.

4. The output prediction device for a substation area suitable for substation area microgrid autonomy according to claim 1, characterized in that: The slave device is connected to and communicates with the inverter of the distributed photovoltaic power generation component, collects real-time information of the photovoltaic component including voltage, current, power and other information with high-frequency time resolution, and sends it to the master device; wherein, there are multiple distributed photovoltaic power generation users in a substation area, and there are multiple distributed photovoltaic power generation component inverters, each inverter is connected to a slave device, and multiple slave devices are connected to a master device together.

5. A prediction method for a power output prediction device for a power station area suitable for autonomous power station area microgrid according to claim 1, characterized in that: The prediction device collects ground cloud images through a sky imager, integrates information such as voltage, current, power, etc. of photovoltaic power generation components and information such as temperature and humidity collected by other related sensors, and predicts the output of distributed photovoltaics in the substation area. The steps include: C. The fluctuation of solar irradiance is the direct cause of the fluctuation of photovoltaic output. This method uses ground-based cloud images to analyze the impact of cloud shading and scattering on solar irradiance. The cloud image data is captured by a sky imager, and the original cloud image data is pre-processed by image enhancement, image denoising, overexposure suppression, etc. D. The ground cloud image after image preprocessing is passed through a multi-layer artificial neural network model consisting of convolutional layers, pooling layers, and normalization layers to extract high-level abstract feature information related to cloud images and photovoltaic output; E. The high-level abstract feature information extracted by the artificial neural network model and the photovoltaic power generation component information collected from the equipment, such as voltage, current, power, etc., are passed through the feature fusion neural network layer composed of a fully connected neural network to generate an initial photovoltaic output prediction value; F. Local meteorological factors, such as water vapor in the air, can also cause small fluctuations in solar irradiance. Temperature can affect the photoelectric conversion efficiency of photovoltaic power generation components and can also affect the output of photovoltaic components. Therefore, it is also necessary to correct the initial photovoltaic output forecast value based on the local meteorological factors near the photovoltaic components to generate the final photovoltaic output forecast value.

Images